fix scheduling reasoning on physical cores directly

document sync experiment

.agents/invariants/PIMSIM_NN_ORACLE_INVARIANT.md

@@ -0,0 +1,26 @@
+# pimsim-nn Oracle Invariant
+
+`backend-simulators/pim/pimsim-nn` is the performance oracle. Its simulation
+behavior defines the hardware model used for Raptor/PIMCOMP comparisons.
+
+## Required invariant
+
+- Do not add instruction or operator support to pimsim-nn.
+- Changes to pimsim-nn must preserve simulation behavior exactly. Acceptable
+  changes are limited to behavior-neutral maintenance proven not to alter
+  simulated timing, scheduling, power, energy, or supported input programs.
+- Unsupported pimsim-nn operations must remain unsupported; do not approximate
+  their timing or map them onto another operation.
+- Validation must report an explicit unsupported-op diagnostic as
+  `UNSUPPORTED`, not as a simulator failure.
+- A compiled artifact with zero active PIM cores has no meaningful
+  non-functional simulation and must be reported as `SKIP`.
+- Adapt compiler inputs to the oracle instead. For YOLO, use
+  `validation/networks/pimcomp_models/yolo11n/yolo11n-pimsim-nn.onnx`, the
+  dedicated pimsim-ready performance artifact with Softmax operations removed.
+  Use `validation/networks/yolo11n/depth_51/yolo11n_depth_51.onnx` for YOLO
+  functional validation; removing Softmax changes the model's numerical
+  behavior, so the pimsim-ready artifact is not a correctness reference.
+
+Any proposed pimsim-nn behavior change requires explicit user authorization and
+must not be introduced as part of a compiler optimization.

.agents/invariants/PIM_SYNCHRONIZATION_INVARIANT.md

@@ -0,0 +1,105 @@
+# PIM Synchronization Invariant
+
+## Scope
+
+This invariant applies to PIM `SYNC`/`WAIT` planning, event-register
+allocation, repeating core bodies, host/global-memory transfers, and pipeline
+restart synchronization.
+
+## ISA contract
+
+`SYNC event, target` and `WAIT event, expected` have static operands. Event
+registers start at zero, persist across instruction-stream restart, and are
+incremented by `SYNC`. `WAIT` succeeds only when the register equals its
+static expected count, then resets it to zero. A plan must therefore guarantee
+exactly the expected number of increments before each wait. It must not use
+generation-dependent registers, treat `WAIT` as `>=`, permit overshoot, or
+consume one event register with independent waits in the same iteration.
+
+Event registers are local to the target core. Allocation must keep READY,
+FREE/reuse, stage-zero barrier, and downstream restart events disjoint on each
+physical core.
+
+## Host/global-memory lifetime
+
+For every host-routed dependency from writer `W` to reader `R`, the repeating
+program must establish:
+
+```text
+ST_N -> READY SYNC -> READY WAIT -> LD_N
+LD_N -> reader END_N -> reader start_N+1 -> FREE SYNC
+     -> writer FREE WAIT -> ST_N+1
+```
+
+The reader emits one FREE `SYNC` per unique physical `(W, R)` relation at body
+entry, before any instruction that can block. The writer consumes each FREE
+event at most once per iteration, with a static count equal to the unique
+readers in that reuse group. The wait is placed immediately before the first
+host store that can overwrite a member of the group, after all preceding
+useful computation. Multiple transfers on one `(W, R)` relation do not create
+additional FREE signals.
+
+Relations with the same writer and first dangerous overwrite boundary belong
+to one FREE group: splitting them cannot unblock an earlier store and only
+adds waits and register pressure. After preserving the READY registers required
+by a core, planning assigns independent groups to distinct overwrite boundaries
+as remaining capacity permits. Capacity-forced merged groups are ordered by
+first overwrite and waited before their earliest member. Unused READY capacity
+must not force all readers into one early aggregate wait.
+
+Host receives lower to READY `WAIT` followed by host load only. They must not
+emit a post-load acknowledgement. Writers must not wait for host readers at
+the end of their body. The repeated reader-entry FREE signal is both the
+initial-slot bootstrap and the release of the previous generation.
+
+## Pipeline restart independence
+
+Host lifetime synchronization does not replace pipeline restart
+synchronization. At the end of every repeating stage-zero body, all unique
+stage-zero physical cores execute a dissemination barrier with
+`ceil(log2(stageZeroCoreCount))` rounds. In round `r`, rank `i` sends one
+signal to rank `(i + 2^r) mod count` and waits for exactly one signal on that
+round's destination-local event register. Each round has its own register, so
+an early signal for a later round cannot overshoot or satisfy another wait.
+
+Every downstream core sends one restart-permission signal to its release-tree
+parent at body entry, after all host FREE signals. The root signals the
+stage-zero leader; other cores signal their binary-tree parent. After the barrier, the lowest-ranked
+stage-zero core waits for the root's permission before sending one restart
+signal to it. Every downstream parent waits for exactly one restart signal
+and for one permission from each existing child before forwarding the restart
+to those children and ending its current body. This reverse permission path
+prevents a fast parent from sending generation `N+1` before a child consumes
+generation `N`, which would overshoot an exact-count restart event.
+
+Thus no downstream core can restart and advance its input generation until
+every stage-zero core has completed the protected generation, and every
+restart event receives exactly one signal between waits. Stage-zero and
+downstream core sets are disjoint; barrier, restart, and restart-permission
+registers are reserved on their respective target cores. Host READY/FREE
+allocation must use only the remaining registers and must not move, merge,
+remove, or weaken this ordering.
+
+## Ownership and verification
+
+Deferred boundary planning owns relation deduplication, event allocation,
+expected counts, and the first dangerous overwrite action. Boundary
+realization owns body-entry FREE signals and materializes planned waits at
+their exact boundaries. Channel lowering preserves those explicit operations
+and lowers host receives as READY wait plus load; it must not reconstruct
+lifetime policy after scheduling information is lost.
+
+Verification must reject non-static lane/core mappings, duplicate physical
+reader representations, register collisions, oversubscribed event capacity,
+and any plan whose generated FREE signal count differs from its static wait
+count. Static and simulator deadlock detectors must model blocking WAITs,
+their remaining matching SYNC producers, missing signals, and exact-count
+overshoot in addition to SEND/RECV cycles. Both detectors must retain the
+per-source contributions accumulated since the last successful WAIT and must not
+add a wait-for edge to a source that already supplied its share of the current
+event value merely because that source has the same static SYNC in a future
+iteration. Structural tests must inspect generated instruction streams. Happens-
+before tests must unroll at least two logical iterations and add only
+same-core restart edges plus matched communication edges. Functional
+simulation must enforce exact-count waits and include stalled or randomized
+legal schedules so correctness never depends on relative core speed.

.agents/invariants/PIPELINE_SCHEDULING_INVARIANT.md

@@ -0,0 +1,56 @@
+# Pipeline Scheduling Invariant
+
+## Scope
+
+This invariant applies to pipeline stage partitioning, physical-core
+assignment, scheduled materialization, deferred transfers, and pipeline
+synchronization.
+
+## Invariant
+
+A scheduled compute operation and all of its lanes belong to exactly one
+pipeline stage. An operation may consume results produced in its own stage or
+the immediately preceding stage only. Therefore every compute-graph edge from
+stage `S` targets stage `S` or `S + 1`; backward edges and dependencies that
+skip a stage are invalid.
+
+Dynamic function inputs are stage-zero sources. Any operation that directly
+consumes one must belong to stage 0. A later stage may consume that data only
+through an explicit result forwarded by the preceding stage.
+
+PEFT assigns physical cores using the injected target topology and their actual
+resident-weight capacity. Pipeline stage capacities cover every physical core
+and may have different sizes when the partitioner predicts a lower maximum
+stage interval. Before stage-local packing, the layout groups physical cores
+by bidirectional link cost; stage cores need not have consecutive IDs. Packing
+scores transfers on those physical links, and communication realization must
+preserve the resulting core identities.
+Synchronization and deferred transfers consume the explicit stage identity;
+they must not infer it from a physical core number after placement.
+
+## Ownership
+
+Physical PEFT supplies the initial placement. Stage partitioning is the first
+phase of pipeline scheduling and owns stage adjacency. It must construct a valid
+operation-level partition before physical-core packing. Operations split for
+physical capacity retain one shared stage identity. Repacking may move work
+only within its assigned stage. Deferred-transfer planning and
+synchronization consume the verified stage assignment; they must not repair
+or reinterpret it.
+
+## Verification
+
+Before scheduled materialization, verify that:
+
+- every compute instance has one valid physical core and stage;
+- all lanes of one compute operation have the same stage;
+- every direct dynamic-function-input consumer belongs to stage 0;
+- every compute-graph edge stays within a stage or advances exactly one stage;
+- every stage-local resident-weight set fits its assigned physical core; and
+- stage capacities cover all physical cores exactly once; and
+- physical placement is a permutation of all target cores.
+
+Pipeline scheduling tests must include an uneven physical-core layout and a
+graph with a long-lived dependency that would cross multiple naive stage
+cuts. End-to-end validation must preserve functional results and exercise the
+existing synchronization lowering without simulator changes.

AGENTS.md

@@ -6,6 +6,9 @@ Before modifying the relevant subsystem, read:
 
 * `.agents/invariants/GRAPH_COMPUTE_BATCH_INVARIANT.md`
 * `.agents/invariants/PERFORMANCE_OPTIMIZATION_INVARIANT.md`
+* `.agents/invariants/PIMSIM_NN_ORACLE_INVARIANT.md`
+* `.agents/invariants/PIM_SYNCHRONIZATION_INVARIANT.md`
+* `.agents/invariants/PIPELINE_SCHEDULING_INVARIANT.md`
 * `.agents/invariants/SPATIAL_TARGET_GENERALITY_INVARIANT.md`
 * Build commands:
   * `cmake --build ./build_release`

README.md

@@ -5,7 +5,7 @@ targeting in-memory computing / processing-in-memory (PIM) architectures. It
 extends ONNX-MLIR with a PIM accelerator and progressively lowers ONNX-MLIR
 through custom MLIR dialects to simulator artifacts.
 
-The current target is the PIM simulator stack under `backend-simulators/pim`.
+The current target is the Pim simulator stack under `backend-simulators/pim`.
 Raptor emits binary per-core `.pim` instruction files by default, plus
 `memory.bin`, `config.json`, and weight binaries. It can also emit per-core JSON
 instruction files with `--pim-emit-json`.
@@ -29,9 +29,9 @@ lowering, scheduling, memory layout, and code-generation optimizations.
 - `backend-simulators/pim/pim-simulator` is the in-tree Rust functional
   simulator used by validation. It reads Raptor's `pim/` artifact directory and
   compares simulator output against native ONNX-MLIR execution.
-- `backend-simulators/pim/pimsim-nn` is the non-functional simulator submodule
-  used internally by validation for latency, power, and energy.
-  The helper scripts in `pimcomp_utils/` are for comparison with PIMCOMP-NN and
+- `backend-simulators/pim/pimsim-nn` contains the non-functional Pimsim
+  simulator used internally by validation for latency, power, and energy.
+  The helper scripts in `pimcomp_utils/` are for comparison with Pimcomp and
   contain local paths; treat them as local utilities, not portable workflows.
 
 ## Compilation pipeline
@@ -43,7 +43,7 @@ them to ONNX-MLIR through generated shim directories under
 High-level lowering flow:
 
 ```
-ONNX-MLIR -> Spatial -> Pim (tensor) -> Pim (bufferized) -> PIM artifacts
+ONNX-MLIR -> Spatial -> Pim (tensor) -> Pim (bufferized) -> Pim artifacts
 ```
 
 1. **ONNX -> Spatial** (`src/PIM/Conversion/ONNXToSpatial`).
@@ -81,20 +81,20 @@ ONNX-MLIR -> Spatial -> Pim (tensor) -> Pim (bufferized) -> PIM artifacts
    addressable accesses, and `PimBufferizationVerification` checks tensor
    absence, contiguity, and copy address spaces.
 
-5. **PIM local-memory planning**
+5. **Pim local-memory planning**
    (`src/PIM/Dialect/Pim/Passes/Transforms/LocalMemoryPlanning`).
    Computes whole-core lifetimes, reuses addresses for non-overlapping
-   allocations, and records the explicit plan in PIM IR. Reusable lifetime
+   allocations, and records the explicit plan in Pim IR. Reusable lifetime
    analysis lives under `src/PIM/Dialect/Pim/Passes/Analyses`.
-6. **PIM verification and code generation** (`src/PIM/Passes/PimCodegen` and
+6. **Pim verification and code generation** (`src/PIM/Passes/PimCodegen` and
    `src/PIM/Compiler`).
-   Verifies the memory plan and other PIM invariants, then emits `.pim` core
+   Verifies the memory plan and other Pim invariants, then emits `.pim` core
    files, weights, and `memory.bin` / `config.json` without rerunning liveness.
 
 Supporting pieces:
 - `src/PIM/Common` - shared IR, filesystem, diagnostics, reports, and utility
   helpers.
-- `src/PIM/Compiler` - PIM compiler options, planned-address materialization, binary
+- `src/PIM/Compiler` - Pim compiler options, planned-address materialization, binary
   instruction format, artifact writing, weight emission, and codegen entry
   points.
 - `src/PIM/Conversion/SpatialToGraphviz` - optional Spatial graphviz conversion
@@ -102,40 +102,97 @@ Supporting pieces:
 - `src/PIM/Passes` - pass registration and auxiliary passes.
 - `src/PIM/PimAccelerator.{cpp,hpp}` - ONNX-MLIR accelerator entry point.
 
-## PIM compiler options
+## Pim compiler options
 
-Pass these to `onnx-mlir` when compiling for PIM. These are all Raptor/PIM-specific
+Pass these to `onnx-mlir` when compiling for Pim. These are all Raptor/Pim-specific
 options; `onnx-mlir --help` lists the inherited ONNX-MLIR options.
 
-- `--maccel=PIM` - select the PIM accelerator.
+- `--maccel=PIM` - select the Pim accelerator. Default: no Pim accelerator.
 - `--EmitSpatial`, `--EmitPim`, `--EmitPimBufferized`,
-  `--EmitPimCodegen` - stop the PIM pipeline at the requested stage. The PIM
-  default is `--EmitPimCodegen`.
-- `--core-count=<N>` - required positive core count for PIM compilation.
-- `--crossbar-size=<N>` - crossbar width/height. Default in code is `128`.
-- `--crossbar-count=<N>` - crossbars per core. Default in code is `64`.
-- `--pim-target-config=<PATH>` - optional PIM target configuration used by the
+  `--EmitPimCodegen` - stop the Pim pipeline at the requested stage. Default:
+  `--EmitPimCodegen` for Pim compilation.
+- `--core-count=<N>` - required positive core count for Pim compilation.
+  Default: none; this option is required.
+- `--crossbar-size=<N>` - required positive crossbar width/height for Pim
+  compilation. Default: none; this option is required.
+- `--crossbar-count=<N>` - required positive crossbar count per core for Pim
+  compilation. Default: none; this option is required.
+- `--pipeline=<N>` - number of throughput pipeline stages; `1` preserves
+  latency scheduling. Default: `1`.
+- `--pim-target-config=<PATH>` - optional Pim target configuration used by the
   target adapter to construct the target-neutral Spatial scheduling cost and
   topology model. Resource values must match the explicit core/crossbar flags.
+  Default: empty; use the built-in target model.
 - `--pim-memory-report=<summary|none>` - emit the concise combined memory report
-  under `reports/memory_report.txt`, or disable it. Default is `summary`.
-- `--pim-only-codegen` - assume input is already bufferized PIM IR and only run
-  the codegen tail.
+  under `reports/memory_report.txt`, or disable it. Default: `summary`.
+- `--pim-only-codegen` - assume input is already bufferized Pim IR and only run
+  the codegen tail. Default: off.
+- `--pim-disable-synchronization` - omit generated `wait` and `sync`
+  instructions for performance ablation. Default: off.
+- `--pim-disable-spatial-planning` - select the first, trivial DenseNCHW layout
+  alternative for every Spatial plan operation, disabling cost-based layout
+  planning while leaving ONNX rewrites and graph-compute merging enabled.
+  Default: off.
+
+### Spatial layout plan variants
+
+Spatial plan operations advertise alternatives as an exact combination of
+operand physical layouts and one result physical layout. Every plan operation
+has the default `DenseNCHW -> DenseNCHW` alternative. The planner can select
+the following additional variants when the operation, tensor shapes, and
+target resources make them legal:
+
+| Physical layout or plan | Meaning and current use |
+|---|---|
+| `DenseNCHW` | Ordinary dense NCHW storage. This is the first alternative and the one selected by `--pim-disable-spatial-planning`. |
+| `NHWCRowStrip` | Row-strip storage for NCHW logical tensors: spatial rows are processed as channel vectors. This enables row-strip lowering through compatible chains. |
+| `Fragmented` | Fragmented physical input accepted by `Flatten`, which reassembles it to dense NCHW. It is not currently selected as a plan result. |
+| `NCHWRowStrip` | A Spatial IR layout enum value reserved for NCHW-oriented row strips; current layout-capability implementations do not advertise it as a plan alternative. |
+
+The operation-specific non-trivial alternatives are:
+
+| Plan operation | Additional alternatives beyond dense NCHW |
+|---|---|
+| `Conv2D` | Dense input to row-strip output, or row-strip input to row-strip output when the target-dependent Conv lowering supports it. |
+| `Flatten` | Fragmented input to dense output, or row-strip input to dense output when legal. |
+| `Relu` | Row-strip input to row-strip output. |
+| `SiLU` | Row-strip input to row-strip output, with a stronger intrinsic cost preference than the generic row-strip variant. |
+| `ResizeNearest` | Row-strip input to row-strip output when its lowering is legal. |
+| `MaxPool2D` | Dense input to row-strip output, or row-strip input to row-strip output. |
+| `GlobalAveragePool` | Dense input to row-strip output, or row-strip input to row-strip output. |
+| `BiasAdd` | Row-strip data input plus a dense bias input to row-strip output when the bias shape is supported. |
+| `Add` | All data inputs row-strip to row-strip output. |
+| `Concat` | All inputs row-strip to row-strip output. |
+
+Cost-based planning scores intrinsic alternative cost, operand layout
+mismatches, and downstream incompatibility, then iterates in alternating
+forward and reverse operation order until the bounded analysis converges.
+Function results are required to remain `DenseNCHW`; explicit materialization
+operations reconcile layout mismatches at boundaries. With
+`--pim-disable-spatial-planning`, the pass still runs and records a valid plan,
+but chooses the first dense alternative for every plan operation. Later graph
+compute merging is unchanged, so elementwise operations such as `Relu` remain
+separate from neighboring parallel operations and can create fan-out/fan-in
+diamonds.
 - `--pim-emit-json` - also emit `core_*.json` instruction files alongside
-  `core_*.pim`.
+  `core_*.pim`. Default: off.
 - `--pim-export-spatial-dataflow=<none|spatial1|spatial2|spatial3|spatial4|all>` -
   control Spatial dataflow CSV reports for the graph, trivially merged graph,
-  scheduled, and realized snapshots under `reports/`. Default is `none`.
+  scheduled, and realized snapshots under `reports/`. Default: `none`.
 - `--pim-conv-lowering=<auto|legacy|depthwise|packed-im2col|streamed-patch|streamed-packed|output-channel-tiled|input-k-tiled|tiled-2d>` -
-  select the convolution lowering strategy. Default is `auto`.
+  select the convolution lowering strategy. Default: `auto`.
 - `--pim-conv-im2col-max-elements=<N>` - maximum globally materialized im2col
-  elements per convolution before streaming. Default is `1048576`.
+  elements per convolution before streaming. Default: `1048576`.
 - `--pim-conv-stream-chunk-positions=<N>` - maximum output positions per
-  streamed convolution chunk. Default is `1024`.
+  streamed convolution chunk. Default: `1024`.
+- `--pim-report-conv-lowering=<true|false>` - emit a bounded convolution
+  lowering report. Default: `true`.
 - `--pim-detect-communication-deadlock` - statically simulate expanded
-  send/receive ordering and reject blocking deadlocks. Default is off.
+  send/receive ordering and reject blocking deadlocks. Default: off.
+- `--pim-verify-bufferization-copy-freedom` - run the expensive official Pim
+  tensor-copy freedom proof before bufferization. Default: off.
 
-## Standard PIM hardware profile
+## Standard Pim hardware profile
 
 Raptor's standard development and YOLO validation profile is:
 
@@ -149,7 +206,8 @@ Canonical compiler flags:
 
 `--crossbar-count=64 --crossbar-size=128 --core-count=144`
 
-`--core-count` remains mandatory and must be passed explicitly to the compiler.
+`--crossbar-size`, `--crossbar-count`, and `--core-count` remain mandatory and
+must be passed explicitly to the compiler.
 
 Example:
 
@@ -159,11 +217,11 @@ Example:
   --crossbar-count=64 --crossbar-size=128 --core-count=144
 ```
 
-This writes PIM artifacts under `/tmp/raptor/pim/`.
+This writes Pim artifacts under `/tmp/raptor/pim/`.
 
 ## Validation
 
-Functional validation compiles ONNX models, compares native ONNX-MLIR and PIM
+Functional validation compiles ONNX models, compares native ONNX-MLIR and Pim
 simulator outputs, and optionally reports latency, power, and energy. See
 [`validation/README.md`](validation/README.md) for prerequisites, usage,
 options, artifacts, and results.
@@ -276,7 +334,7 @@ cd backend-simulators/pim/pim-simulator
 cargo test
 ```
 
-## Repository Layout
+## Repository layout
 
 - `src/PIM/` - PIM accelerator implementation.
 - `test/PIM/` - PIM C++ unit tests.
@@ -284,6 +342,6 @@ cargo test
   slices, and pimsim config generation.
 - `backend-simulators/pim/pim-simulator/` - in-tree Rust functional simulator.
 - `backend-simulators/pim/pimsim-nn/` - non-functional simulator submodule.
-- `pimcomp_utils/` - local comparison helpers for PIMCOMP-NN.
+- `pimcomp_utils/` - local comparison helpers for Pimcomp.
 - `.github/actions/` and `.github/workflows/validate_operations.yml` - CI setup
   for MLIR/Protobuf caching, building Raptor, and validation.

backend-simulators/pim/pim-simulator/src/bin/pim-simulator/main.rs

@@ -11,6 +11,7 @@ use pimcore::cpu::crossbar::Crossbar;
 use pimcore::json_to_instruction::json_to_executor;
 use pimcore::memory_manager::CoreMemory;
 use pimcore::tracing::TRACER;
+use pimcore::{DiagnosticSchedulePolicy, DiagnosticScheduleTarget};
 use serde_json::Value;
 use std::collections::HashMap;
 use std::fs::{self, File};
@@ -53,13 +54,45 @@ struct Args {
     #[arg(long)]
     batch_size: Option<u32>,
 
-    /// Input binary for one iteration; repeat once per batch entry
-    #[arg(long = "input")]
-    inputs: Vec<PathBuf>,
+    /// Directory containing input_*.bin files, one per batch entry
+    #[arg(long = "input-dir")]
+    input_dir: PathBuf,
 
     /// Optional directory for per-iteration output dumps
     #[arg(long)]
     batch_output_dir: Option<PathBuf>,
+
+    /// Optional JSONL shadow provenance trace
+    #[arg(long)]
+    provenance_trace: Option<PathBuf>,
+
+    /// Diagnostic-only barrier between global throughput iterations
+    #[arg(long)]
+    provenance_global_barrier: bool,
+
+    /// Diagnostic-only ready-core delay, formatted as CORE:CYCLES
+    #[arg(long, value_name = "CORE:CYCLES")]
+    provenance_core_stall: Option<String>,
+
+    /// Diagnostic scheduler policy; greedy is the unchanged default
+    #[arg(long, value_enum, default_value_t = DiagnosticSchedulePolicyArg::Greedy)]
+    diagnostic_schedule_policy: DiagnosticSchedulePolicyArg,
+
+    /// Deterministic seed for the randomized diagnostic scheduler
+    #[arg(long, default_value_t = 0)]
+    diagnostic_schedule_seed: u64,
+
+    /// Adversarial target: WCORE:WPC:RCORE:RPC:BEGIN:END[:READER_ITER:WRITER_MIN_ITER]
+    #[arg(long, value_name = "WCORE:WPC:RCORE:RPC:BEGIN:END[:RITER:WMIN]")]
+    diagnostic_schedule_target: Option<String>,
+
+    /// Maximum number of target-consumer deferrals
+    #[arg(long, default_value_t = 10_000)]
+    diagnostic_schedule_deferral_budget: u64,
+
+    /// Diagnostic delay applied when a target reader reaches its PC, formatted as CORE:PC:CYCLES or CORE:PC:ITERATION:CYCLES
+    #[arg(long, value_name = "CORE:PC[:ITERATION]:CYCLES")]
+    diagnostic_target_stall: Option<String>,
 }
 
 #[derive(Clone, Debug, ValueEnum)]
@@ -68,6 +101,13 @@ enum ExecutionMode {
     Throughput,
 }
 
+#[derive(Clone, Copy, Debug, ValueEnum)]
+enum DiagnosticSchedulePolicyArg {
+    Greedy,
+    Randomized,
+    Adversarial,
+}
+
 fn main() -> Result<()> {
     let args = Args::parse();
 
@@ -89,6 +129,33 @@ fn main() -> Result<()> {
         }
     };
     set_memory(&mut executor, memory);
+    if let Some(path) = &args.provenance_trace {
+        executor.enable_provenance(path)?;
+    }
+    executor.set_provenance_global_barrier(args.provenance_global_barrier);
+    if let Some(spec) = args.provenance_core_stall.as_deref() {
+        let (core, cycles) = parse_core_stall(spec)?;
+        executor.set_provenance_core_stall(core, cycles);
+    }
+    executor.set_diagnostic_schedule_policy(match args.diagnostic_schedule_policy {
+        DiagnosticSchedulePolicyArg::Greedy => DiagnosticSchedulePolicy::Greedy,
+        DiagnosticSchedulePolicyArg::Randomized => DiagnosticSchedulePolicy::Randomized,
+        DiagnosticSchedulePolicyArg::Adversarial => DiagnosticSchedulePolicy::Adversarial,
+    });
+    executor.set_diagnostic_schedule_seed(args.diagnostic_schedule_seed);
+    executor.set_diagnostic_schedule_deferral_budget(args.diagnostic_schedule_deferral_budget);
+    if let Some(spec) = args.diagnostic_target_stall.as_deref() {
+        let (core, pc, iteration, cycles) = parse_target_stall(spec)?;
+        executor.set_diagnostic_target_stall(core, pc, iteration, cycles);
+    }
+    if let Some(spec) = args.diagnostic_schedule_target.as_deref() {
+        executor.set_diagnostic_schedule_target(parse_schedule_target(spec)?);
+    } else if matches!(
+        args.diagnostic_schedule_policy,
+        DiagnosticSchedulePolicyArg::Adversarial
+    ) {
+        bail!("adversarial scheduling requires --diagnostic-schedule-target");
+    }
     TRACER
         .lock()
         .unwrap()
@@ -107,6 +174,75 @@ fn main() -> Result<()> {
     Ok(())
 }
 
+fn parse_core_stall(spec: &str) -> Result<(usize, u64)> {
+    let (core, cycles) = spec
+        .split_once(':')
+        .context("--provenance-core-stall must be CORE:CYCLES")?;
+    let core = core.parse().context("invalid stalled core")?;
+    let cycles = cycles.parse().context("invalid stall cycle count")?;
+    if cycles == 0 {
+        bail!("--provenance-core-stall cycles must be positive");
+    }
+    Ok((core, cycles))
+}
+
+fn parse_schedule_target(spec: &str) -> Result<DiagnosticScheduleTarget> {
+    let values: Vec<usize> = spec
+        .split(':')
+        .map(|value| {
+            value
+                .parse()
+                .with_context(|| format!("invalid schedule target field: {value}"))
+        })
+        .collect::<Result<_>>()?;
+    if values.len() != 6 && values.len() != 8 {
+        bail!(
+            "--diagnostic-schedule-target requires WCORE:WPC:RCORE:RPC:BEGIN:END with optional RITER:WMIN"
+        );
+    }
+    if values[5] <= values[4] {
+        bail!("schedule target address end must be greater than begin");
+    }
+    Ok(DiagnosticScheduleTarget {
+        writer_core: values[0],
+        writer_pc: values[1],
+        reader_core: values[2],
+        reader_pc: values[3],
+        address_begin: values[4],
+        address_end: values[5],
+        reader_iteration: (values.len() == 8).then_some(values[6] as u32),
+        writer_min_iteration: (values.len() == 8).then_some(values[7] as u32),
+    })
+}
+
+fn parse_target_stall(spec: &str) -> Result<(usize, usize, Option<u32>, u64)> {
+    let values: Vec<&str> = spec.split(':').collect();
+    if values.len() != 3 && values.len() != 4 {
+        bail!("--diagnostic-target-stall must be CORE:PC:CYCLES or CORE:PC:ITERATION:CYCLES");
+    }
+    let core = values[0].parse().context("invalid target-stall core")?;
+    let pc = values[1].parse().context("invalid target-stall PC")?;
+    let (iteration, cycle_field) = if values.len() == 4 {
+        (
+            Some(
+                values[2]
+                    .parse()
+                    .context("invalid target-stall iteration")?,
+            ),
+            values[3],
+        )
+    } else {
+        (None, values[2])
+    };
+    let cycles = cycle_field
+        .parse()
+        .context("invalid target-stall cycle count")?;
+    if cycles == 0 {
+        bail!("--diagnostic-target-stall cycles must be positive");
+    }
+    Ok((core, pc, iteration, cycles))
+}
+
 fn batch_size(args: &Args) -> Result<u32> {
     match (&args.mode, args.batch_size) {
         (ExecutionMode::Latency, None | Some(1)) => Ok(1),
@@ -142,16 +278,9 @@ fn input_regions(config: &Value) -> Result<Vec<(usize, usize)>> {
 }
 
 fn retrieve_inputs(args: &Args, batch_size: u32) -> Result<Vec<Vec<u8>>> {
-    if args.inputs.len() != batch_size as usize {
-        bail!(
-            "batch size {batch_size} requires {} inputs, got {}",
-            batch_size,
-            args.inputs.len()
-        );
-    }
-    args.inputs
-        .iter()
-        .map(|path| fs::read(path).with_context(|| format!("Failed to read input file: {path:?}")))
+    (0..batch_size)
+        .map(|index| args.input_dir.join(format!("input_{index}.bin")))
+        .map(|path| fs::read(&path).with_context(|| format!("Failed to read input file: {path:?}")))
         .collect()
 }
 

backend-simulators/pim/pim-simulator/src/lib/binary_to_instruction/mod.rs

@@ -80,19 +80,19 @@ fn read_i32_le(bytes: &[u8], offset: usize) -> i32 {
 
 fn parse_binary_records(bytes: &[u8]) -> Result<Vec<InstructionRecord>> {
     ensure!(bytes.len() >= HEADER_SIZE, "binary core file too small");
-    ensure!(&bytes[0..4] == MAGIC, "invalid PIM binary magic");
+    ensure!(&bytes[0..4] == MAGIC, "invalid Pim binary magic");
 
     let version = read_u32_le(bytes, 4);
     ensure!(
         version == VERSION,
-        "unsupported PIM binary version {version}"
+        "unsupported Pim binary version {version}"
     );
 
     let instruction_count = read_u32_le(bytes, 8) as usize;
     let expected_len = HEADER_SIZE + instruction_count * RECORD_SIZE;
     ensure!(
         bytes.len() == expected_len,
-        "PIM binary size mismatch: expected {expected_len} bytes, got {}",
+        "Pim binary size mismatch: expected {expected_len} bytes, got {}",
         bytes.len()
     );
 
@@ -335,7 +335,7 @@ fn append_record(
                 .set_offset_select_value(generic1, 0);
             inst_builder.make_inst(sync, inst_data_builder.build());
         }
-        _ => bail!("unsupported PIM binary opcode {opcode}"),
+        _ => bail!("unsupported Pim binary opcode {opcode}"),
     }
     Ok(())
 }

backend-simulators/pim/pim-simulator/src/lib/cpu/mod.rs

@@ -2,11 +2,14 @@ use crate::utility::AddressArg;
 use anyhow::{Context, Result, ensure};
 use std::{collections::HashMap, fmt::Debug};
 
+use super::{DiagnosticSchedulePolicy, DiagnosticScheduleTarget};
 use crate::{
     cpu::crossbar::Crossbar,
     instruction_set::Instructions,
     memory_manager::{CoreMemory, MemoryStorable, type_traits::TryToUsize},
+    provenance::ProvenanceTracker,
 };
+use serde_json::json;
 
 pub mod crossbar;
 
@@ -14,6 +17,7 @@ pub mod crossbar;
 pub struct CPU<'a> {
     cores: Box<[Core<'a>]>,
     batch_outputs: Option<BatchOutputs>,
+    provenance: Option<ProvenanceTracker>,
 }
 
 #[derive(Debug, Clone)]
@@ -53,9 +57,265 @@ impl<'a> CPU<'a> {
         Self {
             cores: cores.into(),
             batch_outputs: None,
+            provenance: None,
         }
     }
 
+    pub(crate) fn enable_provenance(
+        &mut self,
+        path: impl AsRef<std::path::Path>,
+    ) -> std::io::Result<()> {
+        self.provenance = Some(ProvenanceTracker::new(self.cores.len(), path)?);
+        Ok(())
+    }
+
+    pub(crate) fn begin_provenance_batch(&mut self, batch_size: usize) {
+        if let Some(provenance) = &mut self.provenance {
+            provenance.begin_batch(batch_size);
+        }
+    }
+
+    pub(crate) fn set_execution_context(
+        &mut self,
+        cycle: u64,
+        core: usize,
+        pc: usize,
+        iteration: u32,
+    ) {
+        if let Some(provenance) = &mut self.provenance {
+            provenance.set_context(cycle, core, pc, iteration);
+        }
+    }
+
+    pub(crate) fn provenance_input_store(&mut self, address: usize, size: usize, sample: u32) {
+        if let Some(provenance) = &mut self.provenance {
+            provenance.input_store(address, size, sample);
+        }
+    }
+
+    pub(crate) fn provenance_global_store_from_local(
+        &mut self,
+        core: usize,
+        global_address: usize,
+        local_address: usize,
+        size: usize,
+    ) {
+        if let Some(provenance) = &mut self.provenance {
+            provenance.global_store_from_local(core, global_address, local_address, size);
+        }
+    }
+
+    pub(crate) fn provenance_global_load_to_local(
+        &mut self,
+        core: usize,
+        global_address: usize,
+        local_address: usize,
+        size: usize,
+    ) {
+        if let Some(provenance) = &mut self.provenance {
+            provenance.global_load_to_local(core, global_address, local_address, size);
+        }
+    }
+
+    pub(crate) fn provenance_local_copy(
+        &mut self,
+        core: usize,
+        destination: usize,
+        source: usize,
+        size: usize,
+        operation: &'static str,
+    ) {
+        if let Some(provenance) = &mut self.provenance {
+            provenance.local_copy(core, destination, source, size, operation);
+        }
+    }
+
+    pub(crate) fn provenance_local_strided_copy(
+        &mut self,
+        core: usize,
+        destination: usize,
+        source: usize,
+        element_size: usize,
+        stride: usize,
+        element_count: usize,
+        operation: &'static str,
+    ) {
+        if let Some(provenance) = &mut self.provenance {
+            provenance.local_strided_copy(
+                core,
+                destination,
+                source,
+                element_size,
+                stride,
+                element_count,
+                operation,
+            );
+        }
+    }
+
+    pub(crate) fn provenance_local_transform(
+        &mut self,
+        core: usize,
+        destination: usize,
+        sources: &[(usize, usize)],
+        output_size: usize,
+        operation: &'static str,
+    ) {
+        if let Some(provenance) = &mut self.provenance {
+            provenance.local_transform(core, destination, sources, output_size, operation);
+        }
+    }
+
+    pub(crate) fn provenance_local_broadcast_transform(
+        &mut self,
+        core: usize,
+        destination: usize,
+        source: usize,
+        source_size: usize,
+        output_size: usize,
+        operation: &'static str,
+    ) {
+        if let Some(provenance) = &mut self.provenance {
+            provenance.local_broadcast_transform(
+                core,
+                destination,
+                source,
+                source_size,
+                output_size,
+                operation,
+            );
+        }
+    }
+
+    pub(crate) fn provenance_local_mvm_transform(
+        &mut self,
+        core: usize,
+        destination: usize,
+        source: usize,
+        element_size: usize,
+        output_size: usize,
+        used_rows: &[bool],
+        operation: &'static str,
+    ) {
+        if let Some(provenance) = &mut self.provenance {
+            provenance.local_mvm_transform(
+                core,
+                destination,
+                source,
+                element_size,
+                output_size,
+                used_rows,
+                operation,
+            );
+        }
+    }
+
+    pub(crate) fn provenance_send_transfer(
+        &mut self,
+        sender: usize,
+        receiver: usize,
+        source: usize,
+        destination: usize,
+        size: usize,
+    ) {
+        if let Some(provenance) = &mut self.provenance {
+            provenance.send_transfer(sender, receiver, source, destination, size);
+        }
+    }
+
+    pub(crate) fn finish_provenance(&mut self) {
+        if let Some(provenance) = &mut self.provenance {
+            provenance.flush();
+        }
+    }
+
+    pub(crate) fn provenance_schedule_stall(&mut self, core: usize, remaining_cycles: u64) {
+        if let Some(provenance) = &self.provenance {
+            provenance.schedule_stall(core, remaining_cycles);
+        }
+    }
+
+    pub(crate) fn provenance_schedule_target_stall(
+        &mut self,
+        core: usize,
+        pc: usize,
+        remaining_cycles: u64,
+    ) {
+        if let Some(provenance) = &self.provenance {
+            provenance.schedule_target_stall(core, pc, remaining_cycles);
+        }
+    }
+
+    pub(crate) fn provenance_schedule_config(&mut self, config: super::DiagnosticScheduleConfig) {
+        if let Some(provenance) = &self.provenance {
+            provenance.schedule_config(
+                match config.policy {
+                    DiagnosticSchedulePolicy::Greedy => "greedy",
+                    DiagnosticSchedulePolicy::Randomized => "randomized",
+                    DiagnosticSchedulePolicy::Adversarial => "adversarial",
+                },
+                config.seed,
+                config.target.map(|target| {
+                    json!({
+                        "writer_core": target.writer_core,
+                        "writer_pc": target.writer_pc,
+                        "reader_core": target.reader_core,
+                        "reader_pc": target.reader_pc,
+                        "address_begin": target.address_begin,
+                        "address_end": target.address_end,
+                        "reader_iteration": target.reader_iteration,
+                        "writer_min_iteration": target.writer_min_iteration,
+                    })
+                }),
+                config.deferral_budget,
+                config.fixed_stall,
+                config.fixed_target_stall,
+            );
+        }
+    }
+
+    pub(crate) fn provenance_scheduler_event(
+        &mut self,
+        event: &'static str,
+        core: usize,
+        pc: usize,
+        iteration: u32,
+        reason: &'static str,
+        selected_core: Option<usize>,
+        target: Option<DiagnosticScheduleTarget>,
+        deferrals: u64,
+    ) {
+        if let Some(provenance) = &self.provenance {
+            let target = target.map(|target| {
+                json!({
+                    "writer_core": target.writer_core,
+                    "writer_pc": target.writer_pc,
+                    "reader_core": target.reader_core,
+                    "reader_pc": target.reader_pc,
+                    "address_begin": target.address_begin,
+                    "address_end": target.address_end,
+                    "reader_iteration": target.reader_iteration,
+                    "writer_min_iteration": target.writer_min_iteration,
+                })
+            });
+            provenance.scheduler_event(json!({
+                "event": event,
+                "cycle": self.provenance_cycle(),
+                "core": core,
+                "pc": pc,
+                "core_iteration": iteration,
+                "reason": reason,
+                "selected_core": selected_core,
+                "target": target,
+                "deferrals": deferrals,
+            }));
+        }
+    }
+
+    fn provenance_cycle(&self) -> u64 {
+        self.provenance.as_ref().map_or(0, ProvenanceTracker::cycle)
+    }
+
     pub(crate) fn set_current_iteration(&mut self, iteration: u32) {
         if let Some(batch_outputs) = &mut self.batch_outputs {
             batch_outputs.iteration = iteration as usize;
@@ -93,6 +353,7 @@ impl<'a> CPU<'a> {
         let Self {
             cores,
             batch_outputs,
+            ..
         } = self;
         let (host, cores) = cores.split_at_mut(1);
         let bytes = cores[core - 1].load::<u8>(core_address, size)?[0];

backend-simulators/pim/pim-simulator/src/lib/instruction_set/isa.rs

@@ -285,6 +285,10 @@ where
     let load = loads[0];
     let vec: Cow<[M]> = load.up();
     let matrix = crossbar.load::<M>(crossbar_stored_bytes)?[0];
+    let used_rows: Vec<bool> = matrix
+        .chunks_exact(crossbar_elem_width)
+        .map(|row| row.iter().any(|value| *value != M::from_f32(0.0)))
+        .collect();
 
     // --- FAER IMPLEMENTATION ---
 
@@ -323,6 +327,16 @@ where
 
     let res_up: Cow<[T]> = res.as_slice().up();
     core.execute_store(rd_val, res_up.as_ref());
+    let _ = core;
+    cores.provenance_local_mvm_transform(
+        core_indx as usize,
+        rd_val as usize,
+        r1_val as usize,
+        size_of::<F>(),
+        res_up.len() * size_of::<T>(),
+        &used_rows,
+        "mvmul",
+    );
 
     TRACER.lock().unwrap().post_mvm::<F, M, T>(cores, data);
     Ok(InstructionStatus::Completed)
@@ -389,6 +403,14 @@ where
     );
     let res_up: Cow<[T]> = res.as_slice().up();
     core.execute_store(rd_val, res_up.as_ref());
+    let _ = core;
+    cores.provenance_local_transform(
+        core_indx as usize,
+        rd_val,
+        &[(r1_val, byte_len), (r2_val, byte_len)],
+        byte_len,
+        "vvadd",
+    );
     TRACER.lock().unwrap().post_vvadd::<F, T>(cores, data);
     Ok(InstructionStatus::Completed)
 }
@@ -474,6 +496,13 @@ where
     );
     let res_up: Cow<[T]> = res.as_slice().up();
     core.execute_store(rd_val, res_up.as_ref());
+    cores.provenance_local_transform(
+        core_indx as usize,
+        rd_val,
+        &[(r1_val, byte_len), (r2_val, byte_len)],
+        byte_len,
+        "vvmul",
+    );
     Ok(InstructionStatus::Completed)
 }
 
@@ -780,6 +809,15 @@ where
         );
     }
     core.execute_store(destination, &result)?;
+    cores.provenance_local_strided_copy(
+        core_indx as usize,
+        destination as usize,
+        source,
+        size_of::<F>(),
+        stride,
+        element_count,
+        "vmv",
+    );
     Ok(InstructionStatus::Completed)
 }
 
@@ -799,16 +837,23 @@ pub fn vrsl(cores: &mut CPU, data: InstructionData) -> Result<InstructionStatus>
 #[inline(never)]
 pub fn ld(cores: &mut CPU, data: InstructionData) -> Result<InstructionStatus> {
     TRACER.lock().unwrap().pre_ld(cores, data);
-    let (core, rd, r1, _, imm_len, offset_select, offset_value) =
+    let (core_index, rd, r1, _, imm_len, offset_select, offset_value) =
         data.get_core_rd_r1_r2_immlen_offset();
-    ensure!(core != 0, "LD cannot be used to move from host to host");
-    let (host, core) = cores.host_and_cores(core);
+    ensure!(
+        core_index != 0,
+        "LD cannot be used to move from host to host"
+    );
+    let (r1_val, rd_val) = {
+        let (host, core) = cores.host_and_cores(core_index);
         let r1_val = core.register(r1);
         let rd_val = core.register(rd);
         let r1_val = add_offset_r1(r1_val, offset_select, offset_value);
         let rd_val = add_offset_rd(rd_val, offset_select, offset_value);
         let global_memory = host.load::<u8>(r1_val, imm_len)?;
         core.execute_store(rd_val, global_memory[0])?;
+        (r1_val, rd_val)
+    };
+    cores.provenance_global_load_to_local(core_index as usize, r1_val, rd_val, imm_len as usize);
     TRACER.lock().unwrap().post_ld(cores, data);
     Ok(InstructionStatus::Completed)
 }
@@ -828,6 +873,7 @@ pub fn st(cores: &mut CPU, data: InstructionData) -> Result<InstructionStatus> {
         (rd_val, r1_val)
     };
     cores.store_to_host(core, rd_val, r1_val, imm_len)?;
+    cores.provenance_global_store_from_local(core as usize, rd_val, r1_val, imm_len as usize);
     TRACER.lock().unwrap().post_st(cores, data);
     Ok(InstructionStatus::Completed)
 }
@@ -852,9 +898,9 @@ pub fn lldi(cores: &mut CPU, data: InstructionData) -> Result<InstructionStatus>
 #[inline(never)]
 pub fn lmv(cores: &mut CPU, data: InstructionData) -> Result<InstructionStatus> {
     TRACER.lock().unwrap().pre_lmv(cores, data);
-    let (core, rd, r1, _, imm_len, offset_select, offset_value) =
+    let (core_index, rd, r1, _, imm_len, offset_select, offset_value) =
         data.get_core_rd_r1_r2_immlen_offset();
-    let core = cores.core(core);
+    let core = cores.core(core_index);
     let r1_val = core.register(r1);
     let rd_val = core.register(rd);
     let r1_val = add_offset_r1(r1_val, offset_select, offset_value);
@@ -862,6 +908,8 @@ pub fn lmv(cores: &mut CPU, data: InstructionData) -> Result<InstructionStatus>
     let local_memory = core.load::<u8>(r1_val, imm_len)?;
     let tmp = local_memory[0].to_vec();
     core.execute_store(rd_val, tmp.as_slice());
+    let _ = core;
+    cores.provenance_local_copy(core_index as usize, rd_val, r1_val, imm_len as usize, "lmv");
     TRACER.lock().unwrap().post_lmv(cores, data);
     Ok(InstructionStatus::Completed)
 }
@@ -886,11 +934,21 @@ pub fn recv(cores: &mut CPU, data: InstructionData) -> Result<InstructionStatus>
     Ok(InstructionStatus::Receiving(data))
 }
 
+#[inline(never)]
+pub fn isa_wait(functor: usize) -> bool {
+    (wait as *const () as usize) == functor
+}
+
 #[inline(never)]
 pub fn wait(cores: &mut CPU, data: InstructionData) -> Result<InstructionStatus> {
     Ok(InstructionStatus::Waiting(data))
 }
 
+#[inline(never)]
+pub fn isa_sync(functor: usize) -> bool {
+    (sync as *const () as usize) == functor
+}
+
 #[inline(never)]
 pub fn sync(cores: &mut CPU, data: InstructionData) -> Result<InstructionStatus> {
     Ok(InstructionStatus::Sync(data))

backend-simulators/pim/pim-simulator/src/lib/provenance.rs

@@ -0,0 +1,639 @@
+use serde_json::{Value, json};
+use std::{
+    collections::BTreeSet,
+    fs::File,
+    io::{BufWriter, Write},
+    path::Path,
+    sync::{Arc, Mutex},
+};
+
+#[derive(Clone, Copy, Debug, Default, PartialEq, Eq, PartialOrd, Ord)]
+pub enum Provenance {
+    #[default]
+    Uninitialized,
+    Unknown,
+    Samples(u64),
+}
+
+impl Provenance {
+    pub fn sample(sample: u32) -> Self {
+        if sample < 64 {
+            Self::Samples(1_u64 << sample)
+        } else {
+            Self::Unknown
+        }
+    }
+
+    fn merge(self, other: Self) -> Self {
+        match (self, other) {
+            (Self::Unknown, _) | (_, Self::Unknown) => Self::Unknown,
+            (Self::Uninitialized, value) | (value, Self::Uninitialized) => value,
+            (Self::Samples(left), Self::Samples(right)) => Self::Samples(left | right),
+        }
+    }
+
+    fn merge_all(values: impl IntoIterator<Item = Self>) -> Self {
+        let mut values = values.into_iter();
+        values
+            .next()
+            .map_or(Self::Uninitialized, |first| values.fold(first, Self::merge))
+    }
+
+    fn samples(self) -> Vec<u32> {
+        match self {
+            Self::Samples(mask) => (0..64)
+                .filter(|sample| mask & (1_u64 << sample) != 0)
+                .collect(),
+            Self::Unknown | Self::Uninitialized => Vec::new(),
+        }
+    }
+
+    fn state(self) -> &'static str {
+        match self {
+            Self::Samples(_) => "known",
+            Self::Unknown => "unknown",
+            Self::Uninitialized => "uninitialized",
+        }
+    }
+
+    fn is_mixed(self) -> bool {
+        matches!(self, Self::Samples(mask) if mask.count_ones() > 1)
+    }
+
+    fn json(self) -> Value {
+        json!({
+            "provenance": self.samples(),
+            "provenance_state": self.state(),
+            "mixed": self.is_mixed(),
+        })
+    }
+}
+
+#[derive(Clone, Copy, Debug)]
+struct ExecutionContext {
+    cycle: u64,
+    core: usize,
+    pc: usize,
+    iteration: u32,
+}
+
+impl Default for ExecutionContext {
+    fn default() -> Self {
+        Self {
+            cycle: 0,
+            core: 0,
+            pc: 0,
+            iteration: 0,
+        }
+    }
+}
+
+#[derive(Clone, Copy, Debug, PartialEq, Eq, PartialOrd, Ord)]
+struct Writer {
+    version: u64,
+    cycle: u64,
+    core: usize,
+    pc: usize,
+    iteration: u32,
+    provenance: Provenance,
+}
+
+#[derive(Clone, Copy, Debug, Default)]
+struct GlobalCell {
+    provenance: Provenance,
+    writer: Option<Writer>,
+}
+
+#[derive(Debug)]
+struct TraceSink {
+    output: BufWriter<File>,
+}
+
+impl TraceSink {
+    fn new(path: impl AsRef<Path>) -> std::io::Result<Self> {
+        let file = File::create(path)?;
+        Ok(Self {
+            output: BufWriter::new(file),
+        })
+    }
+
+    fn event(&mut self, value: Value) {
+        serde_json::to_writer(&mut self.output, &value).expect("write provenance event");
+        self.output
+            .write_all(b"\n")
+            .expect("write provenance newline");
+    }
+
+    fn flush(&mut self) {
+        self.output.flush().expect("flush provenance trace");
+    }
+}
+
+#[derive(Debug, Clone)]
+pub struct ProvenanceTracker {
+    global: Vec<GlobalCell>,
+    local: Vec<Vec<Provenance>>,
+    next_version: u64,
+    context: ExecutionContext,
+    sink: Arc<Mutex<TraceSink>>,
+}
+
+impl ProvenanceTracker {
+    pub fn new(core_count: usize, path: impl AsRef<Path>) -> std::io::Result<Self> {
+        if let Some(parent) = path.as_ref().parent() {
+            std::fs::create_dir_all(parent)?;
+        }
+        let sink = Arc::new(Mutex::new(TraceSink::new(path)?));
+        let tracker = Self {
+            global: Vec::new(),
+            local: vec![Vec::new(); core_count],
+            next_version: 0,
+            context: ExecutionContext::default(),
+            sink,
+        };
+        tracker.write(json!({
+            "event": "provenance_trace_start",
+            "schema": 1,
+            "core_count": core_count,
+        }));
+        Ok(tracker)
+    }
+
+    pub fn begin_batch(&mut self, batch_size: usize) {
+        self.global.fill(GlobalCell::default());
+        for memory in &mut self.local {
+            memory.fill(Provenance::Uninitialized);
+        }
+        self.next_version = 0;
+        self.write(json!({
+            "event": "batch_start",
+            "batch_size": batch_size,
+        }));
+    }
+
+    pub fn set_context(&mut self, cycle: u64, core: usize, pc: usize, iteration: u32) {
+        self.context = ExecutionContext {
+            cycle,
+            core,
+            pc,
+            iteration,
+        };
+    }
+
+    pub fn cycle(&self) -> u64 {
+        self.context.cycle
+    }
+
+    pub fn flush(&mut self) {
+        self.sink.lock().unwrap().flush();
+    }
+
+    pub fn schedule_stall(&self, core: usize, remaining_cycles: u64) {
+        self.write(json!({
+            "event": "diagnostic_core_stall",
+            "cycle": self.context.cycle,
+            "core": core,
+            "pc": self.context.pc,
+            "core_iteration": self.context.iteration,
+            "remaining_cycles": remaining_cycles,
+        }));
+    }
+
+    pub fn schedule_target_stall(&self, core: usize, pc: usize, remaining_cycles: u64) {
+        self.write(json!({
+            "event": "diagnostic_target_stall",
+            "cycle": self.context.cycle,
+            "core": core,
+            "pc": pc,
+            "core_iteration": self.context.iteration,
+            "remaining_cycles": remaining_cycles,
+        }));
+    }
+
+    pub fn schedule_config(
+        &self,
+        policy: &'static str,
+        seed: u64,
+        target: Option<Value>,
+        deferral_budget: u64,
+        fixed_stall: Option<(usize, u64)>,
+        fixed_target_stall: Option<(usize, usize, Option<u32>, u64)>,
+    ) {
+        self.write(json!({
+            "event": "scheduler_config",
+            "schedule_policy": policy,
+            "schedule_seed": seed,
+            "target_dependency": target,
+            "deferral_budget": deferral_budget,
+            "fixed_stall": fixed_stall.map(|(core, cycles)| json!({
+                "core": core,
+                "cycles": cycles,
+            })),
+            "fixed_target_stall": fixed_target_stall.map(|(core, pc, iteration, cycles)| json!({
+                "core": core,
+                "pc": pc,
+                "iteration": iteration,
+                "cycles": cycles,
+            })),
+        }));
+    }
+
+    pub fn scheduler_event(&self, event: Value) {
+        self.write(event);
+    }
+
+    fn write(&self, value: Value) {
+        self.sink.lock().unwrap().event(value);
+    }
+
+    fn ensure_global(&mut self, end: usize) {
+        if self.global.len() < end {
+            self.global.resize(end, GlobalCell::default());
+        }
+    }
+
+    fn ensure_local(&mut self, core: usize, end: usize) {
+        if let Some(memory) = self.local.get_mut(core)
+            && memory.len() < end
+        {
+            memory.resize(end, Provenance::Uninitialized);
+        }
+    }
+
+    fn local_tags(&mut self, core: usize, address: usize, size: usize) -> Vec<Provenance> {
+        let Some(end) = address.checked_add(size) else {
+            return vec![Provenance::Unknown; size];
+        };
+        self.ensure_local(core, end);
+        self.local[core][address..end].to_vec()
+    }
+
+    fn store_local_tags(&mut self, core: usize, address: usize, tags: &[Provenance]) {
+        let Some(end) = address.checked_add(tags.len()) else {
+            return;
+        };
+        self.ensure_local(core, end);
+        self.local[core][address..end].copy_from_slice(tags);
+    }
+
+    fn unique_versions(cells: &[GlobalCell]) -> Vec<u64> {
+        cells
+            .iter()
+            .filter_map(|cell| cell.writer.map(|writer| writer.version))
+            .collect::<BTreeSet<_>>()
+            .into_iter()
+            .collect()
+    }
+
+    fn unique_writers(cells: &[GlobalCell]) -> Vec<Writer> {
+        cells
+            .iter()
+            .filter_map(|cell| cell.writer)
+            .collect::<BTreeSet<_>>()
+            .into_iter()
+            .collect()
+    }
+
+    fn writer_json(writer: Writer) -> Value {
+        json!({
+            "version": writer.version,
+            "cycle": writer.cycle,
+            "core": writer.core,
+            "pc": writer.pc,
+            "core_iteration": writer.iteration,
+            "provenance": writer.provenance.samples(),
+            "provenance_state": writer.provenance.state(),
+        })
+    }
+
+    pub fn input_store(&mut self, address: usize, size: usize, sample: u32) {
+        let Some(end) = address.checked_add(size) else {
+            return;
+        };
+        self.ensure_global(end);
+        let provenance = Provenance::sample(sample);
+        self.next_version += 1;
+        let writer = Writer {
+            version: self.next_version,
+            cycle: self.context.cycle,
+            core: 0,
+            pc: 0,
+            iteration: sample,
+            provenance,
+        };
+        let overwritten_versions = Self::unique_versions(&self.global[address..end]);
+        for cell in &mut self.global[address..end] {
+            *cell = GlobalCell {
+                provenance,
+                writer: Some(writer),
+            };
+        }
+        let mut event = json!({
+            "event": "external_input_store",
+            "cycle": self.context.cycle,
+            "core": 0,
+            "pc": 0,
+            "core_iteration": sample,
+            "address": address,
+            "size": size,
+            "sample": sample,
+            "version": writer.version,
+            "overwritten_versions": overwritten_versions,
+        });
+        if let Some(object) = event.as_object_mut() {
+            object.extend(provenance.json().as_object().unwrap().clone());
+        }
+        self.write(event);
+    }
+
+    pub fn global_store_from_local(
+        &mut self,
+        core: usize,
+        global_address: usize,
+        local_address: usize,
+        size: usize,
+    ) {
+        let Some(end) = global_address.checked_add(size) else {
+            return;
+        };
+        let tags = self.local_tags(core, local_address, size);
+        self.ensure_global(end);
+        self.next_version += 1;
+        let provenance = Provenance::merge_all(tags.iter().copied());
+        let writer = Writer {
+            version: self.next_version,
+            cycle: self.context.cycle,
+            core,
+            pc: self.context.pc,
+            iteration: self.context.iteration,
+            provenance,
+        };
+        let overwritten_versions = Self::unique_versions(&self.global[global_address..end]);
+        let overwritten_writers = Self::unique_writers(&self.global[global_address..end]);
+        for (cell, tag) in self.global[global_address..end].iter_mut().zip(tags) {
+            *cell = GlobalCell {
+                provenance: tag,
+                writer: Some(writer),
+            };
+        }
+        let mut event = json!({
+            "event": "global_store",
+            "cycle": self.context.cycle,
+            "core": core,
+            "pc": self.context.pc,
+            "core_iteration": self.context.iteration,
+            "address": global_address,
+            "local_address": local_address,
+            "size": size,
+            "version": writer.version,
+            "overwritten_versions": overwritten_versions,
+            "overwritten_writers": overwritten_writers.into_iter().map(Self::writer_json).collect::<Vec<_>>(),
+        });
+        if let Some(object) = event.as_object_mut() {
+            object.extend(provenance.json().as_object().unwrap().clone());
+        }
+        self.write(event);
+    }
+
+    pub fn global_load_to_local(
+        &mut self,
+        core: usize,
+        global_address: usize,
+        local_address: usize,
+        size: usize,
+    ) {
+        let Some(end) = global_address.checked_add(size) else {
+            return;
+        };
+        self.ensure_global(end);
+        let cells = self.global[global_address..end].to_vec();
+        let tags: Vec<_> = cells.iter().map(|cell| cell.provenance).collect();
+        let provenance = Provenance::merge_all(tags.iter().copied());
+        let writers = Self::unique_writers(&cells);
+        let versions = Self::unique_versions(&cells);
+        self.store_local_tags(core, local_address, &tags);
+        let mut event = json!({
+            "event": "global_load",
+            "cycle": self.context.cycle,
+            "core": core,
+            "pc": self.context.pc,
+            "core_iteration": self.context.iteration,
+            "address": global_address,
+            "local_address": local_address,
+            "size": size,
+            "versions": versions,
+            "last_writers": writers.into_iter().map(Self::writer_json).collect::<Vec<_>>(),
+        });
+        if let Some(object) = event.as_object_mut() {
+            object.extend(provenance.json().as_object().unwrap().clone());
+        }
+        self.write(event);
+    }
+
+    pub fn local_copy(
+        &mut self,
+        core: usize,
+        destination: usize,
+        source: usize,
+        size: usize,
+        operation: &'static str,
+    ) {
+        let tags = self.local_tags(core, source, size);
+        let provenance = Provenance::merge_all(tags.iter().copied());
+        self.store_local_tags(core, destination, &tags);
+        self.local_event(
+            operation,
+            core,
+            destination,
+            size,
+            provenance,
+            &[provenance],
+        );
+    }
+
+    pub fn local_strided_copy(
+        &mut self,
+        core: usize,
+        destination: usize,
+        source: usize,
+        element_size: usize,
+        stride: usize,
+        element_count: usize,
+        operation: &'static str,
+    ) {
+        let mut tags = Vec::with_capacity(element_size.saturating_mul(element_count));
+        for index in 0..element_count {
+            let address =
+                source.saturating_add(index.saturating_mul(stride).saturating_mul(element_size));
+            tags.extend(self.local_tags(core, address, element_size));
+        }
+        let provenance = Provenance::merge_all(tags.iter().copied());
+        self.store_local_tags(core, destination, &tags);
+        self.local_event(
+            operation,
+            core,
+            destination,
+            tags.len(),
+            provenance,
+            &[provenance],
+        );
+    }
+
+    pub fn local_transform(
+        &mut self,
+        core: usize,
+        destination: usize,
+        sources: &[(usize, usize)],
+        output_size: usize,
+        operation: &'static str,
+    ) {
+        let source_tags: Vec<Vec<_>> = sources
+            .iter()
+            .map(|&(address, size)| self.local_tags(core, address, size))
+            .collect();
+        let mut output = Vec::with_capacity(output_size);
+        for index in 0..output_size {
+            let provenance = Provenance::merge_all(
+                source_tags
+                    .iter()
+                    .filter_map(|tags| tags.get(index).copied()),
+            );
+            output.push(provenance);
+        }
+        let provenance = Provenance::merge_all(output.iter().copied());
+        self.store_local_tags(core, destination, &output);
+        let operands: Vec<_> = source_tags
+            .iter()
+            .map(|tags| Provenance::merge_all(tags.iter().copied()))
+            .collect();
+        self.local_event(
+            operation,
+            core,
+            destination,
+            output_size,
+            provenance,
+            &operands,
+        );
+    }
+
+    pub fn local_broadcast_transform(
+        &mut self,
+        core: usize,
+        destination: usize,
+        source: usize,
+        source_size: usize,
+        output_size: usize,
+        operation: &'static str,
+    ) {
+        let tags = self.local_tags(core, source, source_size);
+        let provenance = Provenance::merge_all(tags.iter().copied());
+        self.store_local_tags(core, destination, &vec![provenance; output_size]);
+        self.local_event(
+            operation,
+            core,
+            destination,
+            output_size,
+            provenance,
+            &[provenance],
+        );
+    }
+
+    pub fn local_mvm_transform(
+        &mut self,
+        core: usize,
+        destination: usize,
+        source: usize,
+        element_size: usize,
+        output_size: usize,
+        used_rows: &[bool],
+        operation: &'static str,
+    ) {
+        let mut provenance = Provenance::Uninitialized;
+        for (row, used) in used_rows.iter().copied().enumerate() {
+            if used {
+                provenance = provenance.merge(Provenance::merge_all(self.local_tags(
+                    core,
+                    source + row * element_size,
+                    element_size,
+                )));
+            }
+        }
+        self.store_local_tags(core, destination, &vec![provenance; output_size]);
+        self.local_event(
+            operation,
+            core,
+            destination,
+            output_size,
+            provenance,
+            &[provenance],
+        );
+    }
+
+    pub fn send_transfer(
+        &mut self,
+        sender: usize,
+        receiver: usize,
+        source: usize,
+        destination: usize,
+        size: usize,
+    ) {
+        let tags = self.local_tags(sender, source, size);
+        let provenance = Provenance::merge_all(tags.iter().copied());
+        self.store_local_tags(receiver, destination, &tags);
+        let mut event = json!({
+            "event": "send_recv_transfer",
+            "cycle": self.context.cycle,
+            "pc": self.context.pc,
+            "core_iteration": self.context.iteration,
+            "sender_core": sender,
+            "receiver_core": receiver,
+            "source_address": source,
+            "destination_address": destination,
+            "size": size,
+        });
+        if let Some(object) = event.as_object_mut() {
+            object.extend(provenance.json().as_object().unwrap().clone());
+        }
+        self.write(event);
+    }
+
+    fn local_event(
+        &self,
+        operation: &'static str,
+        core: usize,
+        destination: usize,
+        size: usize,
+        provenance: Provenance,
+        operands: &[Provenance],
+    ) {
+        let mut event = json!({
+            "event": "local_compute",
+            "operation": operation,
+            "cycle": self.context.cycle,
+            "core": core,
+            "pc": self.context.pc,
+            "core_iteration": self.context.iteration,
+            "destination_address": destination,
+            "size": size,
+            "operand_provenance": operands.iter().map(|tag| tag.json()).collect::<Vec<_>>(),
+        });
+        if let Some(object) = event.as_object_mut() {
+            object.extend(provenance.json().as_object().unwrap().clone());
+        }
+        self.write(event);
+        if provenance.is_mixed() {
+            self.write(json!({
+                "event": "cross_sample_data_mix",
+                "cycle": self.context.cycle,
+                "core": core,
+                "pc": self.context.pc,
+                "core_iteration": self.context.iteration,
+                "operation": operation,
+                "destination_address": destination,
+                "size": size,
+                "operand_provenance": operands.iter().map(|tag| tag.json()).collect::<Vec<_>>(),
+                "provenance": provenance.samples(),
+            }));
+        }
+    }
+}

backend-simulators/pim/pim-simulator/src/lib/send_recv.rs

@@ -73,6 +73,7 @@ where
                 let data = inst.data;
                 TRACER.lock().unwrap().pre_recv(cpu, data);
             }
+            {
                 let [sender_core, receiver_core] =
                     cpu.get_multiple_cores([sender.internal_core, receiver.internal_core]);
                 let memory = sender_core
@@ -85,6 +86,14 @@ where
                     })
                     .unwrap();
                 receiver_core.execute_store(receiver.address, memory[0]);
+            }
+            cpu.provenance_send_transfer(
+                sender.internal_core,
+                receiver.internal_core,
+                sender.address,
+                receiver.address,
+                sender.size,
+            );
             {
                 let sender = &mut core_instructions[sender.internal_core];
                 let pc = sender.program_counter;

backend-simulators/pim/pim-simulator/tests/sync.rs

@@ -297,7 +297,7 @@ fn multiple_send_recv_test() {
 }
 
 #[test]
-fn sync_wait_tokens_test() {
+fn sync_wait_exact_count_resets_test() {
     let cpu = common::empty_cpu(2);
     let mut cores = CoreInstructionsBuilder::new(2);
     let mut instructions = InstructionsBuilder::new();
@@ -313,14 +313,68 @@ fn sync_wait_tokens_test() {
     cores.set_core(1, instructions.build());
 
     data.set_core_indx(2).fix_core_indx();
-    for _ in 0..2 {
-        instructions.make_inst(wait, data.set_offset_select_value(0, 1).build());
-    }
+    instructions.make_inst(wait, data.set_offset_select_value(0, 2).build());
     cores.set_core(2, instructions.build());
 
     Executable::new(cpu, cores.build()).execute().unwrap();
 }
 
+#[test]
+fn sync_wait_rejects_overshoot() {
+    let cpu = common::empty_cpu(3);
+    let mut cores = CoreInstructionsBuilder::new(3);
+    let mut instructions = InstructionsBuilder::new();
+    let mut data = InstructionDataBuilder::new();
+
+    data.set_core_indx(1).fix_core_indx();
+    instructions.make_inst(
+        sync,
+        data.set_imm_core(3).set_offset_select_value(0, 0).build(),
+    );
+    cores.set_core(1, instructions.build());
+
+    data.set_core_indx(2).fix_core_indx();
+    instructions.make_inst(
+        sync,
+        data.set_imm_core(3).set_offset_select_value(0, 0).build(),
+    );
+    cores.set_core(2, instructions.build());
+
+    data.set_core_indx(3).fix_core_indx();
+    instructions.make_inst(wait, data.set_offset_select_value(0, 1).build());
+    cores.set_core(3, instructions.build());
+
+    let error = Executable::new(cpu, cores.build()).execute().unwrap_err();
+    assert!(error.to_string().contains("overshot exact value"));
+}
+
+#[test]
+fn sync_wait_deadlock_cycle_is_reported() {
+    let cpu = common::empty_cpu(2);
+    let mut cores = CoreInstructionsBuilder::new(2);
+    let mut instructions = InstructionsBuilder::new();
+    let mut data = InstructionDataBuilder::new();
+
+    data.set_core_indx(1).fix_core_indx();
+    instructions.make_inst(wait, data.set_offset_select_value(0, 1).build());
+    instructions.make_inst(
+        sync,
+        data.set_imm_core(2).set_offset_select_value(0, 0).build(),
+    );
+    cores.set_core(1, instructions.build());
+
+    data.set_core_indx(2).fix_core_indx();
+    instructions.make_inst(wait, data.set_offset_select_value(0, 1).build());
+    instructions.make_inst(
+        sync,
+        data.set_imm_core(1).set_offset_select_value(0, 0).build(),
+    );
+    cores.set_core(2, instructions.build());
+
+    let error = Executable::new(cpu, cores.build()).execute().unwrap_err();
+    assert!(error.to_string().contains("wait event"));
+}
+
 #[test]
 fn blocked_transfers_do_not_starve_sync_producer() {
     let cpu = common::empty_cpu(4);

src/PIM/Common/IR/AddressAnalysis.hpp

@@ -19,7 +19,7 @@ struct ResolvedContiguousAddress {
 };
 
 /// Records compile-time facts used when interpreting address arithmetic and
-/// loop-carried aliases inside PIM regions.
+/// loop-carried aliases inside Pim regions.
 struct StaticValueKnowledge {
   llvm::DenseMap<mlir::Value, int64_t> indexValues;
   llvm::DenseMap<mlir::Value, mlir::Value> aliases;

src/PIM/Common/IR/CoreBlockUtils.cpp

@@ -85,12 +85,12 @@ static mlir::LogicalResult walkPimCoreBlockImpl(mlir::Block& block,
       auto step = resolveIndexValue(forOp.getStep(), knowledge);
       if (failed(lower) || failed(upper) || failed(step)
           || (mode == CoreWalkMode::ExecuteCommunication && *step <= 0)) {
-        forOp.emitOpError() << "requires statically evaluable scf.for bounds for PIM " << purpose;
+        forOp.emitOpError() << "requires statically evaluable scf.for bounds for Pim " << purpose;
         hasFailure = true;
         continue;
       }
       if (*step <= 0) {
-        forOp.emitOpError("requires positive scf.for step for PIM verification");
+        forOp.emitOpError("requires positive scf.for step for Pim verification");
         hasFailure = true;
         continue;
       }
@@ -126,7 +126,7 @@ static mlir::LogicalResult walkPimCoreBlockImpl(mlir::Block& block,
     if (auto ifOp = mlir::dyn_cast<mlir::scf::IfOp>(op)) {
       auto condition = resolveIndexValue(ifOp.getCondition(), knowledge);
       if (failed(condition)) {
-        ifOp.emitOpError() << "requires statically evaluable scf.if condition for PIM " << purpose;
+        ifOp.emitOpError() << "requires statically evaluable scf.if condition for Pim " << purpose;
         hasFailure = true;
         continue;
       }
@@ -147,7 +147,7 @@ static mlir::LogicalResult walkPimCoreBlockImpl(mlir::Block& block,
     if (auto switchOp = mlir::dyn_cast<mlir::scf::IndexSwitchOp>(op)) {
       auto selector = resolveIndexValue(switchOp.getArg(), knowledge);
       if (failed(selector)) {
-        switchOp.emitOpError() << "requires a statically evaluable scf.index_switch selector for PIM " << purpose;
+        switchOp.emitOpError() << "requires a statically evaluable scf.index_switch selector for Pim " << purpose;
         hasFailure = true;
         continue;
       }

src/PIM/Common/IR/CoreBlockUtils.hpp

@@ -14,7 +14,7 @@ namespace onnx_mlir {
 using PimCoreCommunicationPlan = llvm::DenseMap<mlir::Block*, llvm::SmallVector<mlir::Operation*, 8>>;
 
 /// Returns true for ops in a `pim.core` body that only participate in static
-/// address or index computation and therefore do not emit PIM instructions.
+/// address or index computation and therefore do not emit Pim instructions.
 bool isCoreStaticAddressOp(mlir::Operation* op);
 
 /// Walks a `pim.core` body's communication stream, statically unrolling

src/PIM/Common/IR/EntryPointUtils.cpp

@@ -9,7 +9,7 @@ llvm::FailureOr<mlir::func::FuncOp> getPimEntryFunc(mlir::ModuleOp moduleOp) {
 
   llvm::SmallVector<mlir::ONNXEntryPointOp> entryPoints(moduleOp.getOps<mlir::ONNXEntryPointOp>());
   if (entryPoints.size() > 1) {
-    moduleOp.emitError("PIM pipeline requires a single ONNX entry point, but found ") << entryPoints.size();
+    moduleOp.emitError("Pim pipeline requires a single ONNX entry point, but found ") << entryPoints.size();
     return mlir::failure();
   }
   if (!entryPoints.empty()) {
@@ -38,7 +38,7 @@ llvm::FailureOr<mlir::func::FuncOp> getPimEntryFunc(mlir::ModuleOp moduleOp) {
   if (nonExternalFuncs.size() == 1)
     return nonExternalFuncs.front();
 
-  moduleOp.emitError("could not resolve a unique PIM entry function");
+  moduleOp.emitError("could not resolve a unique Pim entry function");
   return mlir::failure();
 }
 

src/PIM/Common/IR/EntryPointUtils.hpp

@@ -5,7 +5,7 @@
 
 namespace onnx_mlir {
 
-/// Resolves the function the PIM pipeline should treat as its entry point.
+/// Resolves the function the Pim pipeline should treat as its entry point.
 /// Prefers ONNX entry-point metadata, then `main_graph`, then the only
 /// non-external function if the module is otherwise unambiguous.
 llvm::FailureOr<mlir::func::FuncOp> getPimEntryFunc(mlir::ModuleOp moduleOp);

src/PIM/Common/IR/WeightUtils.hpp

@@ -32,7 +32,7 @@ struct ResolvedWeightView {
 bool hasWeightAlways(mlir::Operation* op);
 
 /// Tags an op as producing a value that should stay materialized as a reusable
-/// weight across later PIM lowering/codegen stages.
+/// weight across later Pim lowering/codegen stages.
 void markWeightAlways(mlir::Operation* op);
 
 bool isSpatialMvmVmmWeightUse(mlir::OpOperand& use);

src/PIM/Common/Support/CheckedArithmetic.cpp

@@ -11,7 +11,7 @@ namespace onnx_mlir::pim {
 namespace {
 
 static void emitCrashMessage(llvm::StringRef fieldName, llvm::StringRef message) {
-  llvm::errs() << "PIM " << fieldName << " " << message << "\n";
+  llvm::errs() << "Pim " << fieldName << " " << message << "\n";
 }
 
 template <typename To, typename From>
@@ -65,7 +65,7 @@ InFlightDiagnostic emitCheckedArithmeticError(Operation* anchor, llvm::StringRef
 }
 
 InFlightDiagnostic emitCheckedArithmeticError(Location loc, llvm::StringRef fieldName, llvm::StringRef message) {
-  return emitError(loc) << "PIM " << fieldName << " " << message;
+  return emitError(loc) << "Pim " << fieldName << " " << message;
 }
 
 FailureOr<int32_t> checkedI32(int64_t value, Operation* anchor, llvm::StringRef fieldName) {
@@ -174,7 +174,7 @@ FailureOr<uint64_t> getCheckedShapedTypeSizeInBytes(ShapedType type, Location lo
 int32_t checkedI32OrCrash(int64_t value, llvm::StringRef fieldName) {
   if (value < std::numeric_limits<int32_t>::min() || value > std::numeric_limits<int32_t>::max()) {
     emitCrashMessage(fieldName, "is outside representable range");
-    llvm_unreachable("PIM checked arithmetic failure");
+    llvm_unreachable("Pim checked arithmetic failure");
   }
   return static_cast<int32_t>(value);
 }
@@ -182,7 +182,7 @@ int32_t checkedI32OrCrash(int64_t value, llvm::StringRef fieldName) {
 int32_t checkedI32OrCrash(uint64_t value, llvm::StringRef fieldName) {
   if (value > static_cast<uint64_t>(std::numeric_limits<int32_t>::max())) {
     emitCrashMessage(fieldName, "is outside representable range");
-    llvm_unreachable("PIM checked arithmetic failure");
+    llvm_unreachable("Pim checked arithmetic failure");
   }
   return static_cast<int32_t>(value);
 }
@@ -190,7 +190,7 @@ int32_t checkedI32OrCrash(uint64_t value, llvm::StringRef fieldName) {
 uint8_t checkedU8OrCrash(uint64_t value, llvm::StringRef fieldName) {
   if (value > static_cast<uint64_t>(std::numeric_limits<uint8_t>::max())) {
     emitCrashMessage(fieldName, "is outside representable range");
-    llvm_unreachable("PIM checked arithmetic failure");
+    llvm_unreachable("Pim checked arithmetic failure");
   }
   return static_cast<uint8_t>(value);
 }
@@ -198,7 +198,7 @@ uint8_t checkedU8OrCrash(uint64_t value, llvm::StringRef fieldName) {
 size_t checkedSizeOrCrash(int64_t value, llvm::StringRef fieldName) {
   if (value < 0) {
     emitCrashMessage(fieldName, "is outside representable range");
-    llvm_unreachable("PIM checked arithmetic failure");
+    llvm_unreachable("Pim checked arithmetic failure");
   }
   return static_cast<size_t>(value);
 }
@@ -206,7 +206,7 @@ size_t checkedSizeOrCrash(int64_t value, llvm::StringRef fieldName) {
 size_t checkedAddOrCrash(size_t lhs, size_t rhs, llvm::StringRef fieldName) {
   if (rhs > std::numeric_limits<size_t>::max() - lhs) {
     emitCrashMessage(fieldName, "addition overflow");
-    llvm_unreachable("PIM checked arithmetic failure");
+    llvm_unreachable("Pim checked arithmetic failure");
   }
   return lhs + rhs;
 }
@@ -214,7 +214,7 @@ size_t checkedAddOrCrash(size_t lhs, size_t rhs, llvm::StringRef fieldName) {
 size_t checkedMulOrCrash(size_t lhs, size_t rhs, llvm::StringRef fieldName) {
   if (lhs != 0 && rhs > std::numeric_limits<size_t>::max() / lhs) {
     emitCrashMessage(fieldName, "multiplication overflow");
-    llvm_unreachable("PIM checked arithmetic failure");
+    llvm_unreachable("Pim checked arithmetic failure");
   }
   return lhs * rhs;
 }

src/PIM/Common/Support/FileSystemUtils.hpp

@@ -4,7 +4,7 @@
 
 namespace onnx_mlir {
 
-/// Returns the directory that should hold PIM artifacts/debug dumps for the
+/// Returns the directory that should hold Pim artifacts/debug dumps for the
 /// current compiler invocation.
 std::string getOutputDir();
 

src/PIM/Compiler/PimBinaryFormat.hpp

@@ -171,19 +171,19 @@ inline Opcode opcodeFromString(llvm::StringRef opName) {
   for (auto [index, name] : llvm::enumerate(kOpcodeNames))
     if (opName == name)
       return static_cast<Opcode>(index);
-  llvm_unreachable("Unsupported PIM binary opcode");
+  llvm_unreachable("Unsupported Pim binary opcode");
 }
 
 inline llvm::StringRef opcodeToString(Opcode opcode) {
   size_t index = static_cast<size_t>(opcode);
-  assert(index < kOpcodeNames.size() && "Unsupported PIM binary opcode");
+  assert(index < kOpcodeNames.size() && "Unsupported Pim binary opcode");
   return kOpcodeNames[index];
 }
 
 inline InstructionRecord makeInstructionRecord(const llvm::json::Object& instruction) {
   InstructionRecord record;
   std::optional<llvm::StringRef> opName = instruction.getString("op");
-  assert(opName && "Missing op field in PIM instruction");
+  assert(opName && "Missing op field in Pim instruction");
   record.opcode = opcodeFromString(*opName);
   const auto& format = kInstructionJsonFormats[static_cast<size_t>(record.opcode)];
   if (format.rd)

src/PIM/Compiler/PimCodeGen.cpp

@@ -125,7 +125,7 @@ static bool isZeroSplatGlobal(mlir::Value value) {
   return false;
 }
 
-// PIM instruction immediates are serialized as signed int32_t fields today
+// Pim instruction immediates are serialized as signed int32_t fields today
 // (`sldi` goes through checkedI32OrCrash), so local addresses must stay within
 // the non-negative int32_t range.
 static FailureOr<size_t> checkedAlignTo(size_t value, size_t alignment, Operation* anchor, StringRef fieldName) {
@@ -141,7 +141,7 @@ static void printMemoryOverflowDiagnostic(const MemoryValueKey& key,
                                           size_t requestedSize,
                                           size_t currentFirstAvailableAddress,
                                           size_t alignedEndAddress) {
-  llvm::errs() << "PIM local memory allocation overflow\n";
+  llvm::errs() << "Pim local memory allocation overflow\n";
   llvm::errs() << "Requested allocation size: " << requestedSize << " bytes\n";
   llvm::errs() << "Current firstAvailableAddress: " << currentFirstAvailableAddress << "\n";
   llvm::errs() << "Aligned end address: " << alignedEndAddress << "\n";
@@ -187,7 +187,7 @@ size_t PimMemory::allocateAddress(size_t size, const MemoryValueKey& key) {
       size,
       firstAvailableAddress,
       succeeded(checkedAlignedEnd) ? *checkedAlignedEnd : kPimLocalMemoryAddressLimit);
-    llvm_unreachable("PIM local memory allocation overflow");
+    llvm_unreachable("Pim local memory allocation overflow");
   }
   firstAvailableAddress = *checkedAlignedEnd;
   return address;
@@ -276,7 +276,7 @@ void PimMemory::allocateCore(const CompiledCoreMemoryPlan& plan, std::optional<u
   }
   else if (*localArenaSize != plan.arenaSize || reportRow.logicalLocalAllocationCount != plan.logicalAllocationCount
            || reportRow.logicalLocalBytes != plan.logicalBytes)
-    llvm_unreachable("inconsistent PIM local-memory plan across core-batch lanes");
+    llvm_unreachable("inconsistent Pim local-memory plan across core-batch lanes");
   for (const CompiledLocalMemoryEntry& entry : plan.entries) {
     MemoryValueKey key = getMemoryValueKey(entry.value, lane);
     ownedMemEntriesMap[key] = entry.memory;
@@ -352,8 +352,8 @@ size_t PimAcceleratorMemory::getValueAddress(mlir::Value value,
     llvm_unreachable("Missing mem entry");
   }
 
-  size_t byteOffset = pim::checkedSizeOrCrash(resolvedAddress->byteOffset, "resolved PIM byte offset");
-  return pim::checkedAddOrCrash(iter->second.address, byteOffset, "resolved PIM address");
+  size_t byteOffset = pim::checkedSizeOrCrash(resolvedAddress->byteOffset, "resolved Pim byte offset");
+  return pim::checkedAddOrCrash(iter->second.address, byteOffset, "resolved Pim address");
 }
 
 llvm::FailureOr<int64_t> PimAcceleratorMemory::getIndexValue(mlir::Value value,
@@ -544,11 +544,23 @@ void PimCodeGen::setupRdRs1(size_t rdAddress, size_t rdOffset, size_t rs1Address
   genSetRegisterImmediateUnsigned(1, pim::checkedAddOrCrash(rs1Address, rs1Offset, "rs1 address"));
 }
 
-void PimCodeGen::setupRdRs1Rs2(
+std::array<uint8_t, 3> PimCodeGen::setupRdRs1Rs2(
   size_t rdAddress, size_t rdOffset, size_t rs1Address, size_t rs1Offset, size_t rs2Address, size_t rs2Offset) const {
-  genSetRegisterImmediateUnsigned(0, pim::checkedAddOrCrash(rdAddress, rdOffset, "rd address"));
-  genSetRegisterImmediateUnsigned(1, pim::checkedAddOrCrash(rs1Address, rs1Offset, "rs1 address"));
-  genSetRegisterImmediateUnsigned(2, pim::checkedAddOrCrash(rs2Address, rs2Offset, "rs2 address"));
+  size_t rd = pim::checkedAddOrCrash(rdAddress, rdOffset, "rd address");
+  size_t rs1 = pim::checkedAddOrCrash(rs1Address, rs1Offset, "rs1 address");
+  size_t rs2 = pim::checkedAddOrCrash(rs2Address, rs2Offset, "rs2 address");
+  genSetRegisterImmediateUnsigned(0, rd);
+  uint8_t rs1Register = 0;
+  if (rd != rs1) {
+    genSetRegisterImmediateUnsigned(1, rs1);
+    rs1Register = 1;
+  }
+  if (rd == rs2)
+    return {0, rs1Register, 0};
+  if (rs1 == rs2)
+    return {0, rs1Register, rs1Register};
+  genSetRegisterImmediateUnsigned(2, rs2);
+  return {0, rs1Register, 2};
 }
 
 void PimCodeGen::emitMemCopyOp(pim_binary::Opcode opcode,
@@ -664,13 +676,13 @@ void PimCodeGen::codeGenVMVOp(pim::PimVMVOp vmvOp, const StaticValueKnowledge& k
   auto sourceType = cast<ShapedType>(vmvOp.getSource().getType());
   int32_t bitwidth = getVectorElementBitwidthOrCrash(sourceType);
   ensureVectorBitwidth(bitwidth, bitwidth);
-  setupRdRs1Rs2(addressOf(vmvOp.getTarget(), knowledge), *targetOffset,
+  auto registers = setupRdRs1Rs2(addressOf(vmvOp.getTarget(), knowledge), *targetOffset,
                                  addressOf(vmvOp.getSource(), knowledge), *sourceOffset, 0, *sourceStride);
   pim_binary::InstructionRecord instruction;
   instruction.opcode = pim_binary::Opcode::vmv;
-  instruction.rd = 0;
-  instruction.r1 = 1;
-  instruction.r2OrImm = 2;
+  instruction.rd = registers[0];
+  instruction.r1 = registers[1];
+  instruction.r2OrImm = registers[2];
   instruction.generic3 = vmvOp.getLength();
   emitInstruction(instruction);
 }
@@ -694,6 +706,8 @@ void PimCodeGen::codeGenSendOp(pim::PimSendOp sendOp, const StaticValueKnowledge
 
 void PimCodeGen::codeGenWaitOp(
     pim::PimWaitOp waitOp, const StaticValueKnowledge& knowledge) const {
+  if (pimDisableSynchronization)
+    return;
   auto eventRegister = indexOf(waitOp.getEventRegister(), knowledge);
   auto waitValue = indexOf(waitOp.getWaitValue(), knowledge);
   assert(succeeded(eventRegister) && succeeded(waitValue)
@@ -710,6 +724,8 @@ void PimCodeGen::codeGenWaitOp(
 
 void PimCodeGen::codeGenSyncOp(
     pim::PimSyncOp syncOp, const StaticValueKnowledge& knowledge) const {
+  if (pimDisableSynchronization)
+    return;
   auto targetCoreId = indexOf(syncOp.getTargetCoreId(), knowledge);
   auto eventRegister = indexOf(syncOp.getEventRegister(), knowledge);
   assert(succeeded(targetCoreId) && succeeded(eventRegister)
@@ -780,12 +796,13 @@ void PimCodeGen::emitBinaryVectorOp(pim_binary::Opcode opcode,
   auto inputType = cast<ShapedType>(lhs.getType());
   ensureVectorBitwidth(getVectorElementBitwidthOrCrash(inputType),
                        getVectorElementBitwidthOrCrash(cast<ShapedType>(output.getType())));
-  setupRdRs1Rs2(addressOf(output, knowledge), 0, addressOf(lhs, knowledge), 0, addressOf(rhs, knowledge), 0);
+  auto registers = setupRdRs1Rs2(
+    addressOf(output, knowledge), 0, addressOf(lhs, knowledge), 0, addressOf(rhs, knowledge), 0);
   pim_binary::InstructionRecord instruction;
   instruction.opcode = opcode;
-  instruction.rd = 0;
-  instruction.r1 = 1;
-  instruction.r2OrImm = 2;
+  instruction.rd = registers[0];
+  instruction.r1 = registers[1];
+  instruction.r2OrImm = registers[2];
   instruction.generic3 = getVectorElementCountOrCrash(inputType);
   emitInstruction(instruction);
 }
@@ -945,7 +962,7 @@ static LogicalResult executeCompiledCorePlan(
       auto step = node.step.evaluate(knowledge);
       auto forOp = cast<mlir::scf::ForOp>(node.op);
       if (failed(lowerBound) || failed(upperBound) || failed(step) || *step <= 0) {
-        forOp.emitOpError("requires statically evaluable scf.for bounds for PIM codegen");
+        forOp.emitOpError("requires statically evaluable scf.for bounds for Pim codegen");
         return failure();
       }
 
@@ -971,7 +988,7 @@ static LogicalResult executeCompiledCorePlan(
       auto condition = node.condition.evaluate(knowledge);
       auto ifOp = cast<mlir::scf::IfOp>(node.op);
       if (failed(condition)) {
-        ifOp.emitOpError("requires statically evaluable scf.if condition for PIM codegen");
+        ifOp.emitOpError("requires statically evaluable scf.if condition for Pim codegen");
         return failure();
       }
 
@@ -985,7 +1002,7 @@ static LogicalResult executeCompiledCorePlan(
       auto selector = node.condition.evaluate(knowledge);
       auto switchOp = cast<mlir::scf::IndexSwitchOp>(node.op);
       if (failed(selector)) {
-        switchOp.emitOpError("requires a statically evaluable scf.index_switch selector for PIM codegen");
+        switchOp.emitOpError("requires a statically evaluable scf.index_switch selector for Pim codegen");
         return failure();
       }
       const llvm::SmallVectorImpl<CompiledCoreNode>* selectedBody = node.defaultBody.get();
@@ -1171,12 +1188,12 @@ OnnxMlirCompilerErrorCodes onnx_mlir::compileToPimCode(ModuleOp& moduleOp, std::
   }
   auto getCompiledProgram = [&](Operation* op) {
     auto it = compiledPrograms.find(op);
-    assert(it != compiledPrograms.end() && "missing compiled PIM core program");
+    assert(it != compiledPrograms.end() && "missing compiled Pim core program");
     return it->second.get();
   };
   auto getMemoryPlan = [&](Operation* op) {
     auto it = memoryPlans.find(op);
-    assert(it != memoryPlans.end() && "missing PIM core memory plan");
+    assert(it != memoryPlans.end() && "missing Pim core memory plan");
     return it->second.get();
   };
 
@@ -1254,7 +1271,7 @@ OnnxMlirCompilerErrorCodes onnx_mlir::compileToPimCode(ModuleOp& moduleOp, std::
       if (failed(weightView)) {
         std::string message;
         llvm::raw_string_ostream os(message);
-        os << "requires a statically resolvable dense global weight view during PIM codegen; weight="
+        os << "requires a statically resolvable dense global weight view during Pim codegen; weight="
            << vmmOp.getWeight() << " type=" << vmmOp.getWeight().getType();
         result.recordDiagnostic(vmmOp, os.str());
         return failure();
@@ -1262,7 +1279,7 @@ OnnxMlirCompilerErrorCodes onnx_mlir::compileToPimCode(ModuleOp& moduleOp, std::
       if (weightView->shape.size() != 2) {
         std::string message;
         llvm::raw_string_ostream os(message);
-        os << "requires a rank-2 matrix weight view during PIM codegen; resolved shape=[";
+        os << "requires a rank-2 matrix weight view during Pim codegen; resolved shape=[";
         llvm::interleaveComma(weightView->shape, os);
         os << "] weight=" << vmmOp.getWeight() << " type=" << vmmOp.getWeight().getType();
         result.recordDiagnostic(vmmOp, os.str());
@@ -1374,7 +1391,7 @@ OnnxMlirCompilerErrorCodes onnx_mlir::compileToPimCode(ModuleOp& moduleOp, std::
   }
   if (diagnostics.hasFailure())
     diagnostics.emitSuppressedSummary(summaryAnchor ? summaryAnchor : moduleOp.getOperation(),
-                                      "PIM codegen diagnostic(s)");
+                                      "Pim codegen diagnostic(s)");
 
   for (size_t jobIndex = 0; jobIndex < jobs.size(); ++jobIndex)
     if (jobResults[jobIndex].status != CompilerSuccess)
@@ -1440,7 +1457,7 @@ OnnxMlirCompilerErrorCodes onnx_mlir::compileToPimCode(ModuleOp& moduleOp, std::
       if (!batchPerCoreRow)
         batchPerCoreRow = result.reportRow;
       else if (!(*batchPerCoreRow == result.reportRow))
-        llvm_unreachable("one PIM core batch produced inconsistent per-core memory reports");
+        llvm_unreachable("one Pim core batch produced inconsistent per-core memory reports");
     }
 
     uint64_t batchReportId = jobs[group.front()].batchReportId.value_or(0);

src/PIM/Compiler/PimCodeGen.hpp

@@ -176,7 +176,7 @@ class PimCodeGen {
   void genSetRegisterImmediateUnsigned(size_t registerNumber, size_t immediate) const;
   void setupRd(size_t rdAddress, size_t rdOffset) const;
   void setupRdRs1(size_t rdAddress, size_t rdOffset, size_t rs1Address, size_t rs1Offset) const;
-  void setupRdRs1Rs2(
+  std::array<uint8_t, 3> setupRdRs1Rs2(
     size_t rdAddress, size_t rdOffset, size_t rs1Address, size_t rs1Offset, size_t rs2Address, size_t rs2Offset) const;
 
   void emitMemCopyOp(pim_binary::Opcode opcode,

src/PIM/Compiler/PimCompilerOptions.cpp

@@ -9,25 +9,25 @@
 namespace onnx_mlir {
 
 llvm::cl::opt<PimEmissionTargetType> pimEmissionTarget(
-  llvm::cl::desc("[Optional] Choose PIM-related target to emit (once selected it will cancel the other targets):"),
-  llvm::cl::values(clEnumVal(EmitSpatial, "Lower model to spatial IR")),
-  llvm::cl::values(clEnumVal(EmitPim, "Lower model to PIM IR")),
-  llvm::cl::values(clEnumVal(EmitPimBufferized, "Lower model to PIM IR and bufferize it")),
-  llvm::cl::values(clEnumVal(EmitPimCodegen, "Lower model to PIM IR and generate code for PIM")),
+  llvm::cl::desc("[Optional] Choose Pim-related target to emit (once selected it will cancel the other targets):"),
+  llvm::cl::values(clEnumVal(EmitSpatial, "Lower model to Spatial IR")),
+  llvm::cl::values(clEnumVal(EmitPim, "Lower model to Pim IR")),
+  llvm::cl::values(clEnumVal(EmitPimBufferized, "Lower model to Pim IR and bufferize it")),
+  llvm::cl::values(clEnumVal(EmitPimCodegen, "Lower model to Pim IR and generate code for Pim")),
   llvm::cl::init(EmitPimCodegen),
   llvm::cl::cat(OnnxMlirOptions));
 
 llvm::cl::opt<PimMemoryReportLevel> pimMemoryReport(
   "pim-memory-report",
-  llvm::cl::desc("Emit a human-readable PIM memory planning report"),
-  llvm::cl::values(clEnumValN(PimMemoryReportNone, "none", "Do not emit any PIM memory planning report")),
-  llvm::cl::values(clEnumValN(PimMemoryReportSummary, "summary", "Emit a concise PIM memory summary")),
+  llvm::cl::desc("Emit a human-readable Pim memory planning report"),
+  llvm::cl::values(clEnumValN(PimMemoryReportNone, "none", "Do not emit any Pim memory planning report")),
+  llvm::cl::values(clEnumValN(PimMemoryReportSummary, "summary", "Emit a concise Pim memory summary")),
   llvm::cl::init(PimMemoryReportSummary),
   llvm::cl::cat(OnnxMlirOptions));
 
 llvm::cl::opt<PimConvLoweringType> pimConvLowering(
   "pim-conv-lowering",
-  llvm::cl::desc("Convolution lowering strategy for PIM"),
+  llvm::cl::desc("Convolution lowering strategy for Pim"),
   llvm::cl::values(clEnumValN(PimConvLoweringAuto, "auto", "Select the Conv lowering strategy automatically")),
   llvm::cl::values(clEnumValN(PimConvLoweringLegacy, "legacy", "Use the legacy explicit-im2col Conv lowering")),
   llvm::cl::values(clEnumValN(PimConvLoweringDepthwise, "depthwise", "Force the depthwise-specialized Conv lowering")),
@@ -55,20 +55,20 @@ llvm::cl::opt<PimSpatialDataflowExportType> pimExportSpatialDataflow(
   llvm::cl::desc("Emit Gephi-importable CSV dataflow reports for Spatial pipeline snapshots"),
   llvm::cl::values(clEnumValN(SpatialDataflowExportNone, "none", "Do not emit Spatial dataflow CSV reports")),
   llvm::cl::values(
-    clEnumValN(SpatialDataflowExportSpatial1, "spatial1", "Emit spatial1 graph dataflow CSV reports")),
+    clEnumValN(SpatialDataflowExportSpatial1, "spatial1", "Emit Spatial1 graph dataflow CSV reports")),
   llvm::cl::values(
-    clEnumValN(SpatialDataflowExportSpatial2, "spatial2", "Emit spatial2 trivially merged graph dataflow CSV reports")),
+    clEnumValN(SpatialDataflowExportSpatial2, "spatial2", "Emit Spatial2 trivially merged graph dataflow CSV reports")),
   llvm::cl::values(
-    clEnumValN(SpatialDataflowExportSpatial3, "spatial3", "Emit spatial3 scheduled dataflow CSV reports")),
+    clEnumValN(SpatialDataflowExportSpatial3, "spatial3", "Emit Spatial3 scheduled dataflow CSV reports")),
   llvm::cl::values(
-    clEnumValN(SpatialDataflowExportSpatial4, "spatial4", "Emit spatial4 realized dataflow CSV reports")),
+    clEnumValN(SpatialDataflowExportSpatial4, "spatial4", "Emit Spatial4 realized dataflow CSV reports")),
   llvm::cl::values(clEnumValN(SpatialDataflowExportAll, "all", "Emit all Spatial dataflow CSV reports")),
   llvm::cl::init(SpatialDataflowExportNone),
   llvm::cl::cat(OnnxMlirOptions));
 
 llvm::cl::opt<bool>
   pimOnlyCodegen("pim-only-codegen",
-                 llvm::cl::desc("Only generate code for PIM (assume input is already in bufferized PIM IR)"),
+                 llvm::cl::desc("Only generate code for Pim (assume input is already in bufferized Pim IR)"),
                  llvm::cl::init(false),
                  llvm::cl::cat(OnnxMlirOptions));
 
@@ -96,21 +96,37 @@ llvm::cl::opt<bool> pimEmitJson("pim-emit-json",
 
 llvm::cl::opt<bool> pimDetectCommunicationDeadlock(
   "pim-detect-communication-deadlock",
-  llvm::cl::desc("Expensively simulate the statically expanded PIM send/receive order at verification time and fail if a blocking communication deadlock is found"),
+  llvm::cl::desc("Expensively simulate statically expanded Pim SEND/RECV and exact-count SYNC/WAIT order at verification time and fail on a blocking deadlock"),
   llvm::cl::init(false),
   llvm::cl::cat(OnnxMlirOptions));
 
 llvm::cl::opt<bool> pimVerifyBufferizationCopyFreedom(
   "pim-verify-bufferization-copy-freedom",
-  llvm::cl::desc("Run the expensive official PIM tensor-copy freedom proof before bufferization"),
+  llvm::cl::desc("Run the expensive official Pim tensor-copy freedom proof before bufferization"),
+  llvm::cl::init(false),
+  llvm::cl::cat(OnnxMlirOptions));
+
+llvm::cl::opt<bool> pimDisableSynchronization(
+  "pim-disable-synchronization",
+  llvm::cl::desc("Omit Pim wait/sync instructions from generated code for performance ablation"),
+  llvm::cl::init(false),
+  llvm::cl::cat(OnnxMlirOptions));
+
+llvm::cl::opt<bool> pimDisableSpatialPlanning(
+  "pim-disable-spatial-planning",
+  llvm::cl::desc("Select the trivial Spatial layout plan for performance ablation"),
   llvm::cl::init(false),
   llvm::cl::cat(OnnxMlirOptions));
 
 llvm::cl::opt<size_t>
-  crossbarSize("crossbar-size", llvm::cl::desc("Width and height of a single crossbar"), llvm::cl::init(128));
+  crossbarSize("crossbar-size",
+               llvm::cl::desc("Width and height of a single crossbar (required for Pim compilation)"),
+               llvm::cl::init(0));
 
 llvm::cl::opt<size_t>
-  crossbarCountInCore("crossbar-count", llvm::cl::desc("Number of crossbars in each core"), llvm::cl::init(64));
+  crossbarCountInCore("crossbar-count",
+                      llvm::cl::desc("Number of crossbars in each core (required for Pim compilation)"),
+                      llvm::cl::init(0));
 
 llvm::cl::opt<size_t> pipelineStages(
   "pipeline",
@@ -119,32 +135,35 @@ llvm::cl::opt<size_t> pipelineStages(
   llvm::cl::cat(OnnxMlirOptions));
 
 llvm::cl::opt<long> coresCount("core-count",
-                               llvm::cl::desc("Number of cores in the chip. Required for PIM compilation."),
+                               llvm::cl::desc("Number of cores in the chip. Required for Pim compilation."),
                                llvm::cl::init(-1));
 
 llvm::cl::opt<std::string> pimTargetConfig(
   "pim-target-config",
-  llvm::cl::desc("PIM target configuration used to construct the Spatial scheduling cost model"),
+  llvm::cl::desc("Pim target configuration used to construct the Spatial scheduling cost model"),
   llvm::cl::init(""),
   llvm::cl::cat(OnnxMlirOptions));
 
-bool hasExplicitPimCoreCount() { return coresCount.getNumOccurrences() != 0; }
-
-void verifyExplicitPimCoreCount() {
-  if (!hasExplicitPimCoreCount())
-    llvm::report_fatal_error("PIM compilation requires an explicit --core-count=<positive integer>");
+void verifyPimCompilerOptions() {
+  if (coresCount.getNumOccurrences() == 0)
+    llvm::report_fatal_error("Pim compilation requires an explicit --core-count=<positive integer>");
   if (coresCount.getValue() <= 0)
-    llvm::report_fatal_error("PIM compilation requires --core-count to be a positive integer");
-}
-
-void verifyPimPipelineStages() {
+    llvm::report_fatal_error("Pim compilation requires --core-count to be a positive integer");
+  if (crossbarSize.getNumOccurrences() == 0)
+    llvm::report_fatal_error("Pim compilation requires an explicit --crossbar-size=<positive integer>");
+  if (crossbarSize.getValue() == 0)
+    llvm::report_fatal_error("Pim compilation requires --crossbar-size to be a positive integer");
+  if (crossbarCountInCore.getNumOccurrences() == 0)
+    llvm::report_fatal_error("Pim compilation requires an explicit --crossbar-count=<positive integer>");
+  if (crossbarCountInCore.getValue() == 0)
+    llvm::report_fatal_error("Pim compilation requires --crossbar-count to be a positive integer");
   if (pipelineStages.getValue() == 0)
-    llvm::report_fatal_error("PIM compilation requires --pipeline to be positive");
-  if (static_cast<size_t>(coresCount.getValue()) % pipelineStages.getValue() != 0)
-    llvm::report_fatal_error("PIM compilation requires --core-count to be divisible by --pipeline");
+    llvm::report_fatal_error("Pim compilation requires --pipeline to be positive");
+  if (static_cast<size_t>(coresCount.getValue()) < pipelineStages.getValue())
+    llvm::report_fatal_error("Pim compilation requires --pipeline not to exceed --core-count");
   if (crossbarCountInCore.getValue()
       > std::numeric_limits<size_t>::max() / pipelineStages.getValue())
-    llvm::report_fatal_error("PIM compilation --crossbar-count * --pipeline overflows");
+    llvm::report_fatal_error("Pim compilation --crossbar-count * --pipeline overflows");
 }
 
 } // namespace onnx_mlir

src/PIM/Compiler/PimCompilerOptions.hpp

@@ -59,6 +59,8 @@ extern llvm::cl::opt<bool> pimEmitJson;
 extern llvm::cl::opt<bool> pimReportConvLowering;
 extern llvm::cl::opt<bool> pimDetectCommunicationDeadlock;
 extern llvm::cl::opt<bool> pimVerifyBufferizationCopyFreedom;
+extern llvm::cl::opt<bool> pimDisableSynchronization;
+extern llvm::cl::opt<bool> pimDisableSpatialPlanning;
 
 extern llvm::cl::opt<size_t> crossbarSize;
 extern llvm::cl::opt<size_t> crossbarCountInCore;
@@ -68,8 +70,6 @@ extern llvm::cl::opt<std::string> pimTargetConfig;
 extern llvm::cl::opt<uint64_t> pimConvIm2colMaxElements;
 extern llvm::cl::opt<uint64_t> pimConvStreamChunkPositions;
 
-bool hasExplicitPimCoreCount();
-void verifyExplicitPimCoreCount();
-void verifyPimPipelineStages();
+void verifyPimCompilerOptions();
 
 } // namespace onnx_mlir

src/PIM/Compiler/PimCompilerUtils.cpp

@@ -97,7 +97,7 @@ spatial::ConvLoweringStrategy getSpatialConvLoweringStrategy(PimConvLoweringType
   case PimConvLoweringInputKTiled: return spatial::ConvLoweringStrategy::InputKTiled;
   case PimConvLoweringTiled2D: return spatial::ConvLoweringStrategy::Tiled2D;
   }
-  llvm_unreachable("unknown PIM Conv lowering strategy");
+  llvm_unreachable("unknown Pim Conv lowering strategy");
 }
 
 spatial::SpatialDataflowExportStage getPimSpatialDataflowExportStage(
@@ -110,7 +110,7 @@ spatial::SpatialDataflowExportStage getPimSpatialDataflowExportStage(
   case SpatialDataflowExportSpatial4: return spatial::SpatialDataflowExportStage::Spatial4;
   case SpatialDataflowExportAll:      return spatial::SpatialDataflowExportStage::All;
   }
-  llvm_unreachable("unknown PIM Spatial dataflow export stage");
+  llvm_unreachable("unknown Pim Spatial dataflow export stage");
 }
 
 spatial::SpatialTargetResources getPimSpatialTargetResources(const spatial::SchedulingTarget& target) {
@@ -120,7 +120,7 @@ spatial::SpatialTargetResources getPimSpatialTargetResources(const spatial::Sche
   resources.processorCount = target.processorCount;
   resources.vectorWidth = target.vectorWidth;
   if (failed(resources.verify()))
-    llvm::report_fatal_error("PIM target resources are incomplete");
+    llvm::report_fatal_error("Pim target resources are incomplete");
   return resources;
 }
 
@@ -138,7 +138,7 @@ const llvm::json::Object& requireObject(const llvm::json::Object& object,
                                         llvm::StringRef path) {
   const llvm::json::Object* nested = object.getObject(key);
   if (!nested)
-    llvm::report_fatal_error("PIM target config is missing object '" + path + "." + key + "'");
+    llvm::report_fatal_error("Pim target config is missing object '" + path + "." + key + "'");
   return *nested;
 }
 
@@ -151,7 +151,7 @@ Cost getConfigCost(const llvm::json::Object& object,
     return fallback;
   if (!std::isfinite(*number) || *number < 0.0 || (!allowZero && *number == 0.0)
       || *number > static_cast<double>(std::numeric_limits<Cost>::max()))
-    llvm::report_fatal_error("PIM target config field '" + key + "' must be a valid positive number");
+    llvm::report_fatal_error("Pim target config field '" + key + "' must be a valid positive number");
   return static_cast<Cost>(std::ceil(*number));
 }
 
@@ -159,11 +159,11 @@ std::pair<size_t, size_t> getConfigPair(const llvm::json::Object& object,
                                         llvm::StringRef key) {
   const llvm::json::Array* values = object.getArray(key);
   if (!values || values->size() != 2)
-    llvm::report_fatal_error("PIM target config field '" + key + "' must contain two integers");
+    llvm::report_fatal_error("Pim target config field '" + key + "' must contain two integers");
   std::optional<int64_t> first = (*values)[0].getAsInteger();
   std::optional<int64_t> second = (*values)[1].getAsInteger();
   if (!first || !second || *first <= 0 || *second <= 0)
-    llvm::report_fatal_error("PIM target config field '" + key + "' must contain two positive integers");
+    llvm::report_fatal_error("Pim target config field '" + key + "' must contain two positive integers");
   return {static_cast<size_t>(*first), static_cast<size_t>(*second)};
 }
 
@@ -174,7 +174,7 @@ void loadPimInterProcessorLatencies(
     network.getString("net_config_file_path");
   if (!filename)
     llvm::report_fatal_error(
-      "PIM target config is missing network latency file path");
+      "Pim target config is missing network latency file path");
 
   llvm::SmallString<256> networkPath(*filename);
   if (!llvm::sys::path::is_absolute(networkPath)) {
@@ -187,19 +187,19 @@ void loadPimInterProcessorLatencies(
   auto buffer = llvm::MemoryBuffer::getFile(networkPath);
   if (!buffer)
     llvm::report_fatal_error(
-      llvm::Twine("failed to read PIM network config '")
+      llvm::Twine("failed to read Pim network config '")
       + networkPath + "': " + buffer.getError().message());
   auto parsed = llvm::json::parse(buffer.get()->getBuffer());
   if (!parsed)
     llvm::report_fatal_error(
-      llvm::Twine("failed to parse PIM network config '")
+      llvm::Twine("failed to parse Pim network config '")
       + networkPath + "': " + llvm::toString(parsed.takeError()));
   const llvm::json::Object* root = parsed->getAsObject();
   const llvm::json::Object* latencies =
     root ? root->getObject("latency") : nullptr;
   if (!latencies)
     llvm::report_fatal_error(
-      "PIM network config is missing its latency matrix");
+      "Pim network config is missing its latency matrix");
 
   target.interProcessorLatencyNs.assign(
     target.processorCount * target.processorCount, 0);
@@ -210,7 +210,7 @@ void loadPimInterProcessorLatencies(
     const llvm::json::Object* row = latencies->getObject(sourceKey);
     if (!row)
       llvm::report_fatal_error(
-        llvm::Twine("PIM network config is missing latency row ")
+        llvm::Twine("Pim network config is missing latency row ")
         + sourceKey);
     for (size_t destination = 0;
          destination < target.processorCount; ++destination) {
@@ -220,7 +220,7 @@ void loadPimInterProcessorLatencies(
       std::optional<double> latency = row->getNumber(destinationKey);
       if (!latency || !std::isfinite(*latency) || *latency <= 0.0)
         llvm::report_fatal_error(
-          llvm::Twine("PIM network config is missing latency ")
+          llvm::Twine("Pim network config is missing latency ")
           + sourceKey + " -> " + destinationKey);
       Cost roundedLatency = static_cast<Cost>(std::ceil(*latency));
       target.interProcessorLatencyNs[
@@ -244,17 +244,17 @@ spatial::SchedulingTarget getPimSchedulingTarget() {
   auto buffer = llvm::MemoryBuffer::getFile(pimTargetConfig);
   if (!buffer)
     llvm::report_fatal_error(
-      llvm::Twine("failed to read PIM target config '")
+      llvm::Twine("failed to read Pim target config '")
       + pimTargetConfig.getValue() + "': " + buffer.getError().message());
   auto parsed = llvm::json::parse(buffer.get()->getBuffer());
   if (!parsed)
     llvm::report_fatal_error(
-      llvm::Twine("failed to parse PIM target config '")
+      llvm::Twine("failed to parse Pim target config '")
       + pimTargetConfig.getValue() + "': "
       + llvm::toString(parsed.takeError()));
   const llvm::json::Object* root = parsed->getAsObject();
   if (!root)
-    llvm::report_fatal_error("PIM target config must contain a JSON object");
+    llvm::report_fatal_error("Pim target config must contain a JSON object");
 
   const llvm::json::Object& chip = requireObject(*root, "chip_config", "root");
   const llvm::json::Object& core = requireObject(chip, "core_config", "chip_config");
@@ -267,7 +267,7 @@ spatial::SchedulingTarget getPimSchedulingTarget() {
 
   std::optional<int64_t> coreCount = chip.getInteger("core_cnt");
   if (!coreCount || *coreCount <= 0)
-    llvm::report_fatal_error("PIM target config field 'core_cnt' must be a positive integer");
+    llvm::report_fatal_error("Pim target config field 'core_cnt' must be a positive integer");
   target.processorCount = static_cast<size_t>(*coreCount);
   target.residentWeightCapacity =
     getConfigCost(matrix, "xbar_array_count", target.residentWeightCapacity);
@@ -278,7 +278,7 @@ spatial::SchedulingTarget getPimSchedulingTarget() {
       || target.residentWeightCapacity != crossbarCountInCore.getValue()
       || target.matrixRows != crossbarSize.getValue()
       || target.matrixColumns != crossbarSize.getValue())
-    llvm::report_fatal_error("PIM target config resources do not match --core-count, "
+    llvm::report_fatal_error("Pim target config resources do not match --core-count, "
                              "--crossbar-count, and --crossbar-size");
   loadPimInterProcessorLatencies(target, network);
 
@@ -331,8 +331,7 @@ void addPassesPim(OwningOpRef<ModuleOp>& module,
                   PassManager& pm,
                   EmissionTargetType& emissionTarget,
                   std::string outputNameNoExt) {
-  verifyExplicitPimCoreCount();
-  verifyPimPipelineStages();
+  verifyPimCompilerOptions();
   spatial::SchedulingTarget schedulingTarget = getPimSchedulingTarget();
   spatial::SpatialTargetResources targetResources = getPimSpatialTargetResources(schedulingTarget);
 
@@ -352,7 +351,8 @@ void addPassesPim(OwningOpRef<ModuleOp>& module,
     spatial::SpatialDataflowExportStage exportStage =
       getPimSpatialDataflowExportStage(pimExportSpatialDataflow.getValue());
     pm.addPass(createONNXToSpatialPass(targetResources, planningOptions));
-    pm.addPass(createSpatialLayoutPlanningPass(targetResources));
+    pm.addPass(createSpatialLayoutPlanningPass(
+      targetResources, pimDisableSpatialPlanning.getValue()));
     pm.addPass(createLowerSpatialPlansPass(targetResources, planningOptions, exportStage));
     pm.addPass(createTrivialGraphComputeMergePass(
       schedulingTarget.residentWeightCapacity, exportStage));

src/PIM/Compiler/PimCoreProgram.cpp

@@ -46,7 +46,7 @@ static LogicalResult compileCoreEmissionPlan(Block& block, SmallVectorImpl<Compi
       auto upper = compileIndexExpr(forOp.getUpperBound());
       auto step = compileIndexExpr(forOp.getStep());
       if (failed(lower) || failed(upper) || failed(step)) {
-        forOp.emitOpError("requires statically evaluable scf.for bounds for PIM codegen");
+        forOp.emitOpError("requires statically evaluable scf.for bounds for Pim codegen");
         return failure();
       }
       CompiledCoreNode node;
@@ -63,7 +63,7 @@ static LogicalResult compileCoreEmissionPlan(Block& block, SmallVectorImpl<Compi
     if (auto ifOp = dyn_cast<scf::IfOp>(op)) {
       auto condition = compileIndexExpr(ifOp.getCondition());
       if (failed(condition)) {
-        ifOp.emitOpError("requires statically evaluable scf.if condition for PIM codegen");
+        ifOp.emitOpError("requires statically evaluable scf.if condition for Pim codegen");
         return failure();
       }
       CompiledCoreNode node;
@@ -82,7 +82,7 @@ static LogicalResult compileCoreEmissionPlan(Block& block, SmallVectorImpl<Compi
     if (auto switchOp = dyn_cast<scf::IndexSwitchOp>(op)) {
       auto selector = compileIndexExpr(switchOp.getArg());
       if (failed(selector)) {
-        switchOp.emitOpError("requires a statically evaluable scf.index_switch selector for PIM codegen");
+        switchOp.emitOpError("requires a statically evaluable scf.index_switch selector for Pim codegen");
         return failure();
       }
       CompiledCoreNode node;

src/PIM/Conversion/ONNXToSpatial/Common/ComputeRegionBuilder.hpp

@@ -249,7 +249,7 @@ auto createEmptySpatGraphComputeBatch(RewriterT& rewriter,
   if (laneCount <= 0 || laneCount > std::numeric_limits<int32_t>::max())
     return mlir::FailureOr<spatial::SpatGraphComputeBatch>(mlir::failure());
 
-  auto laneCountAttr = pim::getCheckedI32Attr(rewriter, loc, laneCount, "spatial compute_batch lane count");
+  auto laneCountAttr = pim::getCheckedI32Attr(rewriter, loc, laneCount, "Spatial compute_batch lane count");
   if (mlir::failed(laneCountAttr))
     return mlir::FailureOr<spatial::SpatGraphComputeBatch>(mlir::failure());
 

src/PIM/Conversion/ONNXToSpatial/Common/ShapeTilingUtils.hpp

@@ -25,7 +25,7 @@ llvm::SmallVector<mlir::Value> sliceVector(const mlir::Value& vectorToSlice,
                                            mlir::Location loc);
 
 /// Partitions one logical vector into per-core crossbar-sized slices using the
-/// current PIM target geometry.
+/// current Pim target geometry.
 llvm::DenseMap<CoreId, llvm::SmallVector<mlir::Value>> sliceVectorPerCrossbarPerCore(
   const mlir::Value& vectorToSlice,
   mlir::PatternRewriter& rewriter,

src/PIM/Conversion/ONNXToSpatial/Passes/Transforms/LowerSpatialPlansPass.cpp

@@ -46,7 +46,7 @@ struct LowerSpatialPlansPass final
     }
     auto entryFunc = getPimEntryFunc(moduleOp);
     if (failed(entryFunc)) {
-      moduleOp.emitError("failed to locate the PIM entry function during LowerSpatialPlans");
+      moduleOp.emitError("failed to locate the Pim entry function during LowerSpatialPlans");
       signalPassFailure();
       return;
     }

src/PIM/Conversion/ONNXToSpatial/Passes/Transforms/ONNXToSpatialPass.cpp

@@ -158,7 +158,7 @@ void ONNXToSpatialPass::runOnOperation() {
 
   auto entryFunc = getPimEntryFunc(moduleOp);
   if (failed(entryFunc)) {
-    moduleOp.emitError("failed to locate the PIM entry function during ONNX-to-Spatial lowering");
+    moduleOp.emitError("failed to locate the Pim entry function during ONNX-to-Spatial lowering");
     signalPassFailure();
     return;
   }
@@ -245,7 +245,7 @@ void ONNXToSpatialPass::runOnOperation() {
   RewritePatternSet postPatterns(ctx);
   populatePostPatterns(postPatterns, ctx);
   if (failed(applyPartialConversion(*entryFunc, postTarget, std::move(postPatterns)))) {
-    moduleOp.emitError("failed to normalize weight-like Spatial compute operands before Spatial-to-PIM lowering");
+    moduleOp.emitError("failed to normalize weight-like Spatial compute operands before Spatial-to-Pim lowering");
     signalPassFailure();
     return;
   }

src/PIM/Conversion/ONNXToSpatial/Passes/Transforms/SpatialLayoutPlanningPass.cpp

@@ -42,8 +42,9 @@ static SmallVector<spatial::PhysicalLayout> getOperandLayouts(
 class SpatialLayoutAnalysis {
 public:
   SpatialLayoutAnalysis(func::FuncOp funcOp,
-                        const spatial::SpatialTargetResources& target)
-      : funcOp(funcOp), target(target) {}
+                        const spatial::SpatialTargetResources& target,
+                        bool selectTrivialPlan)
+      : funcOp(funcOp), target(target), selectTrivialPlan(selectTrivialPlan) {}
 
   FailureOr<SpatialLayoutSelection> run() {
     SpatialLayoutSelection selection;
@@ -56,6 +57,9 @@ public:
       selection.selectedAlternative[&op] = 0;
     }
 
+    if (selectTrivialPlan)
+      return selection;
+
     const size_t maxRounds = 2 * planOps.size() + 1;
     for (size_t round = 0; round < maxRounds; ++round) {
       bool changed = false;
@@ -168,6 +172,7 @@ private:
 
   func::FuncOp funcOp;
   const spatial::SpatialTargetResources& target;
+  bool selectTrivialPlan;
 };
 
 static LogicalResult materializeMismatchedUses(
@@ -251,8 +256,9 @@ struct SpatialLayoutPlanningPass final
   }
 
   SpatialLayoutPlanningPass() = default;
-  explicit SpatialLayoutPlanningPass(const spatial::SpatialTargetResources& target)
-  : target(target), hasTarget(true) {}
+  SpatialLayoutPlanningPass(const spatial::SpatialTargetResources& target,
+                            bool selectTrivialPlan)
+  : target(target), selectTrivialPlan(selectTrivialPlan), hasTarget(true) {}
 
   void runOnOperation() override {
     ModuleOp moduleOp = getOperation();
@@ -263,13 +269,13 @@ struct SpatialLayoutPlanningPass final
     }
     auto entryFunc = getPimEntryFunc(moduleOp);
     if (failed(entryFunc)) {
-      moduleOp.emitError("failed to locate the PIM entry function during Spatial layout planning");
+      moduleOp.emitError("failed to locate the Pim entry function during Spatial layout planning");
       signalPassFailure();
       return;
     }
 
     func::FuncOp funcOp = *entryFunc;
-    SpatialLayoutAnalysis analysis(funcOp, target);
+    SpatialLayoutAnalysis analysis(funcOp, target, selectTrivialPlan);
     FailureOr<SpatialLayoutSelection> selection = analysis.run();
     if (failed(selection)) {
       signalPassFailure();
@@ -301,6 +307,7 @@ struct SpatialLayoutPlanningPass final
   }
 
   spatial::SpatialTargetResources target;
+  bool selectTrivialPlan = false;
   bool hasTarget = false;
 };
 
@@ -311,8 +318,8 @@ std::unique_ptr<Pass> createSpatialLayoutPlanningPass() {
 }
 
 std::unique_ptr<Pass> createSpatialLayoutPlanningPass(
-    const spatial::SpatialTargetResources& target) {
-  return std::make_unique<SpatialLayoutPlanningPass>(target);
+    const spatial::SpatialTargetResources& target, bool selectTrivialPlan) {
+  return std::make_unique<SpatialLayoutPlanningPass>(target, selectTrivialPlan);
 }
 
 } // namespace onnx_mlir

src/PIM/Conversion/ONNXToSpatial/Patterns/Math/Conv.cpp

@@ -199,7 +199,7 @@ static bool writeConvLoweringReport(const ConvLoweringReportEntry& entry,
     return false;
   }
 
-  reportFile << "# PIM Conv Lowering Report (bounded to 512 rows)\n\n";
+  reportFile << "# Pim conv lowering report (bounded to 512 rows)\n\n";
   reportFile << "## Plan selection\n";
   writeConvReportTableHeader(reportFile, "Selector");
   bool realizationSectionStarted = false;

src/PIM/Conversion/ONNXToSpatial/Patterns/Math/ReduceMean.cpp

@@ -370,6 +370,14 @@ struct ReduceMeanToSpatialCompute : OpConversionPattern<ReduceMeanOp> {
     Location loc = reduceMeanOp.getLoc();
     RankedTensorType leafType = getAllOnesType(inputType, resultType.getElementType());
     RankedTensorType keepdimsType = getKeepdimsType(inputType, resultType.getElementType(), reducedAxes);
+    if (semantics->keepdims != 0 && inputType.getRank() == 4
+        && inputType.getDimSize(0) == 1 && semantics->axes == ArrayRef<int64_t>({2, 3})
+        && resultType == keepdimsType) {
+      auto plan = spatial::SpatGlobalAveragePoolPlanOp::create(
+        rewriter, loc, resultType, adaptor.getData(), spatial::getNCHWLayout(rewriter.getContext()));
+      rewriter.replaceOp(reduceMeanOp, plan.getResult());
+      return success();
+    }
     int64_t laneCount = 1;
     for (auto [dim, isReduced] : llvm::zip_equal(keepdimsType.getShape(), reducedAxes)) {
       if (isReduced)

src/PIM/Conversion/SpatialToPim/BatchCoreLoweringPatterns.cpp

@@ -307,7 +307,7 @@ LogicalResult raptor::SpatialToPimPass::lowerComputeBatchOp(spatial::SpatSchedul
       "resultful compute_batch lowering currently requires a spat.in_parallel terminator");
   }
 
-  auto coreIds = getRequiredScheduledBatchCoreIds(computeBatchOp, "spatial compute_batch core id");
+  auto coreIds = getRequiredScheduledBatchCoreIds(computeBatchOp, "Spatial compute_batch core id");
   if (failed(coreIds))
     return failure();
   SmallVector<Value> batchWeights(computeBatchOp.getWeights().begin(), computeBatchOp.getWeights().end());
@@ -317,7 +317,7 @@ LogicalResult raptor::SpatialToPimPass::lowerComputeBatchOp(spatial::SpatSchedul
 
   rewriter.setInsertionPointAfter(computeBatchOp);
   auto laneCountAttr = pim::getCheckedI32Attr(
-    rewriter, computeBatchOp, static_cast<uint64_t>(computeBatchOp.getLaneCount()), "pim core_batch lane count");
+    rewriter, computeBatchOp, static_cast<uint64_t>(computeBatchOp.getLaneCount()), "Pim core_batch lane count");
   if (failed(laneCountAttr))
     return failure();
   auto coreBatchOp =

src/PIM/Conversion/SpatialToPim/CoreLoweringPatterns.cpp

@@ -373,9 +373,6 @@ LogicalResult raptor::SpatialToPimPass::lowerComputeOp(spatial::SpatScheduledCom
           rewriter, receiveOp->getLoc(), outputBuffer.getType(), zero,
           hostWaitLoad.getHostOffset(), outputBuffer, *hostBuffer, *sizeAttr)
           .getOutput();
-        PimSyncOp::create(
-          rewriter, receiveOp->getLoc(), hostWaitLoad.getSourceCoreId(),
-          hostWaitLoad.getAcknowledgementEventRegister());
       } else {
         received = PimReceiveOp::create(
           rewriter, receiveOp->getLoc(), outputBuffer.getType(), outputBuffer,
@@ -410,7 +407,7 @@ LogicalResult raptor::SpatialToPimPass::lowerComputeOp(spatial::SpatScheduledCom
         continue;
     }
 
-    return computeOp.emitOpError("has an unsupported remaining result use during Spatial-to-PIM lowering");
+    return computeOp.emitOpError("has an unsupported remaining result use during Spatial-to-Pim lowering");
   }
 
   rewriter.setInsertionPoint(yieldOp);
@@ -420,7 +417,7 @@ LogicalResult raptor::SpatialToPimPass::lowerComputeOp(spatial::SpatScheduledCom
   if (!computeOp.getWeights().empty())
     computeWeights.append(computeOp.getWeights().begin(), computeOp.getWeights().end());
   rewriter.setInsertionPointAfter(computeOp);
-  auto checkedCoreId = getRequiredScheduledCoreId(computeOp, "spatial compute core id");
+  auto checkedCoreId = getRequiredScheduledCoreId(computeOp, "Spatial compute core id");
   if (failed(checkedCoreId))
     return failure();
   auto coreIdAttr = pim::getCheckedI32Attr(rewriter, computeOp, static_cast<int64_t>(*checkedCoreId), "pim core id");

src/PIM/Conversion/SpatialToPim/Patterns/ChannelLowering.cpp

@@ -152,10 +152,6 @@ struct HostWaitLoadLowering : OpRewritePattern<spatial::SpatHostWaitLoadOp> {
       Value output = pim::PimMemCopyHostToDevOp::create(
         rewriter, op.getLoc(), outputBuffer.getType(), zero,
         op.getHostOffset(), outputBuffer, *hostBuffer, sizeAttr).getOutput();
-      auto sync = pim::PimSyncOp::create(
-        rewriter, op.getLoc(), op.getSourceCoreId(),
-        op.getAcknowledgementEventRegister());
-      copyRaptorDebugAttrs(op.getOperation(), sync.getOperation());
       return output;
     });
   }

src/PIM/Conversion/SpatialToPim/ReturnPathNormalization.cpp

@@ -734,7 +734,7 @@ raptor::SpatialToPimPass::ReturnPathLoweringResult raptor::SpatialToPimPass::low
     auto storedType = dyn_cast<RankedTensorType>(storedValue.getType());
     if (!storedType) {
       producerOp->emitOpError(
-        "has an unsupported non-ranked concat-return helper yield during Spatial-to-PIM lowering");
+        "has an unsupported non-ranked concat-return helper yield during Spatial-to-Pim lowering");
       return ReturnPathLoweringResult::Failure;
     }
     rewriter.setInsertionPointAfterValue(storedValue);
@@ -748,7 +748,7 @@ raptor::SpatialToPimPass::ReturnPathLoweringResult raptor::SpatialToPimPass::low
       SmallVector<int64_t> destinationIndices;
       if (failed(mapIndicesThroughHelperChain(
             sourceIndices, concatReturnUse->concatShape, concatReturnUse->helperChain, destinationIndices))) {
-        producerOp->emitOpError("has an unsupported concat-return helper chain during Spatial-to-PIM lowering");
+        producerOp->emitOpError("has an unsupported concat-return helper chain during Spatial-to-Pim lowering");
         return ReturnPathLoweringResult::Failure;
       }
 

src/PIM/Conversion/SpatialToPim/SpatialToPimPass.cpp

@@ -88,7 +88,7 @@ void onnx_mlir::raptor::SpatialToPimPass::runOnOperation() {
   operationsToRemove.clear();
   ModuleOp moduleOp = getOperation();
   if (!hasTarget || failed(targetResources.verify())) {
-    moduleOp.emitError("Spatial-to-PIM lowering requires valid injected target resources");
+    moduleOp.emitError("Spatial-to-Pim lowering requires valid injected target resources");
     signalPassFailure();
     return;
   }
@@ -96,7 +96,7 @@ void onnx_mlir::raptor::SpatialToPimPass::runOnOperation() {
 
   auto entryFunc = getPimEntryFunc(moduleOp);
   if (failed(entryFunc)) {
-    moduleOp.emitError("failed to locate the PIM entry function during Spatial-to-PIM lowering");
+    moduleOp.emitError("failed to locate the Pim entry function during Spatial-to-Pim lowering");
     signalPassFailure();
     return;
   }
@@ -135,7 +135,7 @@ void onnx_mlir::raptor::SpatialToPimPass::runOnOperation() {
   RewritePatternSet initialPatterns(ctx);
   populateInitialPatterns(initialPatterns);
   if (failed(applyPartialConversion(moduleOp, target, std::move(initialPatterns)))) {
-    moduleOp.emitError("failed to lower required Spatial ops to the initial PIM form");
+    moduleOp.emitError("failed to lower required Spatial ops to the initial Pim form");
     signalPassFailure();
     return;
   }
@@ -153,7 +153,7 @@ void onnx_mlir::raptor::SpatialToPimPass::runOnOperation() {
   auto returnOp = cast<func::ReturnOp>(funcOp.front().getTerminator());
   addReturnOutputBuffers(returnOp, rewriter);
   if (failed(allocateAndInitializeCoreLocalVariables(funcOp, rewriter))) {
-    funcOp.emitOpError("failed to allocate or initialize core-local tensors during Spatial-to-PIM lowering");
+    funcOp.emitOpError("failed to allocate or initialize core-local tensors during Spatial-to-Pim lowering");
     signalPassFailure();
     return;
   }
@@ -285,7 +285,7 @@ void onnx_mlir::raptor::SpatialToPimPass::runOnOperation() {
   RewritePatternSet communicationPatterns(ctx);
   populateChannelLoweringPatterns(communicationPatterns);
   if (failed(applyFullConversion(funcOp, communicationTarget, std::move(communicationPatterns)))) {
-    funcOp.emitOpError("failed to lower Spatial communication ops to PIM communication ops");
+    funcOp.emitOpError("failed to lower Spatial communication ops to Pim communication ops");
     signalPassFailure();
     return;
   }

src/PIM/Conversion/SpatialToPim/SpatialToPimPass.hpp

@@ -26,7 +26,7 @@ namespace raptor {
 struct SpatialToPimPass : mlir::PassWrapper<SpatialToPimPass, mlir::OperationPass<mlir::ModuleOp>> {
   MLIR_DEFINE_EXPLICIT_INTERNAL_INLINE_TYPE_ID(SpatialToPimPass)
   llvm::StringRef getArgument() const override { return "convert-spatial-to-pim"; }
-  llvm::StringRef getDescription() const override { return "Lower Spatial ops to PIM-ready format"; }
+  llvm::StringRef getDescription() const override { return "Lower Spatial ops to Pim-ready format"; }
 
   SpatialToPimPass() = default;
   explicit SpatialToPimPass(const spatial::SpatialTargetResources& target)

src/PIM/Dialect/Pim/Passes/Transforms/Bufferization/PimBufferizationPass.cpp

@@ -402,7 +402,7 @@ static LogicalResult verifyPimCoresNeedNoTensorCopies(
 
   bufferization::BufferizationState state;
   if (failed(bufferization::insertTensorCopies(*clone, options, state))) {
-    moduleOp.emitError("official one-shot analysis failed while verifying PIM core copy freedom");
+    moduleOp.emitError("official one-shot analysis failed while verifying Pim core copy freedom");
     return failure();
   }
 
@@ -415,10 +415,10 @@ static LogicalResult verifyPimCoresNeedNoTensorCopies(
     Operation* requiredBy = alloc->getUsers().empty()
       ? alloc.getOperation() : *alloc->getUsers().begin();
     diagnostics.report(requiredBy, [](Operation* op) {
-      op->emitOpError("official one-shot bufferization requires a tensor copy inside a PIM core");
+      op->emitOpError("official one-shot bufferization requires a tensor copy inside a Pim core");
     });
   });
-  diagnostics.emitSuppressedSummary(moduleOp, "required PIM core tensor copies");
+  diagnostics.emitSuppressedSummary(moduleOp, "required Pim core tensor copies");
   return success(!diagnostics.hasFailure());
 }
 
@@ -440,7 +440,7 @@ static LogicalResult runOneShotPimBufferization(
   bufferization::BufferizationState state;
   if (failed(bufferization::insertTensorCopies(moduleOp, hostOptions, state))
       || failed(bufferization::bufferizeModuleOp(moduleOp, options, state))) {
-    moduleOp.emitError("Failed to bufferize PIM and Spatial ops");
+    moduleOp.emitError("Failed to bufferize Pim and Spatial ops");
     return failure();
   }
   return success();
@@ -478,7 +478,7 @@ static LogicalResult verifyContiguousRuntimeOperands(ModuleOp moduleOp) {
           if (succeeded(resolveContiguousAddress(operand, knowledge)) || succeeded(compileContiguousAddressExpr(operand)))
             return;
           op.emitOpError() << "operand #" << operandIndex
-                           << " is not backed by contiguous addressable storage after PIM bufferization";
+                           << " is not backed by contiguous addressable storage after Pim bufferization";
           hasFailure = true;
         };
 
@@ -552,7 +552,7 @@ static LogicalResult verifyContiguousRuntimeOperands(ModuleOp moduleOp) {
   });
 
   if (hasFailure) {
-    moduleOp.emitError("PIM bufferization must fully normalize executable runtime operand contiguity before codegen");
+    moduleOp.emitError("Pim bufferization must fully normalize executable runtime operand contiguity before codegen");
     return failure();
   }
   return success();
@@ -589,7 +589,7 @@ static LogicalResult verifyPimCopyAddressSpaces(ModuleOp moduleOp) {
   });
   if (failureCount != 0)
     moduleOp.emitError() << "found " << failureCount
-                         << " PIM copy address-space violation(s); the first is reported above";
+                         << " Pim copy address-space violation(s); the first is reported above";
   return success(failureCount == 0);
 }
 
@@ -673,7 +673,7 @@ static LogicalResult normalizePimMemory(ModuleOp moduleOp, func::FuncOp funcOp)
   GreedyRewriteConfig contiguityConfig;
   contiguityConfig.enableFolding(false);
   if (failed(applyPatternsGreedily(moduleOp, std::move(contiguityPatterns), contiguityConfig))) {
-    moduleOp.emitError("failed to normalize PIM copy contiguity during bufferization");
+    moduleOp.emitError("failed to normalize Pim copy contiguity during bufferization");
     return failure();
   }
   annotateWeightsMemrefs(moduleOp, funcOp);
@@ -684,7 +684,7 @@ static LogicalResult normalizePimMemory(ModuleOp moduleOp, func::FuncOp funcOp)
 static FailureOr<func::FuncOp> requirePimEntryFunc(ModuleOp moduleOp, StringRef phase) {
   auto entryFunc = getPimEntryFunc(moduleOp);
   if (failed(entryFunc)) {
-    moduleOp.emitError("failed to locate the PIM entry function during ") << phase;
+    moduleOp.emitError("failed to locate the Pim entry function during ") << phase;
     return failure();
   }
   return *entryFunc;
@@ -701,12 +701,12 @@ struct PimBufferizationPreparationPass
 
   StringRef getArgument() const override { return "pim-bufferization-preparation"; }
   StringRef getDescription() const override {
-    return "Prepare writable tensor destinations for PIM one-shot bufferization.";
+    return "Prepare writable tensor destinations for Pim one-shot bufferization.";
   }
 
   void runOnOperation() final {
     ModuleOp moduleOp = getOperation();
-    auto funcOp = requirePimEntryFunc(moduleOp, "PIM bufferization preparation");
+    auto funcOp = requirePimEntryFunc(moduleOp, "Pim bufferization preparation");
     if (failed(funcOp)) {
       signalPassFailure();
       return;
@@ -725,7 +725,7 @@ struct PimOneShotBufferizationPass
 
   StringRef getArgument() const override { return "pim-one-shot-bufferization"; }
   StringRef getDescription() const override {
-    return "Run one-shot bufferization for PIM and Spatial tensors.";
+    return "Run one-shot bufferization for Pim and Spatial tensors.";
   }
 
   void runOnOperation() final {
@@ -740,12 +740,12 @@ struct PimMemoryNormalizationPass
 
   StringRef getArgument() const override { return "pim-memory-normalization"; }
   StringRef getDescription() const override {
-    return "Normalize PIM memory copies and verify addressable operands.";
+    return "Normalize Pim memory copies and verify addressable operands.";
   }
 
   void runOnOperation() final {
     ModuleOp moduleOp = getOperation();
-    auto funcOp = requirePimEntryFunc(moduleOp, "PIM memory normalization");
+    auto funcOp = requirePimEntryFunc(moduleOp, "Pim memory normalization");
     if (failed(funcOp)) {
       signalPassFailure();
       return;
@@ -761,20 +761,20 @@ static LogicalResult verifyNoTensorValues(ModuleOp moduleOp) {
     if (failureCount >= 8)
       return;
     if (op->getDialect()->getNamespace() == "tensor") {
-      op->emitOpError("tensor operation remains after PIM bufferization");
+      op->emitOpError("tensor operation remains after Pim bufferization");
       ++failureCount;
       return;
     }
     for (Value value : op->getOperands()) {
       if (isa<TensorType>(value.getType())) {
-        op->emitOpError("tensor operand remains after PIM bufferization");
+        op->emitOpError("tensor operand remains after Pim bufferization");
         ++failureCount;
         return;
       }
     }
     for (Value value : op->getResults()) {
       if (isa<TensorType>(value.getType())) {
-        op->emitOpError("tensor result remains after PIM bufferization");
+        op->emitOpError("tensor result remains after Pim bufferization");
         ++failureCount;
         return;
       }
@@ -782,7 +782,7 @@ static LogicalResult verifyNoTensorValues(ModuleOp moduleOp) {
   });
   if (failureCount != 0)
     moduleOp.emitError() << "found " << failureCount
-                         << " tensor value(s) after PIM bufferization"
+                         << " tensor value(s) after Pim bufferization"
                          << (failureCount == 8 ? " (first 8 reported)" : "");
   return success(failureCount == 0);
 }
@@ -793,7 +793,7 @@ struct PimBufferizationVerificationPass
 
   StringRef getArgument() const override { return "pim-bufferization-verification"; }
   StringRef getDescription() const override {
-    return "Verify tensor elimination, contiguity, and PIM copy address spaces.";
+    return "Verify tensor elimination, contiguity, and Pim copy address spaces.";
   }
 
   void runOnOperation() final {

src/PIM/Dialect/Pim/Passes/Transforms/HostConstantFolding/HostConstantFoldingPass.cpp

@@ -16,7 +16,7 @@ struct HostConstantFoldingPass : PassWrapper<HostConstantFoldingPass, OperationP
   MLIR_DEFINE_EXPLICIT_INTERNAL_INLINE_TYPE_ID(HostConstantFoldingPass)
 
   StringRef getArgument() const override { return "pim-host-constant-folding-pass"; }
-  StringRef getDescription() const override { return "Fold host-side constant expressions before PIM verification"; }
+  StringRef getDescription() const override { return "Fold host-side constant expressions before Pim verification"; }
 
   LogicalResult initialize(MLIRContext* context) override {
     RewritePatternSet owningPatterns(context);
@@ -38,7 +38,7 @@ struct HostConstantFoldingPass : PassWrapper<HostConstantFoldingPass, OperationP
     GreedyRewriteConfig config;
     config.enableFolding();
     if (failed(applyPatternsGreedily(moduleOp, *patterns, config))) {
-      moduleOp.emitError("PIM host constant folding failed in the greedy rewrite driver");
+      moduleOp.emitError("Pim host constant folding failed in the greedy rewrite driver");
       signalPassFailure();
       return;
     }

src/PIM/Dialect/Pim/Passes/Transforms/HostConstantFolding/Patterns/Constant.cpp

@@ -472,7 +472,7 @@ struct FoldConstantHostCopyPattern final : OpRewritePattern<memref::CopyOp> {
   }
 };
 
-// Converts PIM copies from dense globals into direct folded globals before codegen.
+// Converts Pim copies from dense globals into direct folded globals before codegen.
 struct FoldConstantMemCpPattern final : OpRewritePattern<pim::PimMemCopyOp> {
   using OpRewritePattern::OpRewritePattern;
 

src/PIM/Dialect/Pim/Passes/Transforms/InstructionSelection/InstructionSelectionPass.cpp

@@ -40,7 +40,7 @@ struct LowerTransposePattern final : OpRewritePattern<pim::PimTransposeOp> {
     auto sourceType = dyn_cast<MemRefType>(op.getInput().getType());
     auto targetType = dyn_cast<MemRefType>(op.getOutputBuffer().getType());
     if (!sourceType || !targetType || !sourceType.hasStaticShape() || !targetType.hasStaticShape())
-      return op.emitOpError("requires static memref operands before PIM instruction selection");
+      return op.emitOpError("requires static memref operands before Pim instruction selection");
 
     ArrayRef<int64_t> sourceShape = sourceType.getShape();
     size_t rank = sourceShape.size();
@@ -147,7 +147,7 @@ struct InstructionSelectionPass : PassWrapper<InstructionSelectionPass, Operatio
   MLIR_DEFINE_EXPLICIT_INTERNAL_INLINE_TYPE_ID(InstructionSelectionPass)
 
   StringRef getArgument() const override { return "pim-instruction-selection"; }
-  StringRef getDescription() const override { return "Select explicit PIM ISA operations"; }
+  StringRef getDescription() const override { return "Select explicit Pim ISA operations"; }
 
   void runOnOperation() override {
     RewritePatternSet patterns(&getContext());

src/PIM/Dialect/Pim/Passes/Transforms/LocalMemoryPlanning/LocalMemoryPlanning.cpp

@@ -36,7 +36,7 @@ struct PimLocalMemoryPlanningPass : PassWrapper<PimLocalMemoryPlanningPass, Oper
 
   StringRef getArgument() const override { return "pim-local-memory-planning"; }
   StringRef getDescription() const override {
-    return "Plan liveness-based addresses for PIM core-local memory";
+    return "Plan liveness-based addresses for Pim core-local memory";
   }
 
   void runOnOperation() override {
@@ -149,14 +149,14 @@ FailureOr<CoreMemoryPlan> buildCoreMemoryPlan(Operation* coreLikeOp) {
   plan.intervals = std::move(*intervals);
   auto placements = planLocalMemoryPlacements(plan.intervals, kPimLocalMemoryAddressLimit);
   if (failed(placements)) {
-    coreLikeOp->emitError("PIM local-memory plan exceeds the signed int32 address range");
+    coreLikeOp->emitError("Pim local-memory plan exceeds the signed int32 address range");
     return failure();
   }
   plan.placements = std::move(*placements);
   for (const LocalMemoryPlacement& placement : plan.placements) {
     auto end = alignedEnd(placement.address, placement.size, kPimLocalMemoryAddressLimit);
     if (failed(end)) {
-      coreLikeOp->emitError("PIM local-memory plan has invalid address arithmetic");
+      coreLikeOp->emitError("Pim local-memory plan has invalid address arithmetic");
       return failure();
     }
     plan.arenaSize = std::max(plan.arenaSize, *end);

src/PIM/Dialect/Pim/Passes/Transforms/Verification/VerificationPass.cpp

@@ -5,9 +5,11 @@
 #include "mlir/Pass/Pass.h"
 
 #include "llvm/ADT/STLExtras.h"
+#include "llvm/ADT/DenseSet.h"
 #include "llvm/Support/FormatVariadic.h"
 #include "llvm/Support/raw_ostream.h"
 
+#include <functional>
 #include <map>
 #include <string>
 
@@ -117,7 +119,7 @@ verifyLocalMemoryPlan(Operation* coreLikeOp, pim::CappedDiagnosticReporter& diag
   for (StringRef name : kRemovedLocalMemoryPlanAttrNames)
     if (coreLikeOp->hasAttr(name)) {
       diagnostics.report(coreLikeOp, [name](Operation* op) {
-        op->emitError() << "contains removed PIM local-memory planning attribute '" << name << "'";
+        op->emitError() << "contains removed Pim local-memory planning attribute '" << name << "'";
       });
       hasFailure = true;
     }
@@ -137,7 +139,7 @@ verifyLocalMemoryPlan(Operation* coreLikeOp, pim::CappedDiagnosticReporter& diag
   auto analyzed = pim::analyzeLocalMemoryLifetimes(coreLikeOp);
   if (failed(analyzed)) {
     diagnostics.report(coreLikeOp, [](Operation* op) {
-      op->emitError("cannot analyze PIM local-memory lifetimes for plan verification");
+      op->emitError("cannot analyze Pim local-memory lifetimes for plan verification");
     });
     return failure();
   }
@@ -156,7 +158,7 @@ verifyLocalMemoryPlan(Operation* coreLikeOp, pim::CappedDiagnosticReporter& diag
     for (StringRef name : kRemovedLocalMemoryPlanAttrNames)
       if (allocation->hasAttr(name)) {
         diagnostics.report(allocation, [name](Operation* op) {
-          op->emitOpError() << "contains removed PIM local-memory planning attribute '" << name << "'";
+          op->emitOpError() << "contains removed Pim local-memory planning attribute '" << name << "'";
         });
         hasFailure = true;
       }
@@ -171,7 +173,7 @@ verifyLocalMemoryPlan(Operation* coreLikeOp, pim::CappedDiagnosticReporter& diag
     uint64_t address = static_cast<uint64_t>(addressAttr.getInt());
     if (address % 4 != 0 || address > arenaSize || interval.size > arenaSize - address) {
       diagnostics.report(allocation, [&](Operation* op) {
-        op->emitOpError() << "has invalid PIM local-memory range [" << address << ", "
+          op->emitOpError() << "has invalid Pim local-memory range [" << address << ", "
                           << (address <= arenaSize && interval.size <= arenaSize - address
                                 ? address + interval.size
                                 : arenaSize)
@@ -221,7 +223,7 @@ verifyLocalMemoryPlan(Operation* coreLikeOp, pim::CappedDiagnosticReporter& diag
       memref::AllocOp otherAllocation = other.allocation;
       diagnostics.report(allocation, [&](Operation*) {
         auto diagnostic = allocation.emitOpError()
-          << "PIM local-memory plan assigns simultaneously live allocations to overlapping ranges; first range ["
+          << "Pim local-memory plan assigns simultaneously live allocations to overlapping ranges; first range ["
           << conflicting->first << ", " << conflicting->first + other.size << "), second range [" << address
           << ", " << address + interval.size << "), live positions overlap at ["
           << std::max(interval.start, other.start) << ", " << std::min(interval.end, other.end) << "]";
@@ -278,7 +280,9 @@ static bool isHostAddressableValue(Value value, const StaticValueKnowledge& know
 
 enum class CommunicationEventKind {
   Send,
-  Receive
+  Receive,
+  Sync,
+  Wait
 };
 
 struct CommunicationEvent {
@@ -286,14 +290,25 @@ struct CommunicationEvent {
   int64_t coreId = 0;
   int64_t peerCoreId = 0;
   int64_t size = 0;
+  int64_t eventRegister = 0;
+  int64_t waitValue = 0;
   uint64_t ordinal = 0;
   Operation* op = nullptr;
 };
 
 using CommunicationEventVector = SmallVector<CommunicationEvent, 0>;
+using SynchronizationEventKey = std::pair<int64_t, int64_t>;
+using SynchronizationSourceCounts =
+  DenseMap<SynchronizationEventKey, DenseMap<int64_t, int64_t>>;
 
 static StringRef getCommunicationEventKindName(CommunicationEventKind kind) {
-  return kind == CommunicationEventKind::Send ? "send" : "receive";
+  switch (kind) {
+  case CommunicationEventKind::Send: return "send";
+  case CommunicationEventKind::Receive: return "receive";
+  case CommunicationEventKind::Sync: return "sync";
+  case CommunicationEventKind::Wait: return "wait";
+  }
+  llvm_unreachable("unknown communication event kind");
 }
 
 constexpr StringLiteral kRaptorMinChannelIdAttr = "raptor.min_channel_id";
@@ -356,9 +371,15 @@ static std::string formatCommunicationEvent(const CommunicationEvent& event) {
 
   std::string text;
   llvm::raw_string_ostream os(text);
-  os << "core " << event.coreId << " " << getCommunicationEventKindName(event.kind) << " "
-     << (event.kind == CommunicationEventKind::Send ? "to" : "from") << " " << event.peerCoreId << " size "
-     << event.size << "B ordinal " << event.ordinal;
+  os << "core " << event.coreId << " " << getCommunicationEventKindName(event.kind);
+  if (event.kind == CommunicationEventKind::Send || event.kind == CommunicationEventKind::Receive)
+    os << " " << (event.kind == CommunicationEventKind::Send ? "to" : "from") << " " << event.peerCoreId
+       << " size " << event.size << "B";
+  else if (event.kind == CommunicationEventKind::Sync)
+    os << " event " << event.eventRegister << " to " << event.peerCoreId;
+  else
+    os << " event " << event.eventRegister << " value " << event.waitValue;
+  os << " ordinal " << event.ordinal;
   if (minChannelId)
     os << " min_channel " << *minChannelId;
   if (commOrder)
@@ -383,6 +404,9 @@ static std::string formatCommunicationEvent(const CommunicationEvent& event) {
 }
 
 static bool areMatchedCommunicationEvents(const CommunicationEvent& lhs, const CommunicationEvent& rhs) {
+  if ((lhs.kind != CommunicationEventKind::Send && lhs.kind != CommunicationEventKind::Receive)
+      || (rhs.kind != CommunicationEventKind::Send && rhs.kind != CommunicationEventKind::Receive))
+    return false;
   if (lhs.coreId != rhs.peerCoreId || lhs.peerCoreId != rhs.coreId || lhs.size != rhs.size)
     return false;
 
@@ -402,6 +426,27 @@ static void printCounterpartProbe(llvm::raw_ostream& os,
                                   const DenseMap<int64_t, CommunicationEventVector>& coreEvents,
                                   const DenseMap<int64_t, size_t>& programCounters,
                                   const CommunicationEvent& blockedEvent) {
+  if (blockedEvent.kind == CommunicationEventKind::Wait) {
+    os << "  SYNC probes for " << formatCommunicationEvent(blockedEvent) << "\n";
+    bool found = false;
+    for (const auto& [sourceCore, events] : coreEvents) {
+      size_t begin = programCounters.lookup(sourceCore);
+      for (size_t index = begin; index < events.size(); ++index) {
+        const CommunicationEvent& candidate = events[index];
+        if (candidate.kind != CommunicationEventKind::Sync
+            || candidate.peerCoreId != blockedEvent.coreId
+            || candidate.eventRegister != blockedEvent.eventRegister)
+          continue;
+        os << "    core " << sourceCore << " next matching SYNC at ordinal "
+           << index << " (distance +" << index - begin << ")\n";
+        found = true;
+        break;
+      }
+    }
+    if (!found)
+      os << "    no remaining matching SYNC exists\n";
+    return;
+  }
   auto peerEventsIt = coreEvents.find(blockedEvent.peerCoreId);
   if (peerEventsIt == coreEvents.end()) {
     os << "  no local stream was collected for peer core " << blockedEvent.peerCoreId << "\n";
@@ -456,7 +501,7 @@ static void printCounterpartProbe(llvm::raw_ostream& os,
 
 static CommunicationEvent makeCommunicationEvent(
   CommunicationEventKind kind, int64_t coreId, int64_t peerCoreId, int64_t size, uint64_t ordinal, Operation* op) {
-  return CommunicationEvent {kind, coreId, peerCoreId, size, ordinal, op};
+  return CommunicationEvent {kind, coreId, peerCoreId, size, 0, 0, ordinal, op};
 }
 
 static LogicalResult appendCoreCommunicationEvents(Block& block,
@@ -471,7 +516,7 @@ static LogicalResult appendCoreCommunicationEvents(Block& block,
         auto targetCoreId = resolveIndexValue(sendOp.getTargetCoreId(), knowledge);
         if (failed(targetCoreId)) {
           diagnostics.report(&op, [](Operation* illegalOp) {
-            illegalOp->emitOpError("cannot statically resolve send target core for PIM communication deadlock check");
+            illegalOp->emitOpError("cannot statically resolve send target core for Pim communication deadlock check");
           });
           return failure();
         }
@@ -490,7 +535,7 @@ static LogicalResult appendCoreCommunicationEvents(Block& block,
         if (failed(sourceCoreId)) {
           diagnostics.report(&op, [](Operation* illegalOp) {
             illegalOp->emitOpError(
-              "cannot statically resolve receive source core for PIM communication deadlock check");
+              "cannot statically resolve receive source core for Pim communication deadlock check");
           });
           return failure();
         }
@@ -504,6 +549,39 @@ static LogicalResult appendCoreCommunicationEvents(Block& block,
         return success();
       }
 
+      if (auto syncOp = dyn_cast<pim::PimSyncOp>(&op)) {
+        auto targetCoreId = resolveIndexValue(syncOp.getTargetCoreId(), knowledge);
+        auto eventRegister = resolveIndexValue(syncOp.getEventRegister(), knowledge);
+        if (failed(targetCoreId) || failed(eventRegister)) {
+          diagnostics.report(&op, [](Operation* illegalOp) {
+            illegalOp->emitOpError(
+              "cannot statically resolve SYNC operands for Pim communication deadlock check");
+          });
+          return failure();
+        }
+        events.push_back(CommunicationEvent {
+          CommunicationEventKind::Sync, coreId, *targetCoreId, 0,
+          *eventRegister, 0, static_cast<uint64_t>(events.size()), &op});
+        return success();
+      }
+
+      if (auto waitOp = dyn_cast<pim::PimWaitOp>(&op)) {
+        auto eventRegister = resolveIndexValue(waitOp.getEventRegister(), knowledge);
+        auto waitValue = resolveIndexValue(waitOp.getWaitValue(), knowledge);
+        if (failed(eventRegister) || failed(waitValue)) {
+          diagnostics.report(&op, [](Operation* illegalOp) {
+            illegalOp->emitOpError(
+              "cannot statically resolve WAIT operands for Pim communication deadlock check");
+          });
+          return failure();
+        }
+        events.push_back(CommunicationEvent {
+          CommunicationEventKind::Wait, coreId, coreId, 0,
+          *eventRegister, *waitValue,
+          static_cast<uint64_t>(events.size()), &op});
+        return success();
+      }
+
       return success();
     });
 }
@@ -530,7 +608,7 @@ static void printCommunicationWindow(llvm::raw_ostream& os,
 static void printCommunicationDeadlockReport(const DenseMap<int64_t, CommunicationEventVector>& coreEvents,
                                              const DenseMap<int64_t, size_t>& programCounters,
                                              ArrayRef<int64_t> cycle) {
-  llvm::errs() << "\n=== PIM static communication deadlock report ===\n";
+  llvm::errs() << "\n=== Pim static communication/synchronization deadlock report ===\n";
   llvm::errs() << "wait cycle:";
   for (int64_t coreId : cycle)
     llvm::errs() << " " << coreId;
@@ -565,7 +643,7 @@ static void printCommunicationDeadlockReport(const DenseMap<int64_t, Communicati
       continue;
     printCommunicationWindow(llvm::errs(), coreEvents, coreId, pcIt->second);
   }
-  llvm::errs() << "=== end PIM static communication deadlock report ===\n\n";
+  llvm::errs() << "=== end Pim static communication/synchronization deadlock report ===\n\n";
 }
 
 static void emitCommunicationDeadlockCycle(ModuleOp moduleOp,
@@ -576,8 +654,8 @@ static void emitCommunicationDeadlockCycle(ModuleOp moduleOp,
 
   auto diagnostic =
     moduleOp.emitError()
-    << "PIM communication deadlock check found a blocking send/receive cycle while statically simulating the "
-       "expanded per-core communication streams; see the PIM static communication deadlock report above";
+    << "Pim communication deadlock check found a blocking SEND/RECV/WAIT cycle while statically simulating the "
+       "expanded per-core communication streams; see the static deadlock report above";
 
   for (int64_t coreId : cycle) {
     auto eventsIt = coreEvents.find(coreId);
@@ -596,46 +674,82 @@ static void emitCommunicationDeadlockCycle(ModuleOp moduleOp,
 
 static FailureOr<SmallVector<int64_t>>
 findCommunicationWaitCycle(const DenseMap<int64_t, CommunicationEventVector>& coreEvents,
-                           const DenseMap<int64_t, size_t>& programCounters) {
-  for (const auto& [startCoreId, events] : coreEvents) {
-    auto startPcIt = programCounters.find(startCoreId);
-    if (startPcIt == programCounters.end() || startPcIt->second >= events.size())
+                           const DenseMap<int64_t, size_t>& programCounters,
+                           const DenseSet<int64_t>& repeatingCores,
+                           const SynchronizationSourceCounts& sourceCounts) {
+  DenseMap<int64_t, SmallVector<int64_t>> dependencies;
+  for (const auto& [coreId, events] : coreEvents) {
+    size_t pc = programCounters.lookup(coreId);
+    if (pc >= events.size())
       continue;
+    const CommunicationEvent& event = events[pc];
+    if (event.kind == CommunicationEventKind::Send
+        || event.kind == CommunicationEventKind::Receive) {
+      dependencies[coreId].push_back(event.peerCoreId);
+      continue;
+    }
+    if (event.kind != CommunicationEventKind::Wait)
+      continue;
+    int64_t blockedCoreId = coreId;
+    SynchronizationEventKey eventKey {coreId, event.eventRegister};
+    auto contributions = sourceCounts.find(eventKey);
+    for (const auto& [sourceCore, sourceEvents] : coreEvents) {
+      size_t sourcePc = programCounters.lookup(sourceCore);
+      auto matches = [&](const CommunicationEvent& candidate) {
+        return candidate.kind == CommunicationEventKind::Sync
+            && candidate.peerCoreId == blockedCoreId
+            && candidate.eventRegister == event.eventRegister;
+      };
+      int64_t signalsPerPhase = llvm::count_if(sourceEvents, matches);
+      if (repeatingCores.contains(sourceCore))
+        signalsPerPhase /= 2;
+      int64_t contributed = contributions == sourceCounts.end()
+        ? 0 : contributions->second.lookup(sourceCore);
+      if (contributed >= signalsPerPhase)
+        continue;
+      bool canSignal = llvm::any_of(
+        llvm::drop_begin(sourceEvents, sourcePc), matches);
+      if (canSignal)
+        dependencies[coreId].push_back(sourceCore);
+    }
+  }
 
+  DenseSet<int64_t> visited;
   DenseMap<int64_t, size_t> positionInPath;
   SmallVector<int64_t, 8> path;
-    int64_t currentCoreId = startCoreId;
-    while (true) {
-      auto eventsIt = coreEvents.find(currentCoreId);
-      auto pcIt = programCounters.find(currentCoreId);
-      if (eventsIt == coreEvents.end() || pcIt == programCounters.end() || pcIt->second >= eventsIt->second.size())
-        break;
-
-      auto positionIt = positionInPath.find(currentCoreId);
-      if (positionIt != positionInPath.end()) {
-        SmallVector<int64_t> cycle;
-        for (size_t index = positionIt->second; index < path.size(); ++index)
-          cycle.push_back(path[index]);
+  std::function<std::optional<SmallVector<int64_t>>(int64_t)> visit =
+    [&](int64_t coreId) -> std::optional<SmallVector<int64_t>> {
+      auto position = positionInPath.find(coreId);
+      if (position != positionInPath.end())
+        return SmallVector<int64_t>(
+          path.begin() + position->second, path.end());
+      if (!visited.insert(coreId).second)
+        return std::nullopt;
+      positionInPath[coreId] = path.size();
+      path.push_back(coreId);
+      for (int64_t target : dependencies.lookup(coreId))
+        if (auto cycle = visit(target))
           return cycle;
-      }
-
-      positionInPath[currentCoreId] = path.size();
-      path.push_back(currentCoreId);
-      currentCoreId = eventsIt->second[pcIt->second].peerCoreId;
-    }
-  }
-
+      path.pop_back();
+      positionInPath.erase(coreId);
+      return std::nullopt;
+    };
+  for (const auto& [coreId, unused] : dependencies)
+    if (auto cycle = visit(coreId))
+      return *cycle;
   return failure();
 }
 
 static LogicalResult verifyNoStaticCommunicationDeadlock(ModuleOp moduleOp,
                                                          pim::CappedDiagnosticReporter& diagnostics) {
   DenseMap<int64_t, CommunicationEventVector> coreEvents;
+  DenseSet<int64_t> repeatingCores;
   bool hasFailure = false;
 
   for (func::FuncOp funcOp : moduleOp.getOps<func::FuncOp>()) {
     if (funcOp.isExternal())
       continue;
+    bool repeating = funcOp->hasAttr("pim.pipeline_host_buffer_bytes");
 
     for (Operation& op : funcOp.getBody().front().getOperations()) {
       if (auto coreOp = dyn_cast<pim::PimCoreOp>(&op)) {
@@ -648,6 +762,8 @@ static LogicalResult verifyNoStaticCommunicationDeadlock(ModuleOp moduleOp,
                                                  coreEvents[coreId],
                                                  diagnostics)))
           hasFailure = true;
+        if (repeating)
+          repeatingCores.insert(coreId);
         continue;
       }
 
@@ -668,8 +784,11 @@ static LogicalResult verifyNoStaticCommunicationDeadlock(ModuleOp moduleOp,
                                                      coreId,
                                                      laneKnowledge,
                                                      coreEvents[coreId],
-                                                     diagnostics)))
+                                                     diagnostics))) {
               hasFailure = true;
+            } else if (repeating) {
+              repeatingCores.insert(coreId);
+            }
         }
       }
     }
@@ -678,10 +797,20 @@ static LogicalResult verifyNoStaticCommunicationDeadlock(ModuleOp moduleOp,
   if (hasFailure)
     return failure();
 
+  for (int64_t coreId : repeatingCores) {
+    CommunicationEventVector iteration = coreEvents[coreId];
+    for (CommunicationEvent event : iteration) {
+      event.ordinal = coreEvents[coreId].size();
+      coreEvents[coreId].push_back(event);
+    }
+  }
+
   DenseMap<int64_t, size_t> programCounters;
   for (const auto& [coreId, events] : coreEvents)
     programCounters[coreId] = 0;
 
+  DenseMap<SynchronizationEventKey, int64_t> eventCounts;
+  SynchronizationSourceCounts sourceCounts;
   while (true) {
     bool madeProgress = false;
     for (const auto& [coreId, events] : coreEvents) {
@@ -690,6 +819,35 @@ static LogicalResult verifyNoStaticCommunicationDeadlock(ModuleOp moduleOp,
         continue;
 
       const CommunicationEvent& event = events[pc];
+      if (event.kind == CommunicationEventKind::Sync) {
+        SynchronizationEventKey eventKey {
+          event.peerCoreId, event.eventRegister};
+        ++eventCounts[eventKey];
+        ++sourceCounts[eventKey][coreId];
+        ++programCounters[coreId];
+        madeProgress = true;
+        continue;
+      }
+      if (event.kind == CommunicationEventKind::Wait) {
+        int64_t observed = eventCounts.lookup(
+          {coreId, event.eventRegister});
+        if (observed == event.waitValue) {
+          SynchronizationEventKey eventKey {coreId, event.eventRegister};
+          eventCounts[eventKey] = 0;
+          sourceCounts.erase(eventKey);
+          ++programCounters[coreId];
+          madeProgress = true;
+        } else if (observed > event.waitValue) {
+          auto diagnostic = event.op->emitOpError()
+            << "Pim synchronization deadlock check found exact-count WAIT overshoot on core "
+            << coreId << " event " << event.eventRegister << ": expected "
+            << event.waitValue << ", observed " << observed;
+          diagnostic.attachNote()
+            << "WAIT requires exact equality and this event cannot decrease without a successful WAIT";
+          return failure();
+        }
+        continue;
+      }
       auto peerEventsIt = coreEvents.find(event.peerCoreId);
       if (peerEventsIt == coreEvents.end())
         continue;
@@ -720,7 +878,8 @@ static LogicalResult verifyNoStaticCommunicationDeadlock(ModuleOp moduleOp,
     if (allDone)
       return success();
 
-    auto cycle = findCommunicationWaitCycle(coreEvents, programCounters);
+    auto cycle = findCommunicationWaitCycle(
+      coreEvents, programCounters, repeatingCores, sourceCounts);
     if (succeeded(cycle)) {
       emitCommunicationDeadlockCycle(moduleOp, coreEvents, programCounters, *cycle);
       return failure();
@@ -728,8 +887,8 @@ static LogicalResult verifyNoStaticCommunicationDeadlock(ModuleOp moduleOp,
 
     auto diagnostic =
       moduleOp.emitError()
-      << "PIM communication deadlock check stalled without finding a closed wait cycle; this usually means a "
-         "send/receive peer is missing or ordered after a finished core";
+      << "Pim communication deadlock check stalled without finding a closed wait cycle; this usually means a "
+         "SEND/RECV peer or exact-count SYNC signal is missing or ordered after a finished core";
     for (const auto& [coreId, events] : coreEvents) {
       size_t pc = programCounters[coreId];
       if (pc >= events.size())
@@ -746,7 +905,7 @@ struct VerificationPass : PassWrapper<VerificationPass, OperationPass<ModuleOp>>
 
   StringRef getArgument() const override { return "verify-pim-pass"; }
   StringRef getDescription() const override {
-    return "Verify that bufferized PIM IR contains only explicit host/device transfers";
+    return "Verify that bufferized Pim IR contains only explicit host/device transfers";
   }
 
   VerificationPass() {}
@@ -763,7 +922,7 @@ struct VerificationPass : PassWrapper<VerificationPass, OperationPass<ModuleOp>>
     pim::CappedDiagnosticReporter diagnostics;
 
     if (!hasTarget || failed(targetResources.verify())) {
-      moduleOp.emitError("PIM codegen verification requires valid injected target resources");
+      moduleOp.emitError("Pim codegen verification requires valid injected target resources");
       signalPassFailure();
       return;
     }
@@ -792,7 +951,7 @@ struct VerificationPass : PassWrapper<VerificationPass, OperationPass<ModuleOp>>
         return;
 
       diagnostics.report(op, [](Operation* illegalOp) {
-        illegalOp->emitError("illegal Spatial operation reached PIM codegen verification");
+        illegalOp->emitError("illegal Spatial operation reached Pim codegen verification");
       });
     });
 
@@ -833,7 +992,7 @@ struct VerificationPass : PassWrapper<VerificationPass, OperationPass<ModuleOp>>
 
         if (!isAddressOnlyHostOp(&op)) {
           diagnostics.report(&op, [](Operation* illegalOp) {
-            illegalOp->emitOpError("illegal host-side runtime op remains after PIM bufferization; "
+            illegalOp->emitOpError("illegal host-side runtime op remains after Pim bufferization; "
                                    "fold it to constants or lower it into pim.core");
           });
           continue;
@@ -849,7 +1008,7 @@ struct VerificationPass : PassWrapper<VerificationPass, OperationPass<ModuleOp>>
 
     if (diagnostics.hasFailure()) {
       diagnostics.emitSuppressedSummary(moduleOp, "verification failures");
-      moduleOp.emitError("PIM codegen verification failed; see diagnostics above");
+      moduleOp.emitError("Pim codegen verification failed; see diagnostics above");
       hasFailure = true;
     }
 
@@ -928,7 +1087,7 @@ private:
         bool hasFailure = false;
         if (!isSupportedCoreInstructionOp(&op)) {
           diagnostics.report(&op, [](Operation* illegalOp) {
-            illegalOp->emitOpError("unsupported executable op reached PIM codegen verification");
+            illegalOp->emitOpError("unsupported executable op reached Pim codegen verification");
           });
           hasFailure = true;
         }
@@ -990,7 +1149,7 @@ private:
           if (failed(resolveIndexValue(storeOp.getHostTargetOffset(), knowledge))
               || failed(resolveIndexValue(storeOp.getDeviceSourceOffset(), knowledge))) {
             diagnostics.report(&op, [](Operation* illegalOp) {
-              illegalOp->emitOpError("offset operands must be statically evaluable for PIM codegen");
+            illegalOp->emitOpError("offset operands must be statically evaluable for Pim codegen");
             });
             hasFailure = true;
           }
@@ -1006,7 +1165,7 @@ private:
           if (failed(resolveIndexValue(loadOp.getDeviceTargetOffset(), knowledge))
               || failed(resolveIndexValue(loadOp.getHostSourceOffset(), knowledge))) {
             diagnostics.report(&op, [](Operation* illegalOp) {
-              illegalOp->emitOpError("offset operands must be statically evaluable for PIM codegen");
+            illegalOp->emitOpError("offset operands must be statically evaluable for Pim codegen");
             });
             hasFailure = true;
           }
@@ -1022,7 +1181,7 @@ private:
           if (failed(resolveIndexValue(copyOp.getTargetOffset(), knowledge))
               || failed(resolveIndexValue(copyOp.getSourceOffset(), knowledge))) {
             diagnostics.report(&op, [](Operation* illegalOp) {
-              illegalOp->emitOpError("offset operands must be statically evaluable for PIM codegen");
+            illegalOp->emitOpError("offset operands must be statically evaluable for Pim codegen");
             });
             hasFailure = true;
           }
@@ -1032,7 +1191,7 @@ private:
             && failed(resolveIndexValue(receiveOp.getOutputOffset(), knowledge))) {
           diagnostics.report(&op, [](Operation* illegalOp) {
             illegalOp->emitOpError(
-              "output offset must be statically evaluable for PIM codegen");
+              "output offset must be statically evaluable for Pim codegen");
           });
           hasFailure = true;
         }

src/PIM/Dialect/Pim/Pim.td

@@ -11,7 +11,7 @@ include "mlir/Dialect/Bufferization/IR/BufferViewFlowOpInterface.td"
 
 def PimDialect : Dialect {
   let name = "pim";
-  let summary = "A low-level dialect for the PIM coprocessors on ReRAM crossbars";
+  let summary = "A low-level dialect for the Pim coprocessors on ReRAM crossbars";
   let cppNamespace = "::onnx_mlir::pim";
 }
 
@@ -27,7 +27,7 @@ def PimTensor :
 
 def PimCoreOp : PimOp<"core", [SingleBlock,
     DeclareOpInterfaceMethods<OpAsmOpInterface, ["getAsmBlockArgumentNames"]>]> {
-  let summary = "Execute a block on a PIM core";
+  let summary = "Execute a block on a Pim core";
 
   let regions = (region SizedRegion<1>:$body);
 

src/PIM/Dialect/Spatial/Passes/Transforms/MergeComputeNodes/DeferredBoundaryPlanning.cpp

@@ -2,6 +2,8 @@
 #include "DeferredCommunicationScheduling.hpp"
 #include "DeferredTransferPlanning.hpp"
 
+#include "llvm/ADT/DenseSet.h"
+
 namespace onnx_mlir::spatial {
 using namespace mlir;
 namespace {
@@ -256,7 +258,7 @@ static unsigned getBarrierRoundCount(size_t coreCount) {
 
 static LogicalResult assignPipelineSynchronization(
     DeferredTransferPlan &transfers,
-    ArrayRef<BoundaryProgram> boundaries,
+    MutableArrayRef<BoundaryProgram> boundaries,
     size_t synchronizationRegisterCount) {
   bool pipelined = false;
   for (ScheduledInfo &scheduled : transfers.scheduled) {
@@ -284,14 +286,13 @@ static LogicalResult assignPipelineSynchronization(
   });
 
   DenseMap<int64_t, SmallVector<HostTransferRef>> incomingByCore;
+  DenseMap<int64_t, llvm::SmallSetVector<int64_t, 4>> readersByWriter;
   DenseMap<ExternalTransferFamily *, SmallVector<int64_t>> eventRegisters;
   DenseMap<ExternalTransferFamily *, SmallVector<int64_t>> waitValues;
-  DenseMap<ExternalTransferFamily *, SmallVector<int64_t>> acknowledgementRegisters;
   auto initialize = [&](ExternalTransferFamily &family) {
     size_t count = family.targetCores.size();
     eventRegisters.try_emplace(&family, count, 0);
     waitValues.try_emplace(&family, count, 0);
-    acknowledgementRegisters.try_emplace(&family, count, 0);
   };
   for (const BoundaryProgram &boundary : boundaries)
     for (const BoundaryInstruction &instruction : boundary.instructions) {
@@ -307,32 +308,117 @@ static LogicalResult assignPipelineSynchronization(
           int64_t source = family.sourceCores.valueAt(index);
           int64_t target = family.targetCores.valueAt(index);
           incomingByCore[target].push_back({&family, index});
-          ++transfers.hostAcknowledgementCounts[source];
+          readersByWriter[source].insert(target);
         }
       }
     }
 
+  DenseMap<int64_t, SmallVector<SmallVector<int64_t>>>
+    overwriteBoundaryReadersByWriter;
+  DenseSet<std::pair<int64_t, int64_t>> plannedRelations;
+  for (const BoundaryProgram &boundary : boundaries) {
+    DenseMap<int64_t, llvm::SmallSetVector<int64_t, 4>> readersAtBoundary;
+    for (const BoundaryInstruction &instruction : boundary.instructions) {
+      auto *send = std::get_if<EmitSendRun>(&instruction);
+      if (!send || send->slices.empty()
+          || !send->slices.front().family->hostRouted)
+        continue;
+      for (const ScheduledTransferSlice &slice : send->slices)
+        for (size_t offset = 0; offset < slice.transferCount; ++offset) {
+          size_t index = slice.familyOffset + offset;
+          int64_t writer = slice.family->sourceCores.valueAt(index);
+          int64_t reader = slice.family->targetCores.valueAt(index);
+          if (plannedRelations.insert({writer, reader}).second)
+            readersAtBoundary[writer].insert(reader);
+        }
+    }
+    for (const auto &[writer, readers] : readersAtBoundary)
+      overwriteBoundaryReadersByWriter[writer].emplace_back(
+        readers.begin(), readers.end());
+  }
+
   unsigned barrierRounds = getBarrierRoundCount(
     transfers.stageZeroCores.size());
-  bool stageZeroNeedsAcknowledgements = llvm::any_of(
-    transfers.stageZeroCores, [&](int64_t core) {
-      return transfers.hostAcknowledgementCounts.contains(core);
-    });
-  for (auto &[target, incoming] : incomingByCore) {
-    bool needsAcknowledgementRegister =
-      transfers.hostAcknowledgementCounts.contains(target);
-    bool stageZero = llvm::is_contained(transfers.stageZeroCores, target);
-    size_t reserved = stageZero
-      ? barrierRounds + (stageZeroNeedsAcknowledgements ? 1 : 0)
-      : 1 + (needsAcknowledgementRegister ? 1 : 0);
-    if (reserved >= synchronizationRegisterCount) {
-      incoming.front().family->requirement->exchange->deferred.emitOpError(
-        "pipeline synchronization leaves no event register for incoming host transfers");
+  auto reservedRegisterCount = [&](int64_t core) -> size_t {
+    if (llvm::is_contained(transfers.stageZeroCores, core)) {
+      bool releasesDownstream = !transfers.downstreamCores.empty()
+        && core == transfers.stageZeroCores.front();
+      return barrierRounds + releasesDownstream;
+    }
+    auto downstream = llvm::find(transfers.downstreamCores, core);
+    if (downstream == transfers.downstreamCores.end())
+      return 0;
+    size_t rank = downstream - transfers.downstreamCores.begin();
+    bool releasesChildren = 2 * rank + 1 < transfers.downstreamCores.size();
+    return 1 + releasesChildren;
+  };
+  for (int64_t core : transfers.stageZeroCores)
+    if (reservedRegisterCount(core) > synchronizationRegisterCount)
+      return transfers.scheduled.front().op->emitOpError(
+        "pipeline stage-zero synchronization requires more event registers than the target provides");
+  for (int64_t core : transfers.downstreamCores)
+    if (reservedRegisterCount(core) > synchronizationRegisterCount)
+      return transfers.scheduled.front().op->emitOpError(
+        "pipeline downstream restart synchronization requires more event registers than the target provides");
+  SmallVector<int64_t> synchronizationCores;
+  for (const auto &[core, incoming] : incomingByCore)
+    synchronizationCores.push_back(core);
+  for (const auto &[core, readers] : readersByWriter)
+    if (!llvm::is_contained(synchronizationCores, core))
+      synchronizationCores.push_back(core);
+  llvm::sort(synchronizationCores);
+
+  DenseMap<int64_t, SmallVector<unsigned>> freeRegisters;
+  DenseMap<int64_t, SmallVector<unsigned>> freeWaitCounts;
+  DenseMap<int64_t, DenseMap<int64_t, unsigned>> freeGroups;
+  for (int64_t core : synchronizationCores) {
+    auto incomingIt = incomingByCore.find(core);
+    ArrayRef<HostTransferRef> incoming = incomingIt == incomingByCore.end()
+      ? ArrayRef<HostTransferRef>() : ArrayRef(incomingIt->second);
+    size_t readerCount = readersByWriter.lookup(core).size();
+    bool needsFreeRegister = readerCount != 0;
+    size_t reserved = reservedRegisterCount(core);
+    if (reserved + needsFreeRegister > synchronizationRegisterCount
+        || (!incoming.empty()
+            && reserved + needsFreeRegister == synchronizationRegisterCount)) {
+      transfers.scheduled.front().op->emitOpError(
+        "pipeline synchronization leaves no event register for host communication");
       return failure();
     }
-    size_t groupCount = std::min(
-      incoming.size(), synchronizationRegisterCount - reserved);
-    // One wait consumes a complete consecutive group of producer signals.
+    size_t available = synchronizationRegisterCount - reserved;
+    size_t readyRegisterCount = std::min(
+      incoming.size(), available - needsFreeRegister);
+    auto overwriteBoundaries = overwriteBoundaryReadersByWriter.find(core);
+    size_t overwriteBoundaryCount = overwriteBoundaries
+      == overwriteBoundaryReadersByWriter.end()
+      ? 0 : overwriteBoundaries->second.size();
+    size_t freeGroupCount = std::min(
+      overwriteBoundaryCount, available - readyRegisterCount);
+    if (needsFreeRegister) {
+      size_t overwriteReaderCount = 0;
+      if (overwriteBoundaries != overwriteBoundaryReadersByWriter.end())
+        for (ArrayRef<int64_t> readers : overwriteBoundaries->second)
+          overwriteReaderCount += readers.size();
+      if (overwriteReaderCount != readerCount)
+        return transfers.scheduled.front().op->emitOpError(
+          "host reuse readers do not match writer overwrite relations");
+      freeRegisters[core].reserve(freeGroupCount);
+      freeWaitCounts[core].assign(freeGroupCount, 0);
+      for (unsigned group = 0; group < freeGroupCount; ++group)
+        freeRegisters[core].push_back(readyRegisterCount + group);
+      for (auto [ordinal, readers] :
+           llvm::enumerate(overwriteBoundaries->second)) {
+        unsigned group = ordinal * freeGroupCount / overwriteBoundaryCount;
+        for (int64_t reader : readers) {
+          freeGroups[core][reader] = group;
+          ++freeWaitCounts[core][group];
+        }
+      }
+    }
+    if (incoming.empty())
+      continue;
+
+    size_t groupCount = readyRegisterCount;
     SmallVector<size_t> groupSizes(groupCount);
     for (size_t ordinal = 0; ordinal < incoming.size(); ++ordinal)
       ++groupSizes[ordinal * groupCount / incoming.size()];
@@ -341,8 +427,6 @@ static LogicalResult assignPipelineSynchronization(
       size_t group = ordinal * groupCount / incoming.size();
       HostTransferRef transfer = incoming[ordinal];
       eventRegisters[transfer.family][transfer.index] = group;
-      acknowledgementRegisters[transfer.family][transfer.index] =
-        synchronizationRegisterCount - 1;
       if (first[group]) {
         waitValues[transfer.family][transfer.index] = groupSizes[group];
         first[group] = false;
@@ -352,27 +436,70 @@ static LogicalResult assignPipelineSynchronization(
   for (auto &[family, values] : eventRegisters) {
     family->eventRegisters = StaticIntSequence::fromValues(values);
     family->waitValues = StaticIntSequence::fromValues(waitValues[family]);
-    family->acknowledgementEventRegisters =
-      StaticIntSequence::fromValues(acknowledgementRegisters[family]);
   }
 
-  if (!transfers.stageZeroCores.empty()) {
-    size_t reserved = barrierRounds
-                    + (stageZeroNeedsAcknowledgements ? 1 : 0);
-    if (reserved > synchronizationRegisterCount)
-      return transfers.scheduled.front().op->emitOpError(
-        "pipeline stage-zero barrier requires more synchronization registers than the target provides");
+  SmallVector<int64_t> writers;
+  for (const auto &[writer, readers] : readersByWriter)
+    writers.push_back(writer);
+  llvm::sort(writers);
+  for (int64_t writer : writers)
+    for (int64_t reader : readersByWriter[writer]) {
+      unsigned group = freeGroups[writer].lookup(reader);
+      transfers.hostReleaseSignals[reader].push_back(
+        {writer, freeRegisters[writer][group]});
     }
-  if (!transfers.downstreamCores.empty()) {
-    bool needsAcknowledgements = llvm::any_of(
-      transfers.downstreamCores, [&](int64_t core) {
-        return transfers.hostAcknowledgementCounts.contains(core);
-      });
-    if (1 + (needsAcknowledgements ? 1 : 0)
-        > synchronizationRegisterCount)
-      return transfers.scheduled.front().op->emitOpError(
-        "pipeline stage-zero release requires more synchronization registers than the target provides");
+
+  DenseSet<std::pair<int64_t, unsigned>> pendingGroups;
+  for (int64_t writer : writers)
+    for (unsigned group = 0; group < freeRegisters[writer].size(); ++group)
+      pendingGroups.insert({writer, group});
+  for (BoundaryProgram &boundary : boundaries) {
+    SmallVector<BoundaryInstruction, 0> instructions;
+    for (BoundaryInstruction &instruction : boundary.instructions) {
+      auto *send = std::get_if<EmitSendRun>(&instruction);
+      SmallVector<std::pair<unsigned, unsigned>> waits;
+      if (send && !send->slices.empty()
+          && send->slices.front().family->hostRouted) {
+        for (const ScheduledTransferSlice &slice : send->slices) {
+          ExternalTransferFamily &family = *slice.family;
+          unsigned lane = family.requirement->producer->scheduledLane;
+          if (lane >= boundary.key.first->cores.size())
+            return family.requirement->exchange->deferred.emitOpError(
+              "host reuse wait references an invalid writer lane"), failure();
+          for (size_t offset = 0; offset < slice.transferCount; ++offset) {
+            int64_t writer = family.sourceCores.valueAt(
+              slice.familyOffset + offset);
+            if (boundary.key.first->cores[lane] != writer)
+              return family.requirement->exchange->deferred.emitOpError(
+                "host reuse wait writer does not match its scheduled lane"),
+                failure();
+            int64_t reader = family.targetCores.valueAt(
+              slice.familyOffset + offset);
+            unsigned group = freeGroups[writer].lookup(reader);
+            if (!pendingGroups.erase({writer, group}))
+              continue;
+            waits.push_back({lane, group});
           }
+        }
+      }
+      for (auto [lane, group] : waits) {
+        int64_t writer = boundary.key.first->cores[lane];
+        SmallVector<int64_t> registers(boundary.key.first->cores.size());
+        SmallVector<int64_t> counts(boundary.key.first->cores.size());
+        registers[lane] = freeRegisters[writer][group];
+        counts[lane] = freeWaitCounts[writer][group];
+        instructions.push_back(EmitHostReuseWait {
+          LaneSet::range(lane, lane + 1),
+          StaticIntSequence::fromValues(registers),
+          StaticIntSequence::fromValues(counts)});
+      }
+      instructions.push_back(std::move(instruction));
+    }
+    boundary.instructions = std::move(instructions);
+  }
+  if (!pendingGroups.empty())
+    return transfers.scheduled.front().op->emitOpError(
+      "host reuse permission has no writer overwrite boundary");
   return success();
 }
 

src/PIM/Dialect/Spatial/Passes/Transforms/MergeComputeNodes/DeferredBoundaryPlanning.hpp

@@ -34,12 +34,17 @@ struct EmitReceiveAssemblyRun {
   llvm::SmallVector<LaneSet> entryLanes;
   LaneSet lanes;
 };
+struct EmitHostReuseWait {
+  LaneSet lanes;
+  StaticIntSequence eventRegisters = StaticIntSequence::uniform(0, 1);
+  StaticIntSequence waitValues = StaticIntSequence::uniform(0, 1);
+};
 struct ProduceDeferredResult {
   DeferredExchangePlan* exchange = nullptr;
 };
 
 using BoundaryInstruction =
-  std::variant<EmitSendRun, EmitLocalCollectionRun,
+  std::variant<EmitSendRun, EmitHostReuseWait, EmitLocalCollectionRun,
                EmitLocalCollectionLoopRun, EmitReceiveAssemblyRun,
                ProduceDeferredResult>;
 struct BoundaryProgram {

src/PIM/Dialect/Spatial/Passes/Transforms/MergeComputeNodes/DeferredBoundaryRealization.cpp

@@ -10,6 +10,7 @@
 #include "src/Accelerators/PIM/Common/IR/StaticIntSequence.hpp"
 #include "src/Accelerators/PIM/Common/IR/TensorSliceUtils.hpp"
 #include "src/Accelerators/PIM/Common/PimCommon.hpp"
+#include "llvm/ADT/DenseSet.h"
 #include <array>
 namespace onnx_mlir::spatial {
 using namespace mlir;
@@ -23,7 +24,6 @@ struct LogicalTransferMetadataView {
   StaticIntSequenceChain hostOffsets;
   StaticIntSequenceChain eventRegisters;
   StaticIntSequenceChain waitValues;
-  StaticIntSequenceChain acknowledgementEventRegisters;
   StaticIntSequenceChain targetLanes;
   StaticIntSequenceChain localOffsets;
   SmallVector<StaticIntSequenceChain> projectionOffsets;
@@ -78,6 +78,29 @@ static FailureOr<Value> emitLaneCondition(const LaneSet &lanes, Value lane, unsi
   Value selected = active->emitLookup(context.constants.getIndex(0), lane, anchor, context.constants, context.rewriter, loc);
   return arith::CmpIOp::create(context.rewriter, loc, arith::CmpIPredicate::ne, selected, context.constants.getIndex(0)).getResult();
 }
+template <typename Emit>
+static LogicalResult emitForLanes(
+    const LaneSet &active, Value lane, unsigned laneCount, Operation *anchor,
+    DeferredEmissionContext &context, Location loc, Emit emit) {
+  if (active.empty())
+    return success();
+  if (!lane) {
+    if (active.contains(0))
+      emit();
+    return success();
+  }
+  auto condition = emitLaneCondition(
+    active, lane, laneCount, anchor, context, loc);
+  if (failed(condition))
+    return failure();
+  auto conditional = scf::IfOp::create(
+    context.rewriter, loc, TypeRange {}, *condition, false);
+  OpBuilder::InsertionGuard guard(context.rewriter);
+  context.rewriter.setInsertionPoint(
+    conditional.getThenRegion().front().getTerminator());
+  emit();
+  return success();
+}
 static void appendMetadata(const ScheduledTransferSlice &slice, LogicalTransferMetadataView &metadata) {
   ExternalTransferFamily &family = *slice.family;
   LaneInterval familyLanes = family.targetLanes.intervals().front();
@@ -95,8 +118,6 @@ static void appendMetadata(const ScheduledTransferSlice &slice, LogicalTransferM
     metadata.eventRegisters.append(
       family.eventRegisters, familyIndex, count);
     metadata.waitValues.append(family.waitValues, familyIndex, count);
-    metadata.acknowledgementEventRegisters.append(
-      family.acknowledgementEventRegisters, familyIndex, count);
   }
   metadata.targetLanes.append(StaticIntSequence::affine(targetLane, 1, count));
   if (family.requirement->producerLocalOffsets)
@@ -303,20 +324,15 @@ static FailureOr<Value> emitReceiveValue(ArrayRef<ScheduledTransferSlice> slices
   std::optional<StaticIntGrid> hostOffsets;
   std::optional<StaticIntGrid> eventRegisters;
   std::optional<StaticIntGrid> waitValues;
-  std::optional<StaticIntGrid> acknowledgementEventRegisters;
   if (slices.front().family->hostRouted) {
     auto offsets = buildGrid(metadata.hostOffsets);
     auto events = buildGrid(metadata.eventRegisters);
     auto waits = buildGrid(metadata.waitValues);
-    auto acknowledgements = buildGrid(
-      metadata.acknowledgementEventRegisters);
-    if (failed(offsets) || failed(events) || failed(waits)
-        || failed(acknowledgements))
+    if (failed(offsets) || failed(events) || failed(waits))
       return failure();
     hostOffsets = std::move(*offsets);
     eventRegisters = std::move(*events);
     waitValues = std::move(*waits);
-    acknowledgementEventRegisters = std::move(*acknowledgements);
   }
   Value position = lane ? lane : context.constants.getIndex(0);
   Value row = context.constants.getIndex(0);
@@ -335,8 +351,6 @@ static FailureOr<Value> emitReceiveValue(ArrayRef<ScheduledTransferSlice> slices
       eventRegisters->emitLookup(
         row, position, anchor, context.constants, context.rewriter, anchor->getLoc()),
       waitValues->emitLookup(
-        row, position, anchor, context.constants, context.rewriter, anchor->getLoc()),
-      acknowledgementEventRegisters->emitLookup(
         row, position, anchor, context.constants, context.rewriter, anchor->getLoc()));
     receive = op;
     output = op.getOutput();
@@ -406,7 +420,6 @@ static FailureOr<Value> emitReceiveAssembly(const EmitReceiveAssemblyRun &run, V
   std::optional<StaticIntGrid> hostOffsets;
   std::optional<StaticIntGrid> eventRegisters;
   std::optional<StaticIntGrid> waitValues;
-  std::optional<StaticIntGrid> acknowledgementEventRegisters;
   bool hostRouted = run.slices.front().family->hostRouted;
   auto metadataByEntry = buildRectangularReceiveMetadata(run, laneCount);
   if (succeeded(metadataByEntry)) {
@@ -424,15 +437,11 @@ static FailureOr<Value> emitReceiveAssembly(const EmitReceiveAssemblyRun &run, V
         &LogicalTransferMetadataView::eventRegisters);
       auto waits = buildRows(
         &LogicalTransferMetadataView::waitValues);
-      auto acknowledgements = buildRows(
-        &LogicalTransferMetadataView::acknowledgementEventRegisters);
-      if (failed(offsets) || failed(events) || failed(waits)
-          || failed(acknowledgements))
+      if (failed(offsets) || failed(events) || failed(waits))
         return failure();
       hostOffsets = std::move(*offsets);
       eventRegisters = std::move(*events);
       waitValues = std::move(*waits);
-      acknowledgementEventRegisters = std::move(*acknowledgements);
     }
     SmallVector<StaticIntSequence> positionRows;
     for (unsigned position : run.positions)
@@ -485,15 +494,11 @@ static FailureOr<Value> emitReceiveAssembly(const EmitReceiveAssemblyRun &run, V
         &LogicalTransferMetadataView::eventRegisters);
       auto waits = buildGrid(
         &LogicalTransferMetadataView::waitValues);
-      auto acknowledgements = buildGrid(
-        &LogicalTransferMetadataView::acknowledgementEventRegisters);
-      if (failed(offsets) || failed(events) || failed(waits)
-          || failed(acknowledgements))
+      if (failed(offsets) || failed(events) || failed(waits))
         return failure();
       hostOffsets = std::move(*offsets);
       eventRegisters = std::move(*events);
       waitValues = std::move(*waits);
-      acknowledgementEventRegisters = std::move(*acknowledgements);
     }
     SmallVector<StaticIntSequence> positionColumns;
     for (const StaticIntSequenceChain &values : positionsByLane)
@@ -527,8 +532,6 @@ static FailureOr<Value> emitReceiveAssembly(const EmitReceiveAssemblyRun &run, V
         eventRegisters->emitLookup(
           entry, runtimeLane, anchor, context.constants, context.rewriter, loc),
         waitValues->emitLookup(
-          entry, runtimeLane, anchor, context.constants, context.rewriter, loc),
-        acknowledgementEventRegisters->emitLookup(
           entry, runtimeLane, anchor, context.constants, context.rewriter, loc));
       receive = op;
       output = op.getOutput();
@@ -1083,7 +1086,8 @@ static LogicalResult emitLocalCollectionUpdate(const EmitLocalCollectionRun &upd
 }
 
 static FailureOr<SmallVector<Value>> emitInstructions(ArrayRef<BoundaryInstruction> instructions, Value lane, unsigned laneCount,
-                                                      ArrayRef<DeferredResultPlan> results, DeferredEmissionContext &context) {
+                                                      ArrayRef<DeferredResultPlan> results, ScheduledInfo &scheduled,
+                                                      DeferredEmissionContext &context) {
   SmallVector<Value> produced;
   for (size_t instructionIndex = 0;
        instructionIndex < instructions.size(); ++instructionIndex) {
@@ -1108,7 +1112,22 @@ static FailureOr<SmallVector<Value>> emitInstructions(ArrayRef<BoundaryInstructi
           }
         }
     }
-    if (auto send = std::get_if<EmitSendRun>(&instruction)) {
+    if (auto reuse = std::get_if<EmitHostReuseWait>(&instruction)) {
+      Value runtimeLane = lane ? lane : context.constants.getIndex(0);
+      Location loc = scheduled.op->getLoc();
+      if (failed(emitForLanes(
+            reuse->lanes, lane, laneCount, scheduled.op, context, loc, [&]() {
+              SpatWaitOp::create(
+                context.rewriter, loc,
+                emitStaticIntLookup(
+                  reuse->eventRegisters, runtimeLane, scheduled.op,
+                  context.constants, context.rewriter, loc),
+                emitStaticIntLookup(
+                  reuse->waitValues, runtimeLane, scheduled.op,
+                  context.constants, context.rewriter, loc));
+            })))
+        return failure();
+    } else if (auto send = std::get_if<EmitSendRun>(&instruction)) {
       if (failed(emitConditionalSendRun(*send, lane, laneCount, context)))
         return failure();
     } else if (auto update = std::get_if<EmitLocalCollectionRun>(&instruction)) {
@@ -1192,7 +1211,9 @@ static LogicalResult emitBoundary(const BoundaryProgram &boundary, ArrayRef<Defe
   if (auto batch = dyn_cast<SpatScheduledComputeBatch>(boundary.key.first->op))
     lane = *batch.getLaneArgument();
   SmallVector<DeferredExchangePlan *> exchanges = getProducedExchanges(boundary.instructions);
-  auto values = emitInstructions(boundary.instructions, lane, laneCount, results, context);
+  auto values = emitInstructions(
+    boundary.instructions, lane, laneCount, results, *boundary.key.first,
+    context);
   return failed(values) ? failure() : replaceResults(exchanges, *values, replacements);
 }
 
@@ -1203,30 +1224,129 @@ static unsigned getBarrierRoundCount(size_t coreCount) {
   return rounds;
 }
 
+static LogicalResult emitHostReleaseSynchronization(
+    DeferredTransferPlan &transfers, DeferredEmissionContext &context) {
+  struct Emission {
+    HostReleaseSignal signal;
+    LaneSet lanes;
+  };
+  DenseSet<int64_t> emittedReaders;
+  for (ScheduledInfo &scheduled : transfers.scheduled) {
+    SmallVector<Emission> emissions;
+    for (auto [laneIndex, core] : llvm::enumerate(scheduled.cores)) {
+      auto found = transfers.hostReleaseSignals.find(core);
+      if (found == transfers.hostReleaseSignals.end())
+        continue;
+      if (!emittedReaders.insert(core).second)
+        return scheduled.op->emitOpError(
+          "host release reader is represented by multiple scheduled lanes");
+      for (HostReleaseSignal signal : found->second) {
+        auto emission = llvm::find_if(emissions, [&](const Emission &item) {
+          return item.signal.writerCore == signal.writerCore
+              && item.signal.eventRegister == signal.eventRegister;
+        });
+        if (emission == emissions.end()) {
+          emissions.push_back({signal, LaneSet::range(
+            laneIndex, laneIndex + 1)});
+        } else {
+          emission->lanes = emission->lanes.unite(
+            LaneSet::range(laneIndex, laneIndex + 1));
+        }
+      }
+    }
+    if (emissions.empty())
+      continue;
+    Block *block = scheduled.blocks.front();
+    context.rewriter.setInsertionPointToStart(block);
+    Location loc = scheduled.op->getLoc();
+    Value lane;
+    if (auto batch = dyn_cast<SpatScheduledComputeBatch>(scheduled.op))
+      lane = *batch.getLaneArgument();
+    for (const Emission &emission : emissions)
+      if (failed(emitForLanes(
+            emission.lanes, lane, scheduled.cores.size(), scheduled.op,
+            context, loc, [&]() {
+              SpatSyncOp::create(
+                context.rewriter, loc,
+                context.constants.getIndex(emission.signal.writerCore),
+                context.constants.getIndex(emission.signal.eventRegister));
+            })))
+        return failure();
+  }
+  return success();
+}
+
+static LogicalResult emitRestartReleasePermissions(
+    DeferredTransferPlan &transfers, DeferredEmissionContext &context) {
+  if (transfers.downstreamCores.empty())
+    return success();
+  unsigned barrierRounds = getBarrierRoundCount(
+    transfers.stageZeroCores.size());
+  size_t restartRegister = transfers.synchronizationRegisterCount - 1;
+  size_t leaderPermissionRegister = restartRegister - barrierRounds;
+  size_t downstreamPermissionRegister = restartRegister - 1;
+  DenseMap<int64_t, unsigned> downstreamRank;
+  for (auto [rank, core] : llvm::enumerate(transfers.downstreamCores))
+    downstreamRank[core] = rank;
+
+  for (ScheduledInfo &scheduled : transfers.scheduled) {
+    LaneSet lanes;
+    SmallVector<int64_t> targets(scheduled.cores.size());
+    SmallVector<int64_t> registers(scheduled.cores.size());
+    for (auto [lane, core] : llvm::enumerate(scheduled.cores)) {
+      auto rank = downstreamRank.find(core);
+      if (rank == downstreamRank.end())
+        continue;
+      lanes = lanes.unite(LaneSet::range(lane, lane + 1));
+      if (rank->second == 0) {
+        targets[lane] = transfers.stageZeroCores.front();
+        registers[lane] = leaderPermissionRegister;
+      } else {
+        targets[lane] = transfers.downstreamCores[(rank->second - 1) / 2];
+        registers[lane] = downstreamPermissionRegister;
+      }
+    }
+    if (lanes.empty())
+      continue;
+    Block *block = scheduled.blocks.front();
+    context.rewriter.setInsertionPointToStart(block);
+    Location loc = scheduled.op->getLoc();
+    Value lane;
+    if (auto batch = dyn_cast<SpatScheduledComputeBatch>(scheduled.op))
+      lane = *batch.getLaneArgument();
+    Value runtimeLane = lane ? lane : context.constants.getIndex(0);
+    if (failed(emitForLanes(
+          lanes, lane, scheduled.cores.size(), scheduled.op, context, loc,
+          [&]() {
+            Value target = emitStaticIntLookup(
+              StaticIntSequence::fromValues(targets), runtimeLane,
+              scheduled.op, context.constants, context.rewriter, loc);
+            Value eventRegister = emitStaticIntLookup(
+              StaticIntSequence::fromValues(registers), runtimeLane,
+              scheduled.op, context.constants, context.rewriter, loc);
+            SpatSyncOp::create(
+              context.rewriter, loc, target, eventRegister);
+          })))
+      return failure();
+  }
+  return success();
+}
+
 static LogicalResult emitCompletionSynchronization(
     DeferredTransferPlan &transfers, DeferredEmissionContext &context) {
   if (transfers.synchronizationRegisterCount == 0)
     return success();
-  size_t acknowledgementRegister =
+  size_t restartRegister =
     transfers.synchronizationRegisterCount - 1;
   unsigned barrierRounds = getBarrierRoundCount(
     transfers.stageZeroCores.size());
-  bool stageZeroNeedsAcknowledgements = llvm::any_of(
-    transfers.stageZeroCores, [&](int64_t core) {
-      return transfers.hostAcknowledgementCounts.contains(core);
-    });
-  size_t firstBarrierRegister = acknowledgementRegister
-                              - (stageZeroNeedsAcknowledgements ? 1 : 0);
+  size_t firstBarrierRegister = restartRegister;
   DenseMap<int64_t, unsigned> stageZeroRank;
   for (auto [rank, core] : llvm::enumerate(transfers.stageZeroCores))
     stageZeroRank[core] = rank;
   DenseMap<int64_t, unsigned> downstreamRank;
   for (auto [rank, core] : llvm::enumerate(transfers.downstreamCores))
     downstreamRank[core] = rank;
-  auto getReleaseRegister = [&](int64_t core) {
-    return acknowledgementRegister
-         - (transfers.hostAcknowledgementCounts.contains(core) ? 1 : 0);
-  };
 
   for (ScheduledInfo &scheduled : transfers.scheduled) {
     Block *block = scheduled.blocks.front();
@@ -1236,13 +1356,10 @@ static LogicalResult emitCompletionSynchronization(
     if (auto batch = dyn_cast<SpatScheduledComputeBatch>(scheduled.op))
       lane = *batch.getLaneArgument();
 
-    SmallVector<int64_t> acknowledgementCounts, releaseRegisters;
-    SmallVector<int64_t> releaseWaitValues, leftTargets, leftRegisters;
-    SmallVector<int64_t> rightTargets, rightRegisters;
-    LaneSet barrierLanes, leaderLanes, leftLanes, rightLanes;
+    SmallVector<int64_t> leftTargets, rightTargets, childCounts;
+    LaneSet barrierLanes, leaderLanes, downstreamLanes, parentLanes,
+      leftLanes, rightLanes;
     for (auto [index, core] : llvm::enumerate(scheduled.cores)) {
-      acknowledgementCounts.push_back(
-        transfers.hostAcknowledgementCounts.lookup(core));
       if (stageZeroRank.contains(core))
         barrierLanes = barrierLanes.unite(
           LaneSet::range(index, index + 1));
@@ -1252,66 +1369,37 @@ static LogicalResult emitCompletionSynchronization(
 
       auto rank = downstreamRank.find(core);
       if (rank == downstreamRank.end()) {
-        releaseRegisters.push_back(0);
-        releaseWaitValues.push_back(0);
         leftTargets.push_back(core);
-        leftRegisters.push_back(0);
         rightTargets.push_back(core);
-        rightRegisters.push_back(0);
+        childCounts.push_back(0);
         continue;
       }
-      releaseRegisters.push_back(getReleaseRegister(core));
-      releaseWaitValues.push_back(1);
+      downstreamLanes = downstreamLanes.unite(
+        LaneSet::range(index, index + 1));
       size_t left = 2 * rank->second + 1;
       size_t right = left + 1;
+      unsigned childCount = 0;
       if (left < transfers.downstreamCores.size()) {
         int64_t child = transfers.downstreamCores[left];
         leftTargets.push_back(child);
-        leftRegisters.push_back(getReleaseRegister(child));
         leftLanes = leftLanes.unite(LaneSet::range(index, index + 1));
+        ++childCount;
       } else {
         leftTargets.push_back(core);
-        leftRegisters.push_back(0);
       }
       if (right < transfers.downstreamCores.size()) {
         int64_t child = transfers.downstreamCores[right];
         rightTargets.push_back(child);
-        rightRegisters.push_back(getReleaseRegister(child));
         rightLanes = rightLanes.unite(LaneSet::range(index, index + 1));
+        ++childCount;
       } else {
         rightTargets.push_back(core);
-        rightRegisters.push_back(0);
       }
+      childCounts.push_back(childCount);
+      if (childCount != 0)
+        parentLanes = parentLanes.unite(LaneSet::range(index, index + 1));
     }
     Value runtimeLane = lane ? lane : context.constants.getIndex(0);
-    auto emitForLanes = [&](const LaneSet &active, auto emit) -> LogicalResult {
-      if (active.empty())
-        return success();
-      if (!lane) {
-        if (active.contains(0))
-          emit();
-        return success();
-      }
-      auto condition = emitLaneCondition(
-        active, lane, scheduled.cores.size(), scheduled.op, context, loc);
-      if (failed(condition))
-        return failure();
-      auto conditional = scf::IfOp::create(
-        context.rewriter, loc, TypeRange {}, *condition, false);
-      OpBuilder::InsertionGuard guard(context.rewriter);
-      context.rewriter.setInsertionPoint(
-        conditional.getThenRegion().front().getTerminator());
-      emit();
-      return success();
-    };
-    Value acknowledgementCount = emitStaticIntLookup(
-      StaticIntSequence::fromValues(acknowledgementCounts),
-      runtimeLane, scheduled.op,
-      context.constants, context.rewriter, loc);
-    SpatWaitOp::create(
-      context.rewriter, loc,
-      context.constants.getIndex(acknowledgementRegister),
-      acknowledgementCount);
 
     // Dissemination barrier: every round doubles the covered stage-zero peers.
     auto emitBarrier = [&]() {
@@ -1341,45 +1429,64 @@ static LogicalResult emitCompletionSynchronization(
       }
     };
     if (barrierRounds > 0
-        && failed(emitForLanes(barrierLanes, emitBarrier)))
+        && failed(emitForLanes(
+          barrierLanes, lane, scheduled.cores.size(), scheduled.op, context,
+          loc, emitBarrier)))
       return failure();
 
     // Gate downstream restarts so no core advances the simulator input
     // iteration ahead of stage zero.
     if (!transfers.downstreamCores.empty()
-        && failed(emitForLanes(leaderLanes, [&]() {
+        && failed(emitForLanes(
+          leaderLanes, lane, scheduled.cores.size(), scheduled.op, context,
+          loc, [&]() {
+          SpatWaitOp::create(
+            context.rewriter, loc,
+            context.constants.getIndex(firstBarrierRegister - barrierRounds),
+            context.constants.getIndex(1));
           int64_t root = transfers.downstreamCores.front();
           SpatSyncOp::create(
             context.rewriter, loc, context.constants.getIndex(root),
-            context.constants.getIndex(getReleaseRegister(root)));
+            context.constants.getIndex(restartRegister));
         })))
       return failure();
 
-    Value releaseRegister = emitStaticIntLookup(
-      StaticIntSequence::fromValues(releaseRegisters), runtimeLane,
-      scheduled.op, context.constants, context.rewriter, loc);
-    Value releaseWaitValue = emitStaticIntLookup(
-      StaticIntSequence::fromValues(releaseWaitValues), runtimeLane,
+    if (failed(emitForLanes(
+          downstreamLanes, lane, scheduled.cores.size(), scheduled.op,
+          context, loc, [&]() {
+            SpatWaitOp::create(
+              context.rewriter, loc,
+              context.constants.getIndex(restartRegister),
+              context.constants.getIndex(1));
+          })))
+      return failure();
+
+    if (failed(emitForLanes(
+          parentLanes, lane, scheduled.cores.size(), scheduled.op, context,
+          loc, [&]() {
+            Value count = emitStaticIntLookup(
+              StaticIntSequence::fromValues(childCounts), runtimeLane,
               scheduled.op, context.constants, context.rewriter, loc);
             SpatWaitOp::create(
-      context.rewriter, loc, releaseRegister, releaseWaitValue);
+              context.rewriter, loc,
+              context.constants.getIndex(restartRegister - 1), count);
+          })))
+      return failure();
 
-    auto emitChild = [&](ArrayRef<int64_t> targets,
-                         ArrayRef<int64_t> registers) {
+    auto emitChild = [&](ArrayRef<int64_t> targets) {
       Value target = emitStaticIntLookup(
         StaticIntSequence::fromValues(targets), runtimeLane, scheduled.op,
         context.constants, context.rewriter, loc);
-      Value eventRegister = emitStaticIntLookup(
-        StaticIntSequence::fromValues(registers), runtimeLane, scheduled.op,
-        context.constants, context.rewriter, loc);
-      SpatSyncOp::create(context.rewriter, loc, target, eventRegister);
+      SpatSyncOp::create(
+        context.rewriter, loc, target,
+        context.constants.getIndex(restartRegister));
     };
-    if (failed(emitForLanes(leftLanes, [&]() {
-          emitChild(leftTargets, leftRegisters);
-        }))
-        || failed(emitForLanes(rightLanes, [&]() {
-          emitChild(rightTargets, rightRegisters);
-        })))
+    if (failed(emitForLanes(
+          leftLanes, lane, scheduled.cores.size(), scheduled.op, context, loc,
+          [&]() { emitChild(leftTargets); }))
+        || failed(emitForLanes(
+          rightLanes, lane, scheduled.cores.size(), scheduled.op, context,
+          loc, [&]() { emitChild(rightTargets); })))
       return failure();
   }
   return success();
@@ -1399,6 +1506,10 @@ LogicalResult realizeDeferredBoundaries(ArrayRef<BoundaryProgram> boundaries, Ar
     if (failed(emitBoundary(boundary, results, context, replacements)))
       return boundary.key.first->op->emitOpError("phase 2 failed to realize a communication boundary");
   }
+  if (failed(emitRestartReleasePermissions(transfers, context)))
+    return failure();
+  if (failed(emitHostReleaseSynchronization(transfers, context)))
+    return failure();
   return emitCompletionSynchronization(transfers, context);
 }
 

src/PIM/Dialect/Spatial/Passes/Transforms/MergeComputeNodes/DeferredCommunicationModel.hpp

@@ -237,8 +237,6 @@ struct ExternalTransferFamily {
   StaticIntSequence hostOffsets = StaticIntSequence::uniform(0, 1);
   StaticIntSequence eventRegisters = StaticIntSequence::uniform(0, 1);
   StaticIntSequence waitValues = StaticIntSequence::uniform(1, 1);
-  StaticIntSequence acknowledgementEventRegisters =
-    StaticIntSequence::uniform(0, 1);
   bool hostRouted = false;
 };
 

src/PIM/Dialect/Spatial/Passes/Transforms/MergeComputeNodes/DeferredCommunicationRealization.cpp

@@ -6,7 +6,6 @@
 #include "DeferredCommunicationRealization.hpp"
 #include "DeferredCommunicationScheduling.hpp"
 #include "DeferredTransferPlanning.hpp"
-#include "Scheduling/PeftScheduler.hpp"
 #include "src/Accelerators/PIM/Common/PimCommon.hpp"
 #include "src/Accelerators/PIM/Common/Support/CheckedArithmetic.hpp"
 
@@ -34,67 +33,6 @@ static LogicalResult verifyNoEscapingRegionValues(Operation* owner, StringRef ph
                               << escapingUser->getName() << " at " << escapingUser->getLoc();
 }
 
-static LogicalResult placeLogicalProcessorsOnPhysicalCores(DeferredTransferPlan& plan, const SchedulingTarget& target) {
-  std::vector<Cost> logicalTrafficFlits(target.processorCount * target.processorCount, 0);
-  for (const std::unique_ptr<DeferredExchangePlan>& exchange : plan.exchanges)
-    for (const ExternalTransferFamily& transfer : exchange->external) {
-      auto fragmentType = dyn_cast<ShapedType>(transfer.requirement->publicationFragmentType);
-      if (!fragmentType || !fragmentType.hasStaticShape())
-        return exchange->deferred.emitOpError("physical core placement requires a static transfer fragment");
-      auto fragmentBytes = pim::getCheckedShapedTypeSizeInBytes(
-        fragmentType, exchange->deferred, "physical core placement transfer fragment");
-      if (failed(fragmentBytes))
-        return failure();
-      Cost flits = static_cast<Cost>(*fragmentBytes) / target.transferWidthBytes
-                 + (*fragmentBytes % target.transferWidthBytes != 0);
-      for (size_t index = 0; index < transfer.sourceCores.size(); ++index) {
-        size_t sourceLogicalProcessor = static_cast<size_t>(transfer.sourceCores.valueAt(index));
-        size_t targetLogicalProcessor = static_cast<size_t>(transfer.targetCores.valueAt(index));
-        Cost& traffic = logicalTrafficFlits[sourceLogicalProcessor * target.processorCount + targetLogicalProcessor];
-        traffic = checkedAdd(traffic, flits);
-      }
-    }
-
-  std::vector<size_t> physicalCoreForLogicalProcessor =
-    mapLogicalProcessorsToPhysicalCores(logicalTrafficFlits, target);
-  auto getPhysicalCore = [&](int64_t logicalProcessor) {
-    assert(logicalProcessor >= 0 && static_cast<size_t>(logicalProcessor) < physicalCoreForLogicalProcessor.size()
-           && "logical processor is outside the scheduling target");
-    return static_cast<int64_t>(physicalCoreForLogicalProcessor[logicalProcessor]);
-  };
-  auto remap = [&](StaticIntSequence& logicalProcessors) {
-    SmallVector<int64_t> physicalCores;
-    physicalCores.reserve(logicalProcessors.size());
-    for (size_t index = 0; index < logicalProcessors.size(); ++index)
-      physicalCores.push_back(getPhysicalCore(logicalProcessors.valueAt(index)));
-    logicalProcessors = StaticIntSequence::fromValues(physicalCores);
-  };
-
-  for (ScheduledInfo& scheduled : plan.scheduled) {
-    for (int64_t& logicalProcessor : scheduled.cores)
-      logicalProcessor = getPhysicalCore(logicalProcessor);
-    if (isa<SpatScheduledCompute>(scheduled.op)) {
-      scheduled.op->setAttr(
-        kCoreIdAttrName, IntegerAttr::get(IntegerType::get(scheduled.op->getContext(), 32), scheduled.cores.front()));
-    }
-    else {
-      SmallVector<int32_t> physicalCores;
-      physicalCores.reserve(scheduled.cores.size());
-      for (int64_t physicalCore : scheduled.cores)
-        physicalCores.push_back(static_cast<int32_t>(physicalCore));
-      scheduled.op->setAttr(kCoreIdsAttrName, DenseI32ArrayAttr::get(scheduled.op->getContext(), physicalCores));
-    }
-  }
-  for (const std::unique_ptr<ProducedValue>& produced : plan.producedStorage)
-    produced->core = getPhysicalCore(produced->core);
-  for (const std::unique_ptr<DeferredExchangePlan>& exchange : plan.exchanges)
-    for (ExternalTransferFamily& transfer : exchange->external) {
-      remap(transfer.sourceCores);
-      remap(transfer.targetCores);
-    }
-  return success();
-}
-
 static LogicalResult replaceFinalGraphPublications(func::FuncOp funcOp, DeferredTransferPlan& plan) {
   for (Operation& op : funcOp.getOps()) {
     if (!isa<SpatGraphCompute, SpatGraphComputeBatch>(op))
@@ -218,8 +156,6 @@ LogicalResult realizeDeferredCommunication(func::FuncOp funcOp,
     funcOp, materialization, pipelineStages, target.processorCount);
   if (failed(transfers))
     return funcOp.emitOpError("phase 2 failed to build symbolic transfer families");
-  if (failed(placeLogicalProcessorsOnPhysicalCores(*transfers, target)))
-    return failure();
   if (transfers->pipelineHostBufferBytes != 0) {
     auto bytes = pim::checkedCast<int64_t>(
       transfers->pipelineHostBufferBytes, funcOp,

src/PIM/Dialect/Spatial/Passes/Transforms/MergeComputeNodes/DeferredCommunicationScheduling.cpp

@@ -285,7 +285,7 @@ FailureOr<ScheduledCommunicationPlan> scheduleDeferredCommunication(func::FuncOp
     enqueue(stream);
   auto advance = [&] {
     bool changed = false;
-    while (!advanceable.empty()) {
+    while (!advanceable.empty() && ready.empty()) {
       unsigned stream = advanceable.front();
       advanceable.pop();
       StreamProgress& progress = streams[stream];
@@ -305,6 +305,8 @@ FailureOr<ScheduledCommunicationPlan> scheduleDeferredCommunication(func::FuncOp
   ScheduledCommunicationPlan result;
   unsigned finishedGroups = 0;
   while (finishedGroups != groups.size()) {
+    while (!ready.empty() && groups[ready.top()].scheduled)
+      ready.pop();
     bool progressed = advance();
     std::optional<unsigned> chosen;
     unsigned bestExtension = 0;

src/PIM/Dialect/Spatial/Passes/Transforms/MergeComputeNodes/DeferredTransferPlanning.cpp

@@ -32,9 +32,11 @@ static LogicalResult collectScheduledOperations(
     DeferredTransferPlan &plan,
     size_t pipelineStageCount,
     size_t processorCount) {
-  if (pipelineStageCount == 0 || processorCount % pipelineStageCount != 0)
+  if (pipelineStageCount == 0
+      || (pipelineStageCount > 1
+          && materialization.processorStages.size() != processorCount))
     return failure();
-  size_t stageSize = processorCount / pipelineStageCount;
+  plan.processorStages = materialization.processorStages;
   unsigned nextStream = 0;
   for (const ScheduledMaterializationRecord &record :
        materialization.materializedSchedules) {
@@ -56,8 +58,12 @@ static LogicalResult collectScheduledOperations(
       if (core >= processorCount)
         return op.emitOpError("phase 2 scheduled core is outside the target");
       info.cores.push_back(core);
-      if (pipelineStageCount > 1)
-        info.pipelineStages.push_back(core / stageSize);
+      if (pipelineStageCount > 1) {
+        size_t stage = materialization.processorStages[core];
+        if (stage >= pipelineStageCount)
+          return op.emitOpError("phase 2 scheduled core has an invalid pipeline stage");
+        info.pipelineStages.push_back(stage);
+      }
     }
     for (size_t lane = 0; lane < info.cores.size(); ++lane)
       info.streamIds.push_back(nextStream++);

src/PIM/Dialect/Spatial/Passes/Transforms/MergeComputeNodes/DeferredTransferPlanning.hpp

@@ -7,13 +7,20 @@
 
 namespace onnx_mlir::spatial {
 
+struct HostReleaseSignal {
+  int64_t writerCore = -1;
+  unsigned eventRegister = 0;
+};
+
 struct DeferredTransferPlan {
+  std::vector<size_t> processorStages;
   llvm::SmallVector<ScheduledInfo, 0> scheduled;
   llvm::SmallVector<std::unique_ptr<ProducedValue>> producedStorage;
   llvm::DenseMap<int64_t, llvm::SmallVector<ProducedValue*>> producedByGraph;
   llvm::SmallVector<std::unique_ptr<DeferredExchangePlan>> exchanges;
   llvm::SmallVector<unsigned> stepCounts;
-  llvm::DenseMap<int64_t, unsigned> hostAcknowledgementCounts;
+  llvm::DenseMap<int64_t, llvm::SmallVector<HostReleaseSignal>>
+    hostReleaseSignals;
   llvm::SmallVector<int64_t> stageZeroCores;
   llvm::SmallVector<int64_t> downstreamCores;
   size_t synchronizationRegisterCount = 0;

src/PIM/Dialect/Spatial/Passes/Transforms/MergeComputeNodes/ScheduledComputeMaterialization.cpp

@@ -808,7 +808,9 @@ materializeScheduledCompute(func::FuncOp funcOp,
     }
   }
 
-  return ScheduledComputeMaterializationResult {std::move(peftClassPlans), std::move(materializedSchedules), std::move(graphComputeToBlockMap)};
+  return ScheduledComputeMaterializationResult {
+    std::move(peftClassPlans), std::move(materializedSchedules),
+    std::move(graphComputeToBlockMap), schedule.processorStages};
 }
 
 

src/PIM/Dialect/Spatial/Passes/Transforms/MergeComputeNodes/ScheduledComputeMaterialization.hpp

@@ -14,6 +14,7 @@ struct ScheduledComputeMaterializationResult {
   llvm::MapVector<size_t, PeftClassPlan> peftClassPlans;
   std::vector<ScheduledMaterializationRecord> materializedSchedules;
   DenseMap<GraphComputeBlockKey, Block *> graphComputeToBlockMap;
+  std::vector<size_t> processorStages;
 };
 
 FailureOr<BatchFragmentSpec>

src/PIM/Dialect/Spatial/Passes/Transforms/MergeComputeNodes/ScheduledComputeReport.cpp

@@ -111,8 +111,16 @@ void dumpScheduledComputeReport(ModuleOp moduleOp, func::FuncOp funcOp, const Me
      << "  materialized homogeneous runs: " << materializedRuns << "\n"
      << "  largest run: " << largestRun << "\n"
      << "  instances compacted: " << instancesCompacted << "\n"
-     << "  compatible runs rejected: 0\n\n"
-     << "Materialized scheduled ops\n";
+     << "  compatible runs rejected: 0\n";
+  llvm::MapVector<size_t, SmallVector<size_t>> stageCores;
+  for (auto [core, stage] : llvm::enumerate(schedule.processorStages))
+    stageCores[stage].push_back(core);
+  for (const auto &[stage, cores] : stageCores) {
+    os << "  pipeline stage " << stage << " cores=";
+    printIndexedList(os, ArrayRef<size_t>(cores));
+    os << "\n";
+  }
+  os << "\nMaterialized scheduled ops\n";
 
   for (const ScheduledMaterializationRecord &record : records) {
     bool batch = isa<SpatScheduledComputeBatch>(record.scheduledOp);

src/PIM/Dialect/Spatial/Passes/Transforms/MergeComputeNodes/ScheduledSpatialPasses.cpp

@@ -10,8 +10,6 @@
 #include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Passes/Analyses/ONNXToSpatialVerifier.hpp"
 #include "src/Accelerators/PIM/Passes/PIMPasses.h"
 
-#include <limits>
-
 using namespace mlir;
 
 namespace onnx_mlir {
@@ -28,43 +26,12 @@ static bool hasValidTarget(const SchedulingTarget& target) {
 static FailureOr<func::FuncOp> requireEntry(ModuleOp moduleOp) {
   auto entry = getPimEntryFunc(moduleOp);
   if (failed(entry)) {
-    moduleOp.emitError("failed to locate the PIM entry function during Spatial scheduling and realization");
+    moduleOp.emitError("failed to locate the Pim entry function during Spatial scheduling and realization");
     return failure();
   }
   return *entry;
 }
 
-static SchedulingTarget getPipelineSchedulingTarget(
-    const SchedulingTarget& physicalTarget, size_t pipelineStages) {
-  if (pipelineStages == 1)
-    return physicalTarget;
-
-  SchedulingTarget schedulingTarget = physicalTarget;
-  schedulingTarget.processorCount = physicalTarget.processorCount / pipelineStages;
-  schedulingTarget.residentWeightCapacity = checkedMultiply(
-    physicalTarget.residentWeightCapacity, pipelineStages);
-  schedulingTarget.interProcessorLatencyNs.assign(
-    schedulingTarget.processorCount * schedulingTarget.processorCount, 0);
-  Cost latencySum = 0;
-  size_t pairCount = 0;
-  for (size_t source = 0; source < schedulingTarget.processorCount; ++source)
-    for (size_t destination = 0;
-         destination < schedulingTarget.processorCount; ++destination) {
-      Cost latency = physicalTarget.getInterProcessorLatencyNs(
-        source, destination);
-      schedulingTarget.interProcessorLatencyNs[
-        source * schedulingTarget.processorCount + destination] = latency;
-      if (source != destination) {
-        latencySum = checkedAdd(latencySum, latency);
-        ++pairCount;
-      }
-    }
-  schedulingTarget.averageInterProcessorLatencyNs = pairCount == 0
-    ? 0
-    : (latencySum + pairCount - 1) / pairCount;
-  return schedulingTarget;
-}
-
 struct ScheduleAndRealizeSpatialPass final
     : PassWrapper<ScheduleAndRealizeSpatialPass, OperationPass<ModuleOp>> {
   MLIR_DEFINE_EXPLICIT_INTERNAL_INLINE_TYPE_ID(ScheduleAndRealizeSpatialPass)
@@ -88,11 +55,10 @@ struct ScheduleAndRealizeSpatialPass final
       signalPassFailure();
       return;
     }
-    if (pipelineStages == 0 || target.processorCount % pipelineStages != 0
+    PipelineCoreLayout pipelineLayout(target.processorCount, pipelineStages);
+    if (!pipelineLayout.isValid()
         || (pipelineStages > 1
-            && target.synchronizationRegisterCount == 0)
-        || target.residentWeightCapacity
-             > std::numeric_limits<size_t>::max() / pipelineStages) {
+            && target.synchronizationRegisterCount == 0)) {
       moduleOp.emitError("ScheduleAndRealizeSpatial requires valid pipeline stages and resource counts");
       signalPassFailure();
       return;
@@ -104,30 +70,32 @@ struct ScheduleAndRealizeSpatialPass final
     }
     func::FuncOp entryFunc = *entry;
 
-    SchedulingTarget schedulingTarget = getPipelineSchedulingTarget(
-      target, pipelineStages);
     ComputeGraph scheduledGraph;
     MergeScheduleResult schedule;
     for (;;) {
-      MergeSchedulingAnalysis analysis(
-        entryFunc, schedulingTarget,
-        pipelineStages > 1 ? target.processorCount : 0);
-      scheduledGraph = analysis.getGraph();
-      schedule = std::move(analysis.getResult());
+      scheduledGraph = buildComputeGraph(entryFunc, target);
       std::string pipelineError;
-      if (succeeded(applyPipelineScheduling(
-            scheduledGraph, schedule, pipelineStages, target, pipelineError)))
-        break;
-      std::string splitError;
-      if (pipelineStages == 1
-          || failed(splitPipelineWorkload(
-                scheduledGraph, schedule, pipelineStages, target, splitError))) {
-        if (!splitError.empty())
-          pipelineError = splitError;
+      if (pipelineStages > 1) {
+        FailureOr<PipelineWorkloadPreparation> preparation =
+          preparePipelineWorkload(
+            scheduledGraph, pipelineStages, target, pipelineError);
+        if (failed(preparation)) {
           moduleOp.emitError() << pipelineError;
           signalPassFailure();
           return;
         }
+        if (*preparation == PipelineWorkloadPreparation::Changed)
+          continue;
+      }
+      MergeSchedulingAnalysis analysis(
+        scheduledGraph, target, entryFunc.getContext());
+      schedule = std::move(analysis.getResult());
+      if (succeeded(applyPipelineScheduling(
+            scheduledGraph, schedule, pipelineStages, target, pipelineError)))
+        break;
+      moduleOp.emitError() << pipelineError;
+      signalPassFailure();
+      return;
     }
     PatternRewriter rewriter(moduleOp.getContext());
     FailureOr<ScheduledComputeMaterializationResult> materialization =

src/PIM/Dialect/Spatial/Passes/Transforms/MergeComputeNodes/Scheduling/MergeSchedule.hpp

@@ -14,6 +14,7 @@ namespace spatial {
 
 struct MergeScheduleResult {
   size_t processorCount = 0;
+  std::vector<size_t> processorStages;
   std::vector<ComputeInstance> dominanceOrderCompute;
   llvm::DenseMap<ComputeInstance, size_t> computeToCpuMap;
   llvm::DenseMap<ComputeInstance, size_t> computeToCpuSlotMap;

src/PIM/Dialect/Spatial/Passes/Transforms/MergeComputeNodes/Scheduling/MergeSchedulingAnalysis.cpp

@@ -88,22 +88,21 @@ void verifySchedule(const ComputeGraph& graph,
 
 } // namespace
 
-MergeSchedulingAnalysis::MergeSchedulingAnalysis(mlir::Operation* op,
+MergeSchedulingAnalysis::MergeSchedulingAnalysis(const ComputeGraph& computeGraph,
                                                  const SchedulingTarget& schedulingTarget,
-                                                 size_t partitionCount)
-: entryOp(op), target(schedulingTarget), computePartitionCount(partitionCount) {
+                                                 mlir::MLIRContext* context)
+: context(context), target(schedulingTarget), graph(computeGraph) {
   result = run();
 }
 
 MergeScheduleResult MergeSchedulingAnalysis::run() {
-  graph = buildComputeGraph(entryOp, target, computePartitionCount);
   if (!verifyAcyclic(graph))
     llvm::report_fatal_error("merge scheduling: compute graph is cyclic");
 
   MergeScheduleResult schedule = runPeftScheduler(
     graph, PeftScheduleOptions {
       target,
-      entryOp->getContext()});
+      context});
   verifySchedule(graph, schedule, target);
   return schedule;
 }

src/PIM/Dialect/Spatial/Passes/Transforms/MergeComputeNodes/Scheduling/MergeSchedulingAnalysis.hpp

@@ -11,17 +11,15 @@ namespace spatial {
 
 class MergeSchedulingAnalysis {
 public:
-  MergeSchedulingAnalysis(mlir::Operation* op,
+  MergeSchedulingAnalysis(const ComputeGraph& graph,
                           const SchedulingTarget& target,
-                          size_t computePartitionCount = 0);
+                          mlir::MLIRContext* context);
   MergeScheduleResult& getResult() { return result; }
-  const ComputeGraph& getGraph() const { return graph; }
 
 private:
-  mlir::Operation* entryOp = nullptr;
+  mlir::MLIRContext* context = nullptr;
   const SchedulingTarget& target;
-  size_t computePartitionCount = 0;
-  ComputeGraph graph;
+  const ComputeGraph& graph;
   MergeScheduleResult result;
 
   MergeScheduleResult run();

src/PIM/Dialect/Spatial/Passes/Transforms/MergeComputeNodes/Scheduling/PeftScheduler.cpp

@@ -5,7 +5,6 @@
 #include "llvm/Support/FormatVariadic.h"
 
 #include <limits>
-#include <numeric>
 #include <optional>
 #include <queue>
 #include <tuple>
@@ -42,11 +41,6 @@ struct TopologyModel {
   }
 };
 
-Time getAverageTransferTime(const TransferCost& transferCost, const SchedulingTarget& target) {
-  return checkedAdd(transferCost.fixed,
-                    checkedMultiply(transferCost.networkFlits, target.averageInterProcessorLatencyNs));
-}
-
 std::vector<std::vector<size_t>> buildReverseLevels(const ComputeGraph& graph) {
   std::vector<size_t> remainingSuccessors(graph.nodes.size(), 0);
   std::queue<size_t> readySinks;
@@ -83,24 +77,6 @@ std::vector<std::vector<size_t>> buildReverseLevels(const ComputeGraph& graph) {
   return reverseLevels;
 }
 
-void verifyOctTableSize(size_t nodeCount, size_t processorCount) {
-  constexpr size_t kMaxOctTableBytes = 1ull << 35;
-  if (nodeCount == 0 || processorCount == 0)
-    return;
-  if (processorCount > std::numeric_limits<size_t>::max() / sizeof(Time))
-    llvm::report_fatal_error("PEFT scheduler: OCT table size overflow");
-  size_t rowBytes = processorCount * sizeof(Time);
-  if (nodeCount > std::numeric_limits<size_t>::max() / rowBytes)
-    llvm::report_fatal_error("PEFT scheduler: OCT table size overflow");
-  size_t totalBytes = nodeCount * rowBytes;
-  if (totalBytes > kMaxOctTableBytes) {
-    std::string message = llvm::formatv("PEFT scheduler: OCT table would require {0} MiB, exceeding the 1024 MiB guard",
-                                        totalBytes / (1024 * 1024))
-                            .str();
-    llvm::report_fatal_error(llvm::StringRef(message));
-  }
-}
-
 bool hasHighResidentWeightPressure(const ComputeGraph& graph, size_t processorCount, size_t residentWeightCapacity) {
   if (residentWeightCapacity > std::numeric_limits<size_t>::max() / processorCount)
     return false;
@@ -243,73 +219,6 @@ FailureOr<LanePublicationSignatures> buildLanePublicationSignatures(SpatComputeB
 
 } // namespace
 
-std::vector<size_t> mapLogicalProcessorsToPhysicalCores(ArrayRef<Cost> logicalTrafficFlits,
-                                                        const SchedulingTarget& target) {
-  const size_t processorCount = target.processorCount;
-  assert(logicalTrafficFlits.size() == processorCount * processorCount
-         && "logical traffic matrix must cover every processor pair");
-
-  std::vector<size_t> physicalCoreForLogicalProcessor(processorCount);
-  std::iota(physicalCoreForLogicalProcessor.begin(), physicalCoreForLogicalProcessor.end(), 0);
-
-  auto transferCost = [&](size_t sourceLogicalProcessor,
-                          size_t targetLogicalProcessor,
-                          size_t sourcePhysicalCore,
-                          size_t targetPhysicalCore) {
-    Cost traffic = logicalTrafficFlits[sourceLogicalProcessor * processorCount + targetLogicalProcessor];
-    return checkedMultiply(traffic, target.getInterProcessorLatencyNs(sourcePhysicalCore, targetPhysicalCore));
-  };
-
-  for (size_t logicalProcessor = 0; logicalProcessor < processorCount; ++logicalProcessor) {
-    size_t bestPeerLogicalProcessor = logicalProcessor;
-    Cost bestSaving = 0;
-    for (size_t peerLogicalProcessor = 0; peerLogicalProcessor < processorCount; ++peerLogicalProcessor) {
-      if (peerLogicalProcessor == logicalProcessor)
-        continue;
-      size_t physicalCore = physicalCoreForLogicalProcessor[logicalProcessor];
-      size_t peerPhysicalCore = physicalCoreForLogicalProcessor[peerLogicalProcessor];
-      Cost currentCost = 0;
-      Cost swappedCost = 0;
-      for (size_t otherLogicalProcessor = 0; otherLogicalProcessor < processorCount; ++otherLogicalProcessor) {
-        if (otherLogicalProcessor == logicalProcessor || otherLogicalProcessor == peerLogicalProcessor)
-          continue;
-        size_t otherPhysicalCore = physicalCoreForLogicalProcessor[otherLogicalProcessor];
-        currentCost = checkedAdd(
-          currentCost, transferCost(logicalProcessor, otherLogicalProcessor, physicalCore, otherPhysicalCore));
-        currentCost = checkedAdd(
-          currentCost, transferCost(otherLogicalProcessor, logicalProcessor, otherPhysicalCore, physicalCore));
-        currentCost = checkedAdd(
-          currentCost, transferCost(peerLogicalProcessor, otherLogicalProcessor, peerPhysicalCore, otherPhysicalCore));
-        currentCost = checkedAdd(
-          currentCost, transferCost(otherLogicalProcessor, peerLogicalProcessor, otherPhysicalCore, peerPhysicalCore));
-        swappedCost = checkedAdd(
-          swappedCost, transferCost(logicalProcessor, otherLogicalProcessor, peerPhysicalCore, otherPhysicalCore));
-        swappedCost = checkedAdd(
-          swappedCost, transferCost(otherLogicalProcessor, logicalProcessor, otherPhysicalCore, peerPhysicalCore));
-        swappedCost = checkedAdd(
-          swappedCost, transferCost(peerLogicalProcessor, otherLogicalProcessor, physicalCore, otherPhysicalCore));
-        swappedCost = checkedAdd(
-          swappedCost, transferCost(otherLogicalProcessor, peerLogicalProcessor, otherPhysicalCore, physicalCore));
-      }
-      currentCost =
-        checkedAdd(currentCost, transferCost(logicalProcessor, peerLogicalProcessor, physicalCore, peerPhysicalCore));
-      currentCost =
-        checkedAdd(currentCost, transferCost(peerLogicalProcessor, logicalProcessor, peerPhysicalCore, physicalCore));
-      swappedCost =
-        checkedAdd(swappedCost, transferCost(logicalProcessor, peerLogicalProcessor, peerPhysicalCore, physicalCore));
-      swappedCost =
-        checkedAdd(swappedCost, transferCost(peerLogicalProcessor, logicalProcessor, physicalCore, peerPhysicalCore));
-      if (currentCost > swappedCost && currentCost - swappedCost > bestSaving) {
-        bestSaving = currentCost - swappedCost;
-        bestPeerLogicalProcessor = peerLogicalProcessor;
-      }
-    }
-    std::swap(physicalCoreForLogicalProcessor[logicalProcessor],
-              physicalCoreForLogicalProcessor[bestPeerLogicalProcessor]);
-  }
-  return physicalCoreForLogicalProcessor;
-}
-
 MergeScheduleResult runPeftScheduler(const ComputeGraph& graph, const PeftScheduleOptions& options) {
   const size_t nodeCount = graph.nodes.size();
   const size_t processorCount = options.target.processorCount;
@@ -321,36 +230,17 @@ MergeScheduleResult runPeftScheduler(const ComputeGraph& graph, const PeftSchedu
   std::vector<ResidentWeightSet> capacityReservations = planResidentWeightReservations(
     graph, processorCount, options.target.residentWeightCapacity, topology, preferWeightReuse);
 
-  verifyOctTableSize(nodeCount, processorCount);
   std::vector<std::vector<size_t>> reverseLevels = buildReverseLevels(graph);
 
-  // MOCK: Replace this with your actual heterogeneous cost lookup.
-  // If graph.nodes[task] is modified to hold a vector of costs per processor, access it here.
-  auto getComputeCost = [&](size_t task, size_t processor) -> Time { return graph.nodes[task].cost; };
-  std::vector<Time> oct(nodeCount * processorCount, 0);
-  std::vector<Time> minOctPlusComp(nodeCount, 0);
-
-  // 1. O(P(E+V)) Heterogeneous OCT Calculation
+  // Compute costs are identical on every core. The optimistic successor can
+  // stay on the same core for zero transfer cost, so every physical OCT column
+  // is identical even with nonuniform links. Store that exact lower bound once.
+  std::vector<Time> oct(nodeCount, 0);
   for (const std::vector<size_t>& levelNodes : reverseLevels) {
     auto computeNodeOct = [&](size_t levelIndex) {
       size_t task = levelNodes[levelIndex];
-      std::vector<Time> maxVals(processorCount, 0);
-
-      for (const auto& [succ, comm] : graph.successors[task]) {
-        Time valDifferentCpu = addOrMax(minOctPlusComp[succ], getAverageTransferTime(comm, options.target));
-        for (size_t processor = 0; processor < processorCount; ++processor) {
-          Time valSameCpu = addOrMax(oct[succ * processorCount + processor], getComputeCost(succ, processor));
-          Time bestSucc = std::min(valSameCpu, valDifferentCpu);
-          maxVals[processor] = std::max(maxVals[processor], bestSucc);
-        }
-      }
-
-      Time minForPreds = std::numeric_limits<Time>::max();
-      for (size_t processor = 0; processor < processorCount; ++processor) {
-        oct[task * processorCount + processor] = maxVals[processor];
-        minForPreds = std::min(minForPreds, addOrMax(maxVals[processor], getComputeCost(task, processor)));
-      }
-      minOctPlusComp[task] = minForPreds == std::numeric_limits<Time>::max() ? 0 : minForPreds;
+      for (const auto& [succ, comm] : graph.successors[task])
+        oct[task] = std::max(oct[task], addOrMax(oct[succ], graph.nodes[succ].cost));
     };
 
     if (options.context != nullptr)
@@ -367,9 +257,7 @@ MergeScheduleResult runPeftScheduler(const ComputeGraph& graph, const PeftSchedu
   };
   std::vector<RankEntry> ranks(nodeCount);
   auto computeRank = [&](size_t node) {
-    long double rank = 0.0L;
-    for (size_t processor = 0; processor < processorCount; ++processor)
-      rank += static_cast<long double>(oct[node * processorCount + processor]);
+    long double rank = static_cast<long double>(oct[node]);
     ranks[node] = {rank, node, graph.nodes[node].originalOrder};
   };
 
@@ -399,7 +287,6 @@ MergeScheduleResult runPeftScheduler(const ComputeGraph& graph, const PeftSchedu
     }
 
     std::vector<char> scheduled(nodeCount, false);
-    std::vector<ResidentWeightSet> reservations = capacityReservations;
     std::vector<ResidentWeightSet> processorResidentWeights(processorCount);
     std::vector<ScheduledTask> schedules(nodeCount);
     std::vector<std::vector<size_t>> tasksByProcessor(processorCount);
@@ -429,9 +316,9 @@ MergeScheduleResult runPeftScheduler(const ComputeGraph& graph, const PeftSchedu
         unsigned int overlapCount =
           countResidentWeightOverlap(processorResidentWeights[processor], graph.nodes[task].residentWeights);
         size_t residentWeightUnion =
-          getResidentWeightUnionSize(reservations[processor], graph.nodes[task].residentWeights);
+          getResidentWeightUnionSize(capacityReservations[processor], graph.nodes[task].residentWeights);
         smallestResidentWeightUnion = std::min(smallestResidentWeightUnion, residentWeightUnion);
-        if (!graph.nodes[task].residentWeights.empty() && residentWeightUnion > options.target.residentWeightCapacity) {
+        if (!graph.nodes[task].residentWeights.empty() && residentWeightUnion > capacityReservations[processor].size()) {
           residentWeightRejected = true;
           continue;
         }
@@ -442,7 +329,7 @@ MergeScheduleResult runPeftScheduler(const ComputeGraph& graph, const PeftSchedu
           dataReady = std::max(dataReady, addOrMax(predSchedule.endTime, commPenalty));
         }
 
-        Time computeCost = getComputeCost(task, processor);
+        Time computeCost = graph.nodes[task].cost;
         Time est = dataReady;
         Time currentEnd = 0;
         bool foundGap = false;
@@ -461,7 +348,7 @@ MergeScheduleResult runPeftScheduler(const ComputeGraph& graph, const PeftSchedu
           est = std::max(currentEnd, dataReady);
 
         Time eft = addOrMax(est, computeCost);
-        Time oeft = addOrMax(eft, oct[task * processorCount + processor]);
+        Time oeft = addOrMax(eft, oct[task]);
         size_t centerDistance = topology.getCenterDistance(processor);
         size_t taskCount = tasksByProcessor[processor].size();
         bool betterResidentWeightChoice =
@@ -513,7 +400,7 @@ MergeScheduleResult runPeftScheduler(const ComputeGraph& graph, const PeftSchedu
           const ComputeInstance& instance = graph.nodes[task].instance;
           std::string message =
             llvm::formatv("PEFT scheduler: no valid processor for task {0} (lanes {1}..{2}, {3} distinct weights); "
-                          "smallest processor union is {4}, exceeding resident-weight capacity {5}",
+                          "no physical reservation contains the task weights (smallest union {4}, core capacity {5})",
                           graph.nodes[task].originalOrder,
                           instance.laneStart,
                           instance.laneStart + instance.laneCount,
@@ -533,7 +420,6 @@ MergeScheduleResult runPeftScheduler(const ComputeGraph& graph, const PeftSchedu
       schedules[task] = {bestProcessor, bestEst, bestEft};
       scheduled[task] = true;
       ++scheduledCount;
-      insertResidentWeights(reservations[bestProcessor], graph.nodes[task].residentWeights);
       insertResidentWeights(processorResidentWeights[bestProcessor], graph.nodes[task].residentWeights);
 
       auto& timeline = timelineByProcessor[bestProcessor];

src/PIM/Dialect/Spatial/Passes/Transforms/MergeComputeNodes/Scheduling/PeftScheduler.hpp

@@ -21,15 +21,11 @@ inline Time getPeftTransferTime(const TransferCost& transferCost,
   if (sourceProcessor == targetProcessor)
     return 0;
   return checkedAdd(transferCost.fixed,
-                    checkedMultiply(transferCost.networkFlits, target.averageInterProcessorLatencyNs));
+                    checkedMultiply(transferCost.networkFlits, target.getInterProcessorLatencyNs(sourceProcessor, targetProcessor)));
 }
 
-// PEFT assigns logical processors. Physical core IDs are chosen only after
-// materialization exposes the exact transfer traffic.
+// PEFT assigns physical cores using the injected target topology.
 MergeScheduleResult runPeftScheduler(const ComputeGraph& graph, const PeftScheduleOptions& options);
 
-std::vector<size_t> mapLogicalProcessorsToPhysicalCores(llvm::ArrayRef<Cost> logicalTrafficFlits,
-                                                        const SchedulingTarget& target);
-
 } // namespace spatial
 } // namespace onnx_mlir

src/PIM/Dialect/Spatial/Passes/Transforms/MergeComputeNodes/Scheduling/PipelineScheduling.hpp

@@ -2,8 +2,15 @@
 
 #include "mlir/Support/LogicalResult.h"
 
+#include "llvm/ADT/ArrayRef.h"
+#include "llvm/ADT/STLExtras.h"
+
+#include <algorithm>
 #include <cstddef>
+#include <numeric>
+#include <optional>
 #include <string>
+#include <vector>
 
 #include "ComputeGraph.hpp"
 #include "MergeSchedule.hpp"
@@ -11,15 +18,86 @@
 
 namespace onnx_mlir::spatial {
 
+struct PipelineStageRange {
+  size_t begin;
+  size_t size;
+};
+
+class PipelineCoreLayout {
+public:
+  PipelineCoreLayout(size_t processorCount, size_t stageCount)
+      : processorCount(processorCount), stageSizes(stageCount) {
+    if (stageCount == 0)
+      return;
+    size_t baseSize = processorCount / stageCount;
+    size_t largerStageCount = processorCount % stageCount;
+    for (size_t stage = 0; stage < stageCount; ++stage)
+      stageSizes[stage] = baseSize + (stage < largerStageCount);
+  }
+
+  explicit PipelineCoreLayout(llvm::ArrayRef<size_t> stageSizes)
+      : processorCount(std::accumulate(
+          stageSizes.begin(), stageSizes.end(), size_t {0})),
+        stageSizes(stageSizes.begin(), stageSizes.end()) {}
+
+  PipelineCoreLayout(llvm::ArrayRef<size_t> stageSizes,
+                     const SchedulingTarget &target);
+
+  bool isValid() const {
+    return !stageSizes.empty()
+        && llvm::none_of(stageSizes, [](size_t size) { return size == 0; });
+  }
+
+  size_t getStageCount() const { return stageSizes.size(); }
+
+  size_t getProcessorCount() const { return processorCount; }
+
+  llvm::ArrayRef<size_t> getStageSizes() const { return stageSizes; }
+
+  size_t getPhysicalCore(size_t position) const {
+    return physicalCores.empty() ? position : physicalCores[position];
+  }
+
+  PipelineStageRange getStageRange(size_t stage) const {
+    return {std::accumulate(
+              stageSizes.begin(), stageSizes.begin() + stage, size_t {0}),
+            stageSizes[stage]};
+  }
+
+  std::optional<size_t> getStageForCore(size_t core) const {
+    if (!isValid() || core >= processorCount)
+      return std::nullopt;
+    if (!physicalCoreStages.empty())
+      return physicalCoreStages[core];
+    size_t end = 0;
+    for (auto [stage, size] : llvm::enumerate(stageSizes)) {
+      end += size;
+      if (core < end)
+        return stage;
+    }
+    return std::nullopt;
+  }
+
+private:
+  size_t processorCount;
+  std::vector<size_t> stageSizes;
+  std::vector<size_t> physicalCores;
+  std::vector<size_t> physicalCoreStages;
+};
+
 mlir::LogicalResult applyPipelineScheduling(const ComputeGraph& graph,
                                             MergeScheduleResult& schedule,
                                             size_t pipelineStages,
                                             const SchedulingTarget& physicalTarget,
                                             std::string& error);
 
-mlir::LogicalResult splitPipelineWorkload(const ComputeGraph& graph,
-                                          const MergeScheduleResult& schedule,
-                                          size_t pipelineStages,
+enum class PipelineWorkloadPreparation {
+  Ready,
+  Changed,
+};
+
+mlir::FailureOr<PipelineWorkloadPreparation> preparePipelineWorkload(
+    const ComputeGraph& graph, size_t pipelineStages,
     const SchedulingTarget& physicalTarget,
     std::string& error);
 

src/PIM/Dialect/Spatial/Passes/Transforms/MergeComputeNodes/SpatialDataflowCsvExporter.cpp

@@ -193,7 +193,7 @@ FailureOr<TopLevelOpInfo> buildTopLevelOpInfo(Operation& op, bool isScheduled, s
 
   if constexpr (std::is_same_v<ComputeOpTy, SpatScheduledCompute>) {
     if (auto compute = dyn_cast<ComputeOpTy>(&op)) {
-      auto coreId = getOptionalScheduledCoreId(compute, "spatial dataflow export core id");
+      auto coreId = getOptionalScheduledCoreId(compute, "Spatial dataflow export core id");
       if (failed(coreId))
         return failure();
       if (*coreId)
@@ -207,7 +207,7 @@ FailureOr<TopLevelOpInfo> buildTopLevelOpInfo(Operation& op, bool isScheduled, s
 template <typename BatchOpTy>
 FailureOr<SmallVector<int32_t, 8>> getBatchLaneCoreIds(BatchOpTy batch) {
   if constexpr (std::is_same_v<BatchOpTy, SpatScheduledComputeBatch>) {
-    auto coreIds = getOptionalScheduledBatchCoreIds(batch, "spatial dataflow export core ids");
+    auto coreIds = getOptionalScheduledBatchCoreIds(batch, "Spatial dataflow export core ids");
     if (failed(coreIds))
       return failure();
     if (!*coreIds)

src/PIM/Dialect/Spatial/Spatial.td

@@ -13,7 +13,7 @@ include "mlir/Interfaces/SideEffectInterfaces.td"
 
 def SpatialDialect : Dialect {
   let name = "spat";
-  let summary = "Dialect designed for deep learning computation in a spatial architecture";
+  let summary = "Dialect designed for deep learning computation in a Spatial architecture";
   let cppNamespace = "::onnx_mlir::spatial";
   let useDefaultAttributePrinterParser = 0;
   let extraClassDeclaration = [{
@@ -592,15 +592,14 @@ def SpatHostStoreSyncOp : SpatOp<"host_store_sync", []> {
 }
 
 def SpatHostWaitLoadOp : SpatOp<"host_wait_load", []> {
-  let summary = "Wait for producers, load from host memory, and acknowledge consumption";
+  let summary = "Wait for producers and load from host memory";
 
   let arguments = (ins
     Index:$sourceCoreId,
     Index:$targetCoreId,
     Index:$hostOffset,
     Index:$eventRegister,
-    Index:$waitValue,
-    Index:$acknowledgementEventRegister
+    Index:$waitValue
   );
 
   let results = (outs
@@ -610,7 +609,7 @@ def SpatHostWaitLoadOp : SpatOp<"host_wait_load", []> {
   let assemblyFormat = [{
     `from` $sourceCoreId `to` $targetCoreId
       `host_offset` $hostOffset `event` $eventRegister `count` $waitValue
-      `ack` $acknowledgementEventRegister attr-dict `:` type($output)
+      attr-dict `:` type($output)
   }];
 }
 

src/PIM/Passes/PIMPasses.h

@@ -25,7 +25,8 @@ std::unique_ptr<mlir::Pass> createONNXToSpatialPass(
   const spatial::SpatialTargetResources& target,
   const ONNXToSpatialPlanningOptions& options);
 std::unique_ptr<mlir::Pass> createSpatialLayoutPlanningPass();
-std::unique_ptr<mlir::Pass> createSpatialLayoutPlanningPass(const spatial::SpatialTargetResources& target);
+std::unique_ptr<mlir::Pass> createSpatialLayoutPlanningPass(
+  const spatial::SpatialTargetResources& target, bool selectTrivialPlan = false);
 std::unique_ptr<mlir::Pass> createLowerSpatialPlansPass();
 std::unique_ptr<mlir::Pass> createLowerSpatialPlansPass(
   const spatial::SpatialTargetResources& target,

src/PIM/Passes/PimCodegen/EmitPimCodePass.cpp

@@ -12,7 +12,7 @@ namespace {
 struct EmitPimCodePass : PassWrapper<EmitPimCodePass, OperationPass<ModuleOp>> {
   MLIR_DEFINE_EXPLICIT_INTERNAL_INLINE_TYPE_ID(EmitPimCodePass);
   StringRef getArgument() const override { return "emit-pim-code-pass"; }
-  StringRef getDescription() const override { return "Emit PIM simulator code artifacts"; }
+  StringRef getDescription() const override { return "Emit Pim simulator code artifacts"; }
 
   EmitPimCodePass() {}
   EmitPimCodePass(const EmitPimCodePass& pass) {}
@@ -25,7 +25,7 @@ struct EmitPimCodePass : PassWrapper<EmitPimCodePass, OperationPass<ModuleOp>> {
 
     int compiler_error_code = compileToPimCode(moduleOp, pimDir);
     if (compiler_error_code != CompilerSuccess) {
-      moduleOp.emitError() << "failed to emit PIM simulator code artifacts; compiler error code "
+      moduleOp.emitError() << "failed to emit Pim simulator code artifacts; compiler error code "
                            << compiler_error_code;
       signalPassFailure();
     }

test/PIM/CMakeLists.txt

@@ -36,3 +36,13 @@ add_pim_unittest(SpatialSchedulingTargetTest
   LINK_LIBS PRIVATE
   OMPimCompilerUtils
 )
+
+add_test(
+  NAME PimHostReuseSynchronizationTest
+  COMMAND "${PYTHON_EXECUTABLE}"
+          "${CMAKE_SOURCE_DIR}/validation/tools/pim/pimcomp/compare/test_PIMCOMP_adversarial_memory_sync.py"
+          --compiler "$<TARGET_FILE:onnx-mlir>"
+          --simple-model "${CMAKE_SOURCE_DIR}/validation/operations/relu/after_conv/relu_after_conv.onnx"
+          --grouped-model "${CMAKE_SOURCE_DIR}/validation/operations/conv/relu_conv_store/conv_relu_conv_store.onnx"
+)
+set_tests_properties(PimHostReuseSynchronizationTest PROPERTIES LABELS pim-unittest)

test/PIM/SpatialSchedulingTargetTest.cpp

@@ -3,12 +3,183 @@
 #include <string>
 #include <vector>
 
+#include "mlir/IR/BuiltinOps.h"
+
+#include "src/Accelerators/PIM/Dialect/Spatial/Passes/Transforms/MergeComputeNodes/DeferredCommunicationScheduling.hpp"
+#include "src/Accelerators/PIM/Dialect/Spatial/Passes/Transforms/MergeComputeNodes/DeferredTransferPlanning.hpp"
 #include "src/Accelerators/PIM/Dialect/Spatial/Passes/Transforms/MergeComputeNodes/Scheduling/PeftScheduler.hpp"
 #include "src/Accelerators/PIM/Dialect/Spatial/Passes/Transforms/MergeComputeNodes/Scheduling/PipelineScheduling.hpp"
 
 using namespace onnx_mlir::spatial;
 
+static void testPhysicalPeft() {
+  TransferCost transfer {.fixed = 50, .networkFlits = 4};
+  SchedulingTarget line;
+  line.processorCount = 3;
+  line.interProcessorLatencyNs = {
+    0,
+    1,
+    10,
+    1,
+    0,
+    1,
+    10,
+    1,
+    0,
+  };
+  line.averageInterProcessorLatencyNs = 4;
+  assert(getPeftTransferTime(transfer, 0, 1, line) == 54);
+  assert(getPeftTransferTime(transfer, 0, 2, line) == 90);
+  SchedulingTarget reversed = line;
+  reversed.interProcessorLatencyNs = {0, 10, 1, 10, 0, 1, 1, 1, 0};
+  assert(getPeftTransferTime(transfer, 0, 1, reversed) == 90);
+  assert(getPeftTransferTime(transfer, 0, 2, reversed) == 54);
+
+  mlir::MLIRContext context;
+  mlir::OwningOpRef<mlir::ModuleOp> owner =
+    mlir::ModuleOp::create(mlir::UnknownLoc::get(&context));
+  ComputeGraph chain;
+  chain.successors.resize(3);
+  chain.predecessors.resize(3);
+  for (uint32_t task = 0; task < 3; ++task) {
+    ComputeInstance instance {owner->getOperation(), task, 1};
+    ResidentWeight weight;
+    weight.opaqueLane = task;
+    chain.nodes.push_back({instance, 10, {weight}, task});
+    if (task != 0) {
+      chain.successors[task - 1].push_back({task, transfer});
+      chain.predecessors[task].push_back({task - 1, transfer});
+    }
+  }
+  line.residentWeightCapacity = reversed.residentWeightCapacity = 3;
+  for (const SchedulingTarget &target : {line, reversed}) {
+    auto scheduled = runPeftScheduler(chain, {target, &context});
+    size_t firstCore = scheduled.computeToCpuMap.lookup(chain.nodes[0].instance);
+    assert(firstCore == (target.interProcessorLatencyNs == line.interProcessorLatencyNs ? 1 : 2));
+    for (size_t task = 1; task < chain.nodes.size(); ++task) {
+      auto previous = chain.nodes[task - 1].instance;
+      auto current = chain.nodes[task].instance;
+      size_t source = scheduled.computeToCpuMap.lookup(previous);
+      size_t destination = scheduled.computeToCpuMap.lookup(current);
+      assert(source != destination);
+      assert(scheduled.computeToAestMap.lookup(current)
+             == scheduled.computeToAestMap.lookup(previous) + 10
+                  + getPeftTransferTime(transfer, source, destination, target));
+    }
+  }
+}
+
+static void testPipelineInputArrival() {
+  ComputeGraph graph;
+  graph.successors.resize(4);
+  graph.predecessors.resize(4);
+  MergeScheduleResult schedule;
+  schedule.processorCount = 4;
+  for (uint32_t task = 0; task < 4; ++task) {
+    ComputeInstance instance {nullptr, task, 1};
+    ResidentWeight weight;
+    weight.opaqueLane = task;
+    graph.nodes.push_back({instance, 1, {weight}, task});
+    schedule.dominanceOrderCompute.push_back(instance);
+    schedule.computeToCpuMap[instance] = task;
+    schedule.computeToCpuSlotMap[instance] = 0;
+    if (task != 0) {
+      TransferCost transfer {.fixed = 0, .networkFlits = 100};
+      graph.predecessors[task].push_back({task - 1, transfer});
+      graph.successors[task - 1].push_back({task, transfer});
+    }
+  }
+  SchedulingTarget target;
+  target.processorCount = 4;
+  target.residentWeightCapacity = 4;
+  target.interProcessorLatencyNs.assign(16, 1);
+  for (size_t core = 0; core < 4; ++core)
+    target.interProcessorLatencyNs[core * 4 + core] = 0;
+  std::string error;
+  assert(mlir::succeeded(applyPipelineScheduling(graph, schedule, 2, target, error)));
+  for (size_t first : {0, 2}) {
+    size_t source = schedule.computeToCpuMap.lookup(graph.nodes[first].instance);
+    size_t destination = schedule.computeToCpuMap.lookup(graph.nodes[first + 1].instance);
+    assert(schedule.processorStages[source] == first / 2);
+    // An idle remote core would finish later than waiting for the local core.
+    assert(source == destination);
+  }
+}
+
+static void testReadyCommunication() {
+  mlir::MLIRContext context;
+  mlir::OwningOpRef<mlir::ModuleOp> owner =
+    mlir::ModuleOp::create(mlir::UnknownLoc::get(&context));
+  auto type = mlir::RankedTensorType::get({1}, mlir::Float32Type::get(&context));
+  mlir::Block payloads;
+  auto payload = payloads.addArgument(type, owner->getLoc());
+  DeferredTransferPlan plan;
+  plan.stepCounts = {3, 1, 1};
+  plan.scheduled.resize(3);
+  for (auto [index, info] : llvm::enumerate(plan.scheduled)) {
+    info.op = owner->getOperation();
+    info.streamIds.push_back(index);
+  }
+  ProducedValue producer;
+  producer.scheduled = &plan.scheduled[0];
+  producer.payload = payload;
+  // Two consumers exercise removal of already-scheduled ready-queue entries.
+  for (unsigned target = 1; target <= 2; ++target) {
+    auto exchange = std::make_unique<DeferredExchangePlan>();
+    exchange->target = &plan.scheduled[target];
+    exchange->exchangeId = target;
+    RequirementFamily requirement;
+    requirement.producer = &producer;
+    requirement.publicationFragmentType = type;
+    exchange->requirements.push_back(requirement);
+    ExternalTransferFamily transfer;
+    transfer.requirement = &exchange->requirements.front();
+    transfer.sourceScheduled = producer.scheduled;
+    transfer.targetScheduled = exchange->target;
+    transfer.targetLanes = LaneSet::all(1);
+    transfer.targetStreams = onnx_mlir::StaticIntSequence::uniform(target, 1);
+    exchange->external.push_back(transfer);
+    plan.exchanges.push_back(std::move(exchange));
+  }
+  auto scheduled = scheduleDeferredCommunication({}, plan);
+  assert(mlir::succeeded(scheduled));
+  assert(scheduled->slices.size() == 2);
+  for (const auto &slice : scheduled->slices)
+    assert(slice.sourceInsertionStep == 1);
+}
+
 int main() {
+  testPipelineInputArrival();
+  testReadyCommunication();
+  PipelineCoreLayout unevenLayout(138, 4);
+  assert(unevenLayout.isValid());
+  assert(unevenLayout.getStageRange(0).begin == 0);
+  assert(unevenLayout.getStageRange(0).size == 35);
+  assert(unevenLayout.getStageRange(1).begin == 35);
+  assert(unevenLayout.getStageRange(1).size == 35);
+  assert(unevenLayout.getStageRange(2).begin == 70);
+  assert(unevenLayout.getStageRange(2).size == 34);
+  assert(unevenLayout.getStageRange(3).begin == 104);
+  assert(unevenLayout.getStageRange(3).size == 34);
+  assert(unevenLayout.getStageForCore(34) == 0);
+  assert(unevenLayout.getStageForCore(35) == 1);
+  assert(unevenLayout.getStageForCore(69) == 1);
+  assert(unevenLayout.getStageForCore(70) == 2);
+  assert(unevenLayout.getStageForCore(137) == 3);
+  assert(!unevenLayout.getStageForCore(138));
+
+  PipelineCoreLayout dynamicLayout(std::vector<size_t> {2, 4, 1, 3});
+  assert(dynamicLayout.isValid());
+  assert(dynamicLayout.getProcessorCount() == 10);
+  assert(dynamicLayout.getStageRange(0).begin == 0);
+  assert(dynamicLayout.getStageRange(1).begin == 2);
+  assert(dynamicLayout.getStageRange(2).begin == 6);
+  assert(dynamicLayout.getStageRange(3).begin == 7);
+  assert(dynamicLayout.getStageForCore(1) == 0);
+  assert(dynamicLayout.getStageForCore(2) == 1);
+  assert(dynamicLayout.getStageForCore(6) == 2);
+  assert(dynamicLayout.getStageForCore(9) == 3);
+
   TransferCost transfer {.fixed = 50, .networkFlits = 4};
 
   SchedulingTarget fast;
@@ -26,36 +197,7 @@ int main() {
   assert(fast.getInterProcessorLatencyNs(0, 1) == 3);
   assert(slow.getInterProcessorLatencyNs(0, 1) == 10);
 
-  SchedulingTarget line;
-  line.processorCount = 3;
-  line.interProcessorLatencyNs = {
-    0,
-    1,
-    10,
-    1,
-    0,
-    1,
-    10,
-    1,
-    0,
-  };
-  std::vector<Cost> logicalTrafficFlits(9, 0);
-  logicalTrafficFlits[2] = 100;
-  assert(mapLogicalProcessorsToPhysicalCores(logicalTrafficFlits, line) == std::vector<size_t>({1, 0, 2}));
-
-  SchedulingTarget alreadyPlaced = line;
-  alreadyPlaced.interProcessorLatencyNs = {
-    0,
-    10,
-    1,
-    10,
-    0,
-    1,
-    1,
-    1,
-    0,
-  };
-  assert(mapLogicalProcessorsToPhysicalCores(logicalTrafficFlits, alreadyPlaced) == std::vector<size_t>({0, 1, 2}));
+  testPhysicalPeft();
 
   ComputeGraph graph;
   graph.successors.resize(6);
@@ -71,16 +213,16 @@ int main() {
     graph.instanceToIndex[instance] = task;
   }
 
-  MergeScheduleResult pipelineSchedule;
-  pipelineSchedule.processorCount = 2;
-  pipelineSchedule.dominanceOrderCompute.reserve(graph.nodes.size());
+  MergeScheduleResult physicalSchedule;
+  physicalSchedule.processorCount = 4;
+  physicalSchedule.dominanceOrderCompute.reserve(graph.nodes.size());
   for (size_t task = 0; task < graph.nodes.size(); ++task) {
     const ComputeInstance& instance = graph.nodes[task].instance;
-    pipelineSchedule.dominanceOrderCompute.push_back(instance);
+    physicalSchedule.dominanceOrderCompute.push_back(instance);
     size_t cpu = task < 4 ? 0 : 1;
-    pipelineSchedule.computeToCpuMap[instance] = cpu;
-    pipelineSchedule.computeToCpuSlotMap[instance] = task < 4 ? task : task - 4;
-    pipelineSchedule.computeToAestMap[instance] = task;
+    physicalSchedule.computeToCpuMap[instance] = cpu;
+    physicalSchedule.computeToCpuSlotMap[instance] = task < 4 ? task : task - 4;
+    physicalSchedule.computeToAestMap[instance] = task;
   }
 
   SchedulingTarget physical = fast;
@@ -92,21 +234,90 @@ int main() {
     3, 3, 0, 3,
     3, 3, 3, 0,
   };
+  SchedulingTarget clustered = physical;
+  clustered.interProcessorLatencyNs = {
+    0, 20, 1, 20,
+    20, 0, 20, 1,
+    1, 20, 0, 20,
+    20, 1, 20, 0,
+  };
+  PipelineCoreLayout clusteredLayout(std::vector<size_t> {2, 2}, clustered);
+  assert(clusteredLayout.getPhysicalCore(0) == 0);
+  assert(clusteredLayout.getPhysicalCore(1) == 2);
+  assert(clusteredLayout.getStageForCore(0) == 0);
+  assert(clusteredLayout.getStageForCore(2) == 0);
+  assert(clusteredLayout.getStageForCore(1) == 1);
+  assert(clusteredLayout.getStageForCore(3) == 1);
+  PipelineCoreLayout uniformLayout(std::vector<size_t> {2, 2}, physical);
+  assert(uniformLayout.getPhysicalCore(1) == 1);
   std::string pipelineError;
+  ComputeGraph preparationGraph = graph;
+  ResidentWeight extraWeight;
+  extraWeight.opaqueLane = graph.nodes.size();
+  preparationGraph.nodes[2].residentWeights.push_back(extraWeight);
+  auto preparation = preparePipelineWorkload(
+    preparationGraph, 2, physical, pipelineError);
+  assert(mlir::succeeded(preparation));
+  assert(*preparation == PipelineWorkloadPreparation::Ready);
+  extraWeight.opaqueLane++;
+  preparationGraph.nodes[2].residentWeights.push_back(extraWeight);
+  assert(mlir::failed(preparePipelineWorkload(
+    preparationGraph, 2, physical, pipelineError)));
+  assert(pipelineError.find("physical core's crossbars") != std::string::npos);
+  pipelineError.clear();
+  ComputeGraph emptyGraph;
+  MergeScheduleResult emptySchedule;
+  emptySchedule.processorCount = 4;
+  assert(mlir::succeeded(applyPipelineScheduling(
+    emptyGraph, emptySchedule, 2, physical, pipelineError)));
+  assert(emptySchedule.processorCount == physical.processorCount);
+  assert(emptySchedule.processorStages == std::vector<size_t>({0, 0, 1, 1}));
+
+  MergeScheduleResult pipelineSchedule = physicalSchedule;
   assert(mlir::succeeded(applyPipelineScheduling(
     graph, pipelineSchedule, 2, physical, pipelineError)));
   assert(pipelineSchedule.processorCount == 4);
-  assert(pipelineSchedule.computeToCpuMap.lookup(graph.nodes[0].instance) == 0);
-  assert(pipelineSchedule.computeToCpuMap.lookup(graph.nodes[1].instance) == 0);
-  assert(pipelineSchedule.computeToCpuMap.lookup(graph.nodes[2].instance) == 2);
-  assert(pipelineSchedule.computeToCpuMap.lookup(graph.nodes[3].instance) == 2);
-  assert(pipelineSchedule.computeToCpuMap.lookup(graph.nodes[4].instance) == 1);
-  assert(pipelineSchedule.computeToCpuMap.lookup(graph.nodes[5].instance) == 3);
+  size_t predecessorCore = pipelineSchedule.computeToCpuMap.lookup(
+    graph.nodes[1].instance);
+  size_t successorCore = pipelineSchedule.computeToCpuMap.lookup(
+    graph.nodes[2].instance);
   assert(pipelineSchedule.computeToAestMap.lookup(graph.nodes[2].instance)
          >= pipelineSchedule.computeToAestMap.lookup(graph.nodes[1].instance)
-              + graph.nodes[1].cost + 4);
+              + graph.nodes[1].cost
+              + getPeftTransferTime(
+                  TransferCost {.fixed = 1, .networkFlits = 1},
+                  predecessorCore, successorCore, physical));
+  assert(pipelineSchedule.processorStages[predecessorCore]
+         <= pipelineSchedule.processorStages[successorCore]);
+  assert(pipelineSchedule.processorStages[successorCore]
+         <= pipelineSchedule.processorStages[predecessorCore] + 1);
   assert(pipelineSchedule.equivalentClass.empty());
 
+  graph.successors[0].push_back(
+    {5, TransferCost {.fixed = 1, .networkFlits = 1}});
+  graph.predecessors[5].push_back(
+    {0, TransferCost {.fixed = 1, .networkFlits = 1}});
+  SchedulingTarget fourStagePhysical = physical;
+  fourStagePhysical.processorCount = 8;
+  fourStagePhysical.interProcessorLatencyNs.assign(64, 3);
+  for (size_t core = 0; core < 8; ++core)
+    fourStagePhysical.interProcessorLatencyNs[core * 8 + core] = 0;
+  MergeScheduleResult fourStageSchedule = physicalSchedule;
+  fourStageSchedule.processorCount = 8;
+  assert(mlir::succeeded(applyPipelineScheduling(
+    graph, fourStageSchedule, 4, fourStagePhysical, pipelineError)));
+  for (size_t task = 0; task < graph.nodes.size(); ++task)
+    for (const auto &[predecessor, cost] : graph.predecessors[task]) {
+      (void)cost;
+      size_t sourceStage = fourStageSchedule.processorStages[
+        fourStageSchedule.computeToCpuMap.lookup(
+          graph.nodes[predecessor].instance)];
+      size_t targetStage = fourStageSchedule.processorStages[
+        fourStageSchedule.computeToCpuMap.lookup(graph.nodes[task].instance)];
+      assert(sourceStage <= targetStage);
+      assert(targetStage <= sourceStage + 1);
+    }
+
   ComputeGraph communicationGraph;
   communicationGraph.successors.resize(5);
   communicationGraph.predecessors.resize(5);
@@ -116,7 +327,7 @@ int main() {
     {4, TransferCost {.fixed = 0, .networkFlits = 1}});
   const Cost communicationCosts[] = {6, 4, 6, 4, 1};
   MergeScheduleResult communicationSchedule;
-  communicationSchedule.processorCount = 2;
+  communicationSchedule.processorCount = 4;
   for (uint32_t task = 0; task < 5; ++task) {
     ComputeInstance instance {nullptr, task, 1};
     ResidentWeight weight;
@@ -136,15 +347,22 @@ int main() {
   MergeScheduleResult fastCommunicationSchedule = communicationSchedule;
   assert(mlir::succeeded(applyPipelineScheduling(
     communicationGraph, fastCommunicationSchedule, 2, fastPipeline, pipelineError)));
-  assert(fastCommunicationSchedule.computeToCpuMap.lookup(
-           communicationGraph.nodes[2].instance) == 2);
 
   SchedulingTarget slowPipeline = fastPipeline;
   slowPipeline.averageInterProcessorLatencyNs = 10;
+  slowPipeline.interProcessorLatencyNs.assign(16, 10);
+  for (size_t core = 0; core < 4; ++core)
+    slowPipeline.interProcessorLatencyNs[core * 4 + core] = 0;
   MergeScheduleResult slowCommunicationSchedule = communicationSchedule;
   assert(mlir::succeeded(applyPipelineScheduling(
     communicationGraph, slowCommunicationSchedule, 2, slowPipeline, pipelineError)));
-  assert(slowCommunicationSchedule.computeToCpuMap.lookup(
-           communicationGraph.nodes[2].instance) < 2);
+  size_t sourceCore = slowCommunicationSchedule.computeToCpuMap.lookup(
+    communicationGraph.nodes[4].instance);
+  size_t targetCore = slowCommunicationSchedule.computeToCpuMap.lookup(
+    communicationGraph.nodes[3].instance);
+  assert(slowCommunicationSchedule.processorStages[sourceCore]
+         <= slowCommunicationSchedule.processorStages[targetCore]);
+  assert(slowCommunicationSchedule.processorStages[targetCore]
+         <= slowCommunicationSchedule.processorStages[sourceCore] + 1);
   return EXIT_SUCCESS;
 }

tools/analyze_spatial_ir_cardinality.py

@@ -238,7 +238,7 @@ def print_report(path: Path, counts: Counter, groups: dict[tuple[str, str], Chai
 
 
 def main() -> None:
-    parser = argparse.ArgumentParser(description="Analyze repeated Spatial/PIM tensor IR cardinality patterns.")
+    parser = argparse.ArgumentParser(description="Analyze repeated Spatial/Pim tensor IR cardinality patterns.")
     parser.add_argument("paths", nargs="+", help="MLIR files to analyze.")
     parser.add_argument("--limit", type=int, default=12, help="Maximum number of hot chains to print per file.")
     args = parser.parse_args()

validation/.gitignore

@@ -1,22 +1,12 @@
-operations/**/inputs
-operations/**/outputs
-operations/**/raptor
-operations/**/runner
-operations/**/simulation
+operations/**/artifacts
 operations/**/*.csv
 !operations/validation_results.csv
 
-networks/**/inputs
-networks/**/outputs
-networks/**/raptor
-networks/**/pimcomp
-networks/**/runner
-networks/**/simulation
-networks/**/real_image_val
+networks/**/artifacts
 networks/**/*.png
 networks/**/*.jpg
 networks/**/*.csv
 !networks/validation_results.csv
 !networks/full_net/validation_results.csv
-!networks/pimcomp_models/validation_results.csv
-!networks/pimcomp_models/results.csv
+!networks/pimcomp_models/results_comparison.csv
+!networks/pimcomp_models/results_ablation.csv

validation/README.md

@@ -1,14 +1,14 @@
-# Raptor Validation
+# Raptor validation
 
 `validate.py` validates every ONNX model below a selected directory. For each
 model it can:
 
 1. compile an ONNX-MLIR reference library and runner;
 2. generate deterministic random inputs;
-3. compile PIM artifacts with Raptor;
-4. run the reference implementation and functional PIM simulator;
+3. compile Pim artifacts with Raptor;
+4. run the reference implementation and functional Pim simulator;
 5. compare their outputs;
-6. run `pimsim-nn` to report latency, throughput, power, and energy.
+6. run Pimsim to report latency, throughput, power, and energy.
 
 Run the script from the repository root with the repository Python environment.
 
@@ -61,29 +61,12 @@ Validate a network or network slice:
 The script discovers them recursively and writes `validation_results.csv` in
 that directory while retaining separate latency and throughput terminal tables.
 
-## Raptor vs PIMCOMP comparison
+## Pim validation tools
 
-The PIMCOMP paper-model suite has a one-command multi-architecture comparison:
-
-```bash
-.venv/bin/python validation/tools/pim/pimcomp/compare/run_pimcomp_paper_latency.py
-```
-
-The runner verifies PIMCOMP's population-200, 1000-iteration GA settings,
-expects Raptor and the existing `third_party/PIMCOMP-NN/build` tree to already
-be built, and compares the configured paper models in parallel. Set `--jobs` to
-control comparison workers;
-the runner leaves `OMP_NUM_THREADS` at its environment default. Use
-`--models vgg8` for one model or `--dry-run` to print the commands.
-
-Generated artifacts are stored beside each model and ignored by Git. The
-comparison reuses model-level `common/inputs/`, `common/outputs/`,
-`common/runner/` across architectures, modes, and pipelines. Pimsim configs and
-network meshes are referenced directly from
-`validation/pimsim_configs/pimcomp/`. Architecture- and pipeline-specific `raptor/`, `simulation/`, and
-PIMCOMP artifacts stay under each comparison directory. See
-[`networks/pimcomp_models/README.md`](networks/pimcomp_models/README.md) for
-profiles, model provenance, limitations, and remote execution.
+- [Pimcomp model suite](networks/pimcomp_models/README.md)
+- [Pimcomp model comparison tools](tools/pim/pimcomp/compare/README.md)
+- [Pimcomp correctness study](tools/pim/pimcomp/correctness/README.md)
+- [Raptor compiler ablation study](tools/pim/ablation/README.md)
 
 ## Validation modes
 
@@ -92,7 +75,7 @@ uses one input, while throughput uses `--pipeline=4` with four distinct inputs.
 Both modes reuse the generated input batch, native runner, and reference
 outputs, and every throughput output is compared with its own reference.
 
-Use `--compile-only` to build the reference runner and PIM artifacts without
+Use `--compile-only` to build the reference runner and Pim artifacts without
 executing either implementation:
 
 ```bash
@@ -139,23 +122,23 @@ count with `-j` or `--jobs`:
 | `--onnx-include-dir PATH` | ONNX-MLIR runtime include directory. Required unless `--clean` is used. |
 | `--operations-dir PATH` | Directory tree containing models. Defaults to `validation/operations`. |
 | `--simulator-dir PATH` | Functional `pim-simulator` crate directory. Defaults to the in-tree simulator. |
-| `--non-functional-simulator-build-dir PATH` | `pimsim-nn` build directory. Defaults to the in-tree build. |
+| `--non-functional-simulator-build-dir PATH` | Pimsim build directory. Defaults to the in-tree build. |
 | `--pimcomp-config {arch-a,arch-b,arch-c}` | Non-functional hardware/timing profile. Defaults to `arch-a`. |
-| `--skip-non-functional-simulation` | Skip `pimsim-nn` latency, throughput, power, and energy measurement. |
-| `--no-fast` | Disable fast throughput convergence for authoritative full-duration `pimsim-nn` experiments. |
+| `--skip-non-functional-simulation` | Skip Pimsim latency, throughput, power, and energy measurement. |
+| `--no-fast` | Disable fast throughput convergence for authoritative full-duration Pimsim experiments. |
 | `--threshold FLOAT` | Absolute output-comparison tolerance. Defaults to `1e-3`. |
 | `--relative-threshold FLOAT` | Relative output-comparison tolerance. Defaults to `1e-5`. |
 | `--seed INT` | Seed for generated inputs. Defaults to `0`. |
 | `--crossbar-size INT` | Crossbar dimensions passed to Raptor. Defaults to the Arch-A value, `128`. |
 | `--crossbar-count INT` | Crossbars per core passed to Raptor. Defaults to the Arch-A value, `96`. |
-| `--core-count INT` | PIM core count passed to Raptor. Defaults to the Arch-A value, `168`. |
+| `--core-count INT` | Pim core count passed to Raptor. Defaults to the Arch-A value, `168`. |
 | `--raptor-extra-arg=ARG` | Additional Raptor compiler argument. Repeat for multiple arguments. |
 | `--command-timeout-seconds FLOAT` | Timeout for each compiler, runner, and simulator subprocess. Defaults to `1000000.0`. |
 | `-j INT`, `--jobs INT` | Parallel validation workers. Defaults to all available CPUs and must be at least one. |
 | `--clean` | Remove generated validation artifacts and exit. |
-| `--compile-only` | Compile reference and PIM artifacts without execution or comparison. |
+| `--compile-only` | Compile reference and Pim artifacts without execution or comparison. |
 | `--run-only` | Reuse compiled artifacts and perform execution, simulation, and comparison. |
-| `--verbose` | Print passing per-stage and subprocess logs, plus average PIM pass timings. |
+| `--verbose` | Print passing per-stage and subprocess logs, plus average Pim pass timings. |
 
 Arguments beginning with `--` that are passed through to Raptor should use the
 equals form:
@@ -166,7 +149,7 @@ equals form:
 
 ## Hardware profiles and non-functional simulation
 
-The selected PIMCOMP profile must match `--core-count`, `--crossbar-count`, and
+The selected Pimcomp profile must match `--core-count`, `--crossbar-count`, and
 `--crossbar-size`. A mismatch disables only non-functional simulation and
 prints the incompatible values; functional validation still runs.
 
@@ -189,27 +172,30 @@ comparison. A non-functional simulation failure remains visible as `ERROR` in
 the corresponding latency, throughput, power, or energy columns but does not
 change a functional PASS.
 
-`pimsim-nn` does not currently implement the `vsoftmax` instruction. When its
+Pimsim does not currently implement the `vsoftmax` instruction. When its
 explicit unsupported-op diagnostic is encountered, Softmax validations retain
 their functional PASS and show `UNSUPPORTED` in the non-functional columns.
-Other `pimsim-nn` failures remain `ERROR`.
+Other Pimsim failures remain `ERROR`.
 
 ## Generated artifacts
 
-Artifacts are written beside each model:
+Generated files are grouped below an `artifacts/` directory beside each model
+or operation case. This keeps checked-in ONNX files and generated trees
+separate and lets `--clean` remove the complete workspace, including stale
+validation lock files:
 
 | Path | Contents |
 |---|---|
-| `inputs.csv` | Generated inputs, one batch entry per line. |
-| `inputs/`, `outputs/`, `runner/` | Inputs, reference outputs, and the runner shared by latency and throughput validation. |
-| `raptor/pim/`, `simulation/latency/` | Latency PIM artifacts and functional simulator outputs. |
-| `raptor/throughput/pim/`, `simulation/throughput/` | Pipeline-4, batch-4 throughput PIM artifacts and functional simulator outputs. |
-| `common/inputs/` | Shared generated input CSV files. |
-| `common/outputs/` | Shared ONNX-MLIR reference output CSV files. |
-| `common/runner/` | Shared reference runner source, build tree, and library. |
-| `<arch>/<mode>[/pipelineN]/raptor/` | Architecture- and pipeline-specific Raptor MLIR and PIM artifacts. |
-| `<arch>/<mode>[/pipelineN]/simulation/` | Functional simulator outputs for that comparison. |
-| `<arch>/<mode>[/pipelineN]/pimcomp/` | PIMCOMP graph, instruction, simulator, and comparison-report artifacts. |
+| `artifacts/inputs.csv` | Generated inputs, one batch entry per line. |
+| `artifacts/inputs/`, `artifacts/outputs/`, `artifacts/runner/` | Inputs, reference outputs, and the runner shared by latency and throughput validation. |
+| `artifacts/raptor/pim/`, `artifacts/simulation/latency/` | Latency Pim artifacts and functional simulator outputs. |
+| `artifacts/raptor/throughput/pim/`, `artifacts/simulation/throughput/` | Pipeline-4, batch-4 throughput Pim artifacts and functional simulator outputs. |
+| `artifacts/common/inputs/` | Shared generated input CSV files. |
+| `artifacts/common/outputs/` | Shared ONNX-MLIR reference output CSV files. |
+| `artifacts/common/runner/` | Shared reference runner source, build tree, and library. |
+| `artifacts/<arch>/<mode>[/pipelineN]/raptor/` | Architecture- and pipeline-specific Raptor MLIR and Pim artifacts. |
+| `artifacts/<arch>/<mode>[/pipelineN]/simulation/` | Functional simulator outputs for that comparison. |
+| `artifacts/<arch>/<mode>[/pipelineN]/pimcomp/` | Pimcomp graph, instruction, simulator, and comparison-report artifacts. |
 
 Each comparison's `raptor/` directory may include `spatial0.mlir`,
 `spatial1_graph.mlir`, `spatial2_trivial_merged.mlir`,
@@ -240,22 +226,23 @@ The generated operation inventory is documented in
 
 ## Manual functional simulator tracing
 
-After validation has produced a `raptor/pim/` directory, rerun the functional
+After validation has produced an `artifacts/raptor/pim/` directory, rerun the functional
 simulator with tracing from its crate directory:
 
 ```bash
 cd backend-simulators/pim/pim-simulator
 cargo run --no-default-features --features tracing --release \
   --package pim-simulator --bin pim-simulator -- \
-  -f /path/to/workspace/raptor/pim \
-  -o /path/to/workspace/simulation/out.bin \
+  -f /path/to/workspace/artifacts/raptor/pim \
+  -o /path/to/workspace/artifacts/simulation/out.bin \
   -d <addr0>,<size0>,<addr1>,<size1>,... \
   --mode latency \
-  --input /path/to/workspace/simulation/inputs/input_0.bin
+  --batch-size 1 \
+  --input-dir /path/to/workspace/artifacts/simulation/inputs
 ```
 
-Throughput mode additionally requires `--batch-size N` and exactly `N`
-`--input` arguments. Each input binary concatenates the model tensors in graph
+Throughput mode additionally requires `--batch-size N` and at least `N`
+`input_<index>.bin` files in `--input-dir`. Each input binary concatenates the model tensors in graph
 input order. The comparison validator also writes one native reference and one
 `simulation/*_iterations/output_*.bin` dump per batch entry, and checks every
 entry rather than only the final output.
@@ -268,7 +255,7 @@ validator normally derives the `-d` address and byte ranges from
 
 The final table reports latency and throughput functional pass/fail state plus
 non-functional latency, throughput, power, and energy. The summary includes pass/fail totals, non-functional simulation
-counts, total measured latency, and average PIM pass timings when `--verbose`
+counts, total measured latency, and average Pim pass timings when `--verbose`
 is enabled.
 
 - Exit status `0`: all discovered models passed, or cleanup completed.

validation/networks/pimcomp_models/README.md

@@ -1,38 +1,38 @@
-# PIMCOMP comparison models
+# Pimcomp comparison models
 
 This directory contains the four networks evaluated in
-[PIMCOMP: An End-to-End DNN Compiler for Processing-In-Memory Accelerators](https://arxiv.org/pdf/2411.09159):
+[Pimcomp: An End-to-End DNN Compiler for Processing-In-Memory Accelerators](https://arxiv.org/pdf/2411.09159):
 VGG-8, ResNet-18, ResNet-34, and GoogLeNet. It also contains YOLO11n as an
 additional compiler comparison model.
 
-See the runner-generated [results.csv](results.csv) for the current comparison
+See the runner-generated [results_comparison.csv](results_comparison.csv) for the current comparison
 results. Rows are retained separately for each model, architecture, mode, and
 pipeline. It records separate `PASS`/`FAIL` functional-validation fields for
-the Raptor and PIMCOMP artifacts; rows without a generated report contain `NA`.
-Running the runner with `--arch arch-b` or `--arch arch-c` appends those
-architecture rows without replacing the existing `arch-a` entries.
+the Raptor and Pimcomp artifacts; rows without a generated report contain `NA`.
+Use `--archs` to select the architecture rows to generate; existing rows for
+other architectures remain unchanged.
 
 ## Models and provenance
 
 | Directory    | Model                | Input         | Provenance                                                                                                                                                                                                 |
 |--------------|----------------------|---------------|------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
 | `resnet18/`  | ResNet-18 v1         | `1x3x224x224` | Symlink to the complete [ONNX Model Zoo `resnet18-v1-7`](https://huggingface.co/onnxmodelzoo/resnet18-v1-7) model already present at `../resnet18/depth_68/resnet18_depth_68.onnx`.                        |
-| `resnet34/`  | ResNet-34 v1         | `1x3x224x224` | [ONNX Model Zoo `resnet34-v1-7`](https://huggingface.co/onnxmodelzoo/resnet34-v1-7), with its symbolic batch fixed to 1 as PIMCOMP's frontend does.                                                        |
+| `resnet34/`  | ResNet-34 v1         | `1x3x224x224` | [ONNX Model Zoo `resnet34-v1-7`](https://huggingface.co/onnxmodelzoo/resnet34-v1-7), with its symbolic batch fixed to 1 as Pimcomp's frontend does.                                                        |
 | `googlenet/` | GoogLeNet            | `1x3x224x224` | Unmodified [ONNX Model Zoo `googlenet-12`](https://huggingface.co/onnxmodelzoo/googlenet-12).                                                                                                              |
-| `vgg8/`      | VGG-8 reconstruction | `1x1x28x28`   | Reconstruction of the [PIMCOMP VGG-8 benchmark](https://arxiv.org/html/2411.09159#S8.SS1), with six convolution and two fully connected layers.                                                            |
+| `vgg8/`      | VGG-8 reconstruction | `1x1x28x28`   | Reconstruction of the [Pimcomp VGG-8 benchmark](https://arxiv.org/html/2411.09159#S8.SS1), with six convolution and two fully connected layers.                                                            |
 | `yolo11n/`   | YOLO11n detection    | `1x3x640x640` | Derived from the canonical local model at `../yolo11n/depth_51/yolo11n_depth_51.onnx`, exported from [Ultralytics YOLO11n](https://github.com/ultralytics/ultralytics/blob/main/docs/en/models/yolo11.md). |
 
-`googlenet/googlenet-12-pimsim-nn.onnx` is the explicit pimsim-nn-ready GoogLeNet model.
+`googlenet/googlenet-12-pimsim-nn.onnx` is the explicit Pimsim-ready GoogLeNet model.
 It removes the two LRN nodes and terminal Softmax from the original model,
-so that the comparison covers only operations scheduled by PIMCOMP and supported by pimsim-nn.
+so that the comparison covers only operations scheduled by Pimcomp and supported by Pimsim.
 
-`yolo11n/yolo11n-pimsim-nn.onnx` is the explicit pimsim-nn-ready YOLO11n model.
+`yolo11n/yolo11n-pimsim-nn.onnx` is the explicit Pimsim-ready YOLO11n model.
 It removes the Softmax nodes from the original model,
-so that the compiled artifact can be simulated in pimsim-nn.
+so that the compiled artifact can be simulated in Pimsim.
 
 ## Unsupported and ignored operations
 
-PIMCOMP's frontend accepts exactly these ONNX operations:
+Pimcomp's frontend accepts exactly these ONNX operations:
 
 ```text
 Add, AveragePool, BatchNormalization, Clip, Concat, Conv, Dropout, Flatten,
@@ -40,14 +40,14 @@ Gather, Gemm, GlobalAveragePool, LRN, MatMul, MaxPool, Mul, Pad, Relu, Reshape,
 Shape, Sigmoid, Softmax, Squeeze, Sub, Sum, Tanh, Transpose, Unsqueeze
 ```
 
-`Constant` is consumed as frontend metadata rather than emitted as a PIMCOMP
+`Constant` is consumed as frontend metadata rather than emitted as a Pimcomp
 node. Every other ONNX operation is unsupported: the frontend prints
 `operation: <type> not considered` and stops at the first occurrence. Thus the
 complete unsupported set is the complement of the allowlist above for the
 model's ONNX opset. In particular, YOLO11n contains unsupported `Split` and
 `Resize` nodes.
 
-PIMCOMP's low-latency scheduler, hierarchy mapper, and genetic algorithm use
+Pimcomp's low-latency scheduler, hierarchy mapper, and genetic algorithm use
 this complete explicit no-consider set:
 
 ```text
@@ -63,7 +63,7 @@ specific Shape-Gather-Unsqueeze-Concat shape chain, and merge Pad into its
 consumer. These transformations do not make an otherwise standalone ignored
 operation timed.
 
-`pimsim-nn` consumes PIM ISA instructions. It supports every named opcode
+Pimsim consumes Pim ISA instructions. It supports every named opcode
 in the shared serialized range except `vsoftmax` (opcode 21),
 which is rejected explicitly in both JSON and binary input. It
 silently ignores no opcode; unknown names and numbers are errors.
@@ -71,12 +71,12 @@ silently ignores no opcode; unknown names and numbers are errors.
 These boundaries explain the dedicated artifacts:
 
 - GoogLeNet's two LRN nodes and terminal Softmax perform real computation but
-  are ignored by PIMCOMP, so the common latency artifact removes them. Its
+  are ignored by Pimcomp, so the common latency artifact removes them. Its
   inference Dropout and shape-only Reshape can remain without adding compute.
 - YOLO11n's latency artifact bypasses exactly its two Softmax nodes so it can
-  run in `pimsim-nn`. Every other node, including MatMul, Transpose, and the
+  run in Pimsim. Every other node, including MatMul, Transpose, and the
   final detection-decoding tail, remains present and timed by Raptor. No
-  PIMCOMP latency is reported because its frontend stops at `Split` and also
+  Pimcomp latency is reported because its frontend stops at `Split` and also
   lacks `Resize`; compiling that prefix would not represent YOLO11n.
 
 The authoritative lists are in
@@ -85,16 +85,16 @@ The authoritative lists are in
 [`ISA.h`](../../../backend-simulators/pim/pimsim-nn/src/isa/ISA.h), and
 [`Instruction.cpp`](../../../backend-simulators/pim/pimsim-nn/src/isa/Instruction.cpp).
 
-The PIMCOMP authors did not publish the ONNX checkpoints used by the paper.
-Running PIMCOMP's frontend on the three Model Zoo files above produces JSON
-graphs exactly equal to PIMCOMP-NN's bundled `resnet18.json`, `resnet34.json`,
+The Pimcomp authors did not publish the ONNX checkpoints used by the paper.
+Running Pimcomp's frontend on the three Model Zoo files above produces JSON
+graphs exactly equal to Pimcomp's bundled `resnet18.json`, `resnet34.json`,
 and `googlenet.json`.
 
-There is no VGG-8 artifact in the ONNX Model Zoo or any PIMCOMP-NN revision.
+There is no VGG-8 artifact in the ONNX Model Zoo or any Pimcomp revision.
 The included VGG-8 therefore has deterministic random weights and is suitable
 for compiler and simulator comparison, not paper-accuracy reproduction. The
 paper also says that VGG-8 and ResNet-18 were trained on MNIST, while the
-published PIMCOMP graphs and ResNet Model Zoo artifacts use ImageNet shapes.
+published Pimcomp graphs and ResNet Model Zoo artifacts use ImageNet shapes.
 
 Current SHA-256 checksums:
 
@@ -113,10 +113,10 @@ The files in
 [`../../pimsim_configs/pimcomp/`](../../pimsim_configs/pimcomp/)
 encode Table V's explicit resource parameters.
 Each profile subdirectory contains pre-generated latency and throughput
-`pimsim-nn` configs plus its matching mesh; comparison and validation reference
+Pimsim configs plus its matching mesh; comparison and validation reference
 these canonical artifacts directly.
 
-| Config                       |             Cores | Crossbars/core | Crossbar   |  Cell | PIMCOMP layout  |
+| Config                       |             Cores | Crossbars/core | Crossbar   |  Cell | Pimcomp layout  |
 |------------------------------|------------------:|---------------:|------------|------:|-----------------|
 | `arch-a/latency_config.json` |               168 |             96 | `128x128`  | 2-bit | `12x14`         |
 | `arch-b/latency_config.json` |               138 |            128 | `128x128`  | 2-bit | `6x23`          |
@@ -125,10 +125,10 @@ these canonical artifacts directly.
 `adc_count` is 16, matching the paper's 16-bit fixed-point weight precision.
 The paper does not give a two-dimensional core topology for Arch-A/B, so the
 factorizations above preserve core count but cannot reproduce unpublished NoC
-placement details. Released PIMCOMP-NN has no chip-count field; Arch-C is
+placement details. Released Pimcomp has no chip-count field; Arch-C is
 therefore flattened to 64 cores and does not model chip boundaries.
 
-The remaining latency and power values come from PIMCOMP-NN's released default
+The remaining latency and power values come from Pimcomp's released default
 configuration. Consequently, instruction/resource comparisons are
 reproducible, but absolute paper power and energy numbers are not.
 
@@ -147,9 +147,9 @@ cmake --build third_party/PIMCOMP-NN/build --target PIMCOMP-NN
 
 Do not build either project with `ninja` directly.
 
-## Compile with PIMCOMP
+## Compile with Pimcomp
 
-PIMCOMP-NN reads `third_party/PIMCOMP-NN/config.json` directly. Back it up,
+Pimcomp reads `third_party/PIMCOMP-NN/config.json` directly. Back it up,
 select one paper profile, and restore it when the shell exits:
 
 ```bash
@@ -162,7 +162,7 @@ trap 'cp "$CONFIG_BACKUP" "$PIMCOMP/config.json"' EXIT
 cp "$PIMCOMP_CONFIGS/arch-a/latency_config.json" "$PIMCOMP/config.json"
 ```
 
-The Model Zoo files map exactly to PIMCOMP's bundled model names, so compile
+The Model Zoo files map exactly to Pimcomp's bundled model names, so compile
 them directly:
 
 ```bash
@@ -179,7 +179,7 @@ cd "$PIMCOMP/build"
 ./PIMCOMP-NN -m=googlenet -r=balance -p=element -o=YES -v=YES -s=YES
 ```
 
-VGG-8 first needs PIMCOMP's JSON frontend. Use a temporary ONNX copy because
+VGG-8 first needs Pimcomp's JSON frontend. Use a temporary ONNX copy because
 the released frontend rewrites the input batch dimension in place:
 
 ```bash
@@ -201,62 +201,66 @@ random placement code occasionally segfaults; an unchanged retry succeeded in
 the observed cases.
 
 The paper's optimizer uses a genetic algorithm with population 200 and up to
-1000 iterations. The checked-out PIMCOMP submodule already has both paper
+1000 iterations. The checked-out Pimcomp submodule already has both paper
 settings in `backend/GeneticAlgorithm.h`; select them with `-r=GA`. Fitness
 evaluation uses OpenMP and bounded bandwidth timelines. Set `OMP_NUM_THREADS`
 to control its parallelism; otherwise OpenMP uses the available CPUs. The GA
 uses the fixed seed `1`, so repeated serial and parallel runs are reproducible.
 
-## Compare Raptor and PIMCOMP
+## Compare Raptor and Pimcomp
 
 The comparison driver uses one random input and one native ONNX-MLIR reference,
-compiles both instruction streams, runs both through `pimsim-nn`, runs
+compiles both instruction streams, runs both through Pimsim, runs
 functional validation through `pim-simulator`, and writes Markdown and JSON
 reports.
 
-To reproduce all configured architectures and both latency/throughput modes,
-use the model-by-model runner. It verifies the paper GA settings, expects Raptor
+To reproduce the default `arch-a`/`arch-b` architectures and both
+latency/throughput modes, use the model-by-model runner. Use `--archs` to
+specify a different architecture set. It verifies the paper GA settings, expects Raptor
 and the existing `third_party/PIMCOMP-NN/build` tree to already be built, then
-runs the comparisons in parallel and regenerates `results.csv` from the JSON
+runs the comparisons in parallel and regenerates `results_comparison.csv` from the JSON
 reports:
 
 ```bash
-.venv/bin/python validation/tools/pim/pimcomp/compare/run_pimcomp_paper_latency.py
+.venv/bin/python validation/tools/pim/pimcomp/compare/run_pimcomp_models.py
 ```
 
-Use `--arch arch-a --mode latency` for only the Arch-A latency experiment.
+Use `--archs arch-a --mode latency` for only the Arch-A latency experiment, or
+`--archs arch-a arch-b arch-c` to run all three architectures.
 
-Each model directory has a shared ignored `common/` directory containing
+Each model directory has a shared ignored `artifacts/common/` directory containing
 `inputs/`, `outputs/`, and the native `runner/`.
-Pimsim-nn configs and network meshes remain canonical under
+Pimsim configs and network meshes remain canonical under
 `validation/pimsim_configs/pimcomp/` and are referenced in place.
-Architecture- and pipeline-specific `raptor/`, `simulation/`, and PIMCOMP
-artifacts remain under each comparison directory; PIMCOMP outputs are prepared
+Architecture- and pipeline-specific `raptor/`, `simulation/`, and Pimcomp
+artifacts remain under each `artifacts/<arch>/<mode>[/pipelineN]/` comparison
+directory; ablation variants use
+`artifacts/<arch>/<mode>[/pipelineN]/ablation/<variant>/`. Pimcomp outputs are prepared
 once per model/architecture/mode and linked into the other pipeline directories;
 `comparison_report.{md,json}`
 live under its `pimcomp/`. The frontend regenerates
-one isolated `models/JSON/` graph because PIMCOMP requires that relative
+one isolated `models/JSON/` graph because Pimcomp requires that relative
 layout; it is removed after a successful backend run and the shared submodule
 model directory is never modified. The original ONNX model is passed to the
 frontend unchanged.
-PIMCOMP's source tree and build directory remain unchanged at runtime. Use
+Pimcomp's source tree and build directory remain unchanged at runtime. Use
 `--models vgg8` to run one model, `--mode throughput` to select one mode,
 `--pipeline 4` to select one throughput pipeline, `--only raptor` or
 `--only pimcomp` to reuse the other compiler's existing artifacts, `--dry-run`
 to inspect every command, or `--out-dir PATH` to keep results outside
 `validation/`. Use `--clean` to remove generated comparison artifacts and
-summaries. Selecting a subset replaces only those comparison rows and
-recomputes the aggregate `results.csv`; missing shared inputs, outputs, or the
+summaries, including stale reference lock files. Selecting a subset replaces only those comparison rows and
+recomputes the aggregate `results_comparison.csv`; missing shared inputs, outputs, or the
 reference runner are generated even for an isolated run. Use `--jobs 4` to cap
 parallel comparisons. The per-stage timeout is unlimited by default; pass a
 positive `--timeout-seconds` value to impose one.
 Throughput comparisons default to `pimsim-nn --fast` with a 1000 ms
 convergence deadline. Add `--no-fast` for authoritative full-duration runs.
-PIMCOMP receives the original ONNX model, and its frontend applies native
+Pimcomp receives the original ONNX model, and its frontend applies native
 BatchNormalization fusion when the graph matches its supported Conv/Gemm pattern.
 The runner continues after a failed model so all reports are produced.
 
-The known PIMCOMP batch-scheduling correctness issue and a reproducible
+The known Pimcomp batch-scheduling correctness issue and a reproducible
 reference-intermediate prefill experiment are documented in
 [`validation/tools/pim/pimcomp/correctness/README.md`](../../tools/pim/pimcomp/correctness/README.md).
 
@@ -265,9 +269,9 @@ Arch-A low-latency example:
 ```bash
 RAPTOR_ROOT=$PWD
 
-"$RAPTOR_ROOT/.venv/bin/python" "$RAPTOR_ROOT/validation/tools/pim/pimcomp/compare/compare_raptor_pimcomp.py" \
+"$RAPTOR_ROOT/.venv/bin/python" "$RAPTOR_ROOT/validation/tools/pim/pimcomp/compare/compare_raptor_pimcomp_model.py" \
   --model "$RAPTOR_ROOT/validation/networks/pimcomp_models/resnet34/resnet34-v1-7.onnx" \
-  --out-dir "$RAPTOR_ROOT/validation/networks/pimcomp_models/resnet34/arch-a/latency" \
+  --out-dir "$RAPTOR_ROOT/validation/networks/pimcomp_models/resnet34/artifacts/arch-a/latency" \
   --pimcomp-config "$RAPTOR_ROOT/validation/pimsim_configs/pimcomp/arch-a/latency_config.json" \
   --core-count 168 \
   --crossbar-count 96 \
@@ -279,10 +283,10 @@ RAPTOR_ROOT=$PWD
 ```
 
 Use the same command with
-`yolo11n/yolo11n-pimsim-nn.onnx` to probe YOLO11n. Released PIMCOMP-NN cannot
+`yolo11n/yolo11n-pimsim-nn.onnx` to probe YOLO11n. Released Pimcomp cannot
 compile it: the frontend stops at `/model.2/Split`, and it also has no mapping
 for YOLO11n's two nearest-neighbor `Resize` nodes. Treating the emitted prefix
-as YOLO11n would produce a misleading latency, so no PIMCOMP number is
+as YOLO11n would produce a misleading latency, so no Pimcomp number is
 reported for this model.
 
 For Arch-A high throughput, use `--pimsim-mode throughput
@@ -303,18 +307,18 @@ generated report.
 The functional and non-functional simulators intentionally consume different
 artifacts:
 
-- Raptor and PIMCOMP are validated against the native ONNX-MLIR reference as
+- Raptor and Pimcomp are validated against the native ONNX-MLIR reference as
   FP32 programs in the Rust simulator. Raptor's emitted program is already
-  FP32. The PIMCOMP-to-Rust export expands its element-addressed storage and
+  FP32. The Pimcomp-to-Rust export expands its element-addressed storage and
   byte-sized transfers to FP32, emits `setbw 32, 32`, and keeps vector
   `imm_len` fields as element counts.
-- PIMCOMP's original `SimulationInfo.gz` is copied unchanged for `pimsim-nn`.
-  PIMCOMP hardcodes `setbw 8, 8` and one byte per element without performing
+- Pimcomp's original `SimulationInfo.gz` is copied unchanged for Pimsim.
+  Pimcomp hardcodes `setbw 8, 8` and one byte per element without performing
   numerical quantization; this artifact is used only for latency estimation.
 - Raptor's original FP32 artifact remains unchanged for functional validation.
   A separate `raptor/pimsim_nn/` view uses `setbw 8, 8` and scales its
   byte-addressed storage and transfer sizes from four bytes to one byte per
-  element. Vector `imm_len` fields remain element counts. Like PIMCOMP's
+  element. Vector `imm_len` fields remain element counts. Like Pimcomp's
   artifact, this view is not numerically valid and is used only for a fair
   non-functional comparison.
 
@@ -326,10 +330,10 @@ Current Raptor status:
 
 - VGG-8, ResNet-18, fixed-batch ResNet-34, and GoogLeNet compile on Arch-A.
 - Use `googlenet-12-pimsim-nn.onnx` for the paper-matched latency comparison.
-  It removes the two LRN nodes and terminal softmax that PIMCOMP does not
+  It removes the two LRN nodes and terminal softmax that Pimcomp does not
   schedule.
 - Raptor currently accepts one square `--crossbar-size`; Arch-C's rectangular
-  `512x1024` arrays can therefore be compiled by PIMCOMP but not compared
+  `512x1024` arrays can therefore be compiled by Pimcomp but not compared
   exactly with Raptor.
 
 Do not change the hardware profile to bypass either limitation; that would no
@@ -365,7 +369,7 @@ python3 -m venv .venv
 .venv/bin/python -m pip install numpy onnx onnxruntime colorama
 
 # Run every configured comparison in parallel.
-.venv/bin/python validation/tools/pim/pimcomp/compare/run_pimcomp_paper_latency.py
+.venv/bin/python validation/tools/pim/pimcomp/compare/run_pimcomp_models.py
 ```
 
 Copy reports back without transferring large compiler artifacts:

validation/networks/pimcomp_models/results.csv

@@ -1,41 +0,0 @@
-model,arch,mode,raptor_pipeline,pimcomp_pipeline,raptor_functional_validation,pimcomp_functional_validation,raptor_throughput_samples_s,pimcomp_throughput_samples_s,raptor_latency_ms,pimcomp_latency_ms,raptor_power_mw,pimcomp_power_mw,raptor_energy_pj,pimcomp_energy_pj,better_compiler,speedup
-vgg8,arch-a,latency,1,element,PASS,PASS,NA,NA,1.465778,7.985074,325.627854,200.111367,477298145.040001,1597904071.120000,raptor,5.45
-vgg8,arch-b,latency,1,element,PASS,PASS,NA,NA,1.438869,7.152125,304.673458,173.768633,438385194.040001,1242814988.120001,raptor,4.97
-vgg8,arch-c,latency,1,element,FAIL,PASS,NA,NA,NA,NA,NA,NA,NA,NA,NA,NA
-vgg8,arch-a,throughput,2,batch,PASS,PASS,1480.000000,1380.000000,0.674197,0.725548,456.064662,475.694103,307477506.200000,345138738.300000,raptor,1.07
-vgg8,arch-b,throughput,2,batch,PASS,PASS,1160.000000,1080.000000,0.861527,0.921878,333.045167,408.540967,286927542.600000,376624916.800000,raptor,1.07
-vgg8,arch-c,throughput,2,batch,FAIL,PASS,NA,NA,NA,NA,NA,NA,NA,NA,NA,NA
-vgg8,arch-a,throughput,4,batch,PASS,PASS,2160.000000,1380.000000,0.462342,0.725548,446.528150,475.694103,206448818.800000,345138738.300000,raptor,1.57
-vgg8,arch-c,throughput,4,batch,FAIL,PASS,NA,NA,NA,NA,NA,NA,NA,NA,NA,NA
-vgg8,arch-a,throughput,8,batch,PASS,PASS,831.000000,1380.000000,1.202894,0.725548,331.341726,475.694103,398569128.700000,345138738.300000,pimcomp,1.66
-vgg8,arch-c,throughput,8,batch,FAIL,PASS,NA,NA,NA,NA,NA,NA,NA,NA,NA,NA
-resnet18,arch-a,latency,1,element,PASS,PASS,NA,NA,28.099951,58.855175,312.513413,237.590194,8781611597.119984,13983412446.119972,raptor,2.09
-resnet18,arch-b,latency,1,element,PASS,PASS,NA,NA,34.938563,65.084209,254.439797,200.986958,8889760875.119959,13081077194.119965,raptor,1.86
-resnet18,arch-c,latency,1,element,FAIL,PASS,NA,NA,NA,NA,NA,NA,NA,NA,NA,NA
-resnet18,arch-a,throughput,2,batch,PASS,FAIL,20.000000,76.000000,50.000000,13.149606,263.148477,479.788072,13157423870.000000,6309024249.000000,pimcomp,3.80
-resnet18,arch-b,throughput,2,batch,PASS,FAIL,26.300000,83.200000,38.016529,12.020906,250.553072,483.029728,9525158116.000000,5806454916.000000,pimcomp,3.16
-resnet18,arch-c,throughput,2,batch,FAIL,FAIL,NA,NA,NA,NA,NA,NA,NA,NA,NA,NA
-resnet18,arch-a,throughput,4,batch,PASS,FAIL,31.700000,76.000000,31.578947,13.149606,336.949886,479.788072,10640522720.000000,6309024249.000000,pimcomp,2.40
-resnet18,arch-c,throughput,4,batch,FAIL,FAIL,NA,NA,NA,NA,NA,NA,NA,NA,NA,NA
-resnet18,arch-a,throughput,8,batch,PASS,FAIL,46.400000,76.000000,21.566110,13.149606,319.292391,479.788072,6885894954.000000,6309024249.000000,pimcomp,1.64
-resnet18,arch-c,throughput,8,batch,FAIL,FAIL,NA,NA,NA,NA,NA,NA,NA,NA,NA,NA
-resnet34,arch-a,latency,1,element,PASS,PASS,NA,NA,45.781484,91.608751,326.833876,248.044564,14962939889.679951,22723052668.680016,raptor,2.00
-resnet34,arch-b,latency,1,element,PASS,PASS,NA,NA,72.119607,94.519582,239.192092,215.513439,17250439680.679901,20370240175.680019,raptor,1.31
-resnet34,arch-c,latency,1,element,FAIL,PASS,NA,NA,NA,NA,NA,NA,NA,NA,NA,NA
-resnet34,arch-a,throughput,2,batch,FAIL,FAIL,NA,40.800000,NA,24.522761,NA,506.131154,NA,12411733160.000000,NA,NA
-resnet34,arch-b,throughput,2,batch,PASS,FAIL,11.600000,43.300000,86.250000,23.076923,260.049856,489.500435,22429300100.000000,11296163880.000000,pimcomp,3.73
-resnet34,arch-c,throughput,2,batch,FAIL,FAIL,NA,NA,NA,NA,NA,NA,NA,NA,NA,NA
-resnet34,arch-a,throughput,4,batch,FAIL,FAIL,NA,40.800000,NA,24.522761,NA,506.131154,NA,12411733160.000000,NA,NA
-resnet34,arch-c,throughput,4,batch,FAIL,FAIL,NA,NA,NA,NA,NA,NA,NA,NA,NA,NA
-resnet34,arch-a,throughput,8,batch,PASS,FAIL,24.300000,40.800000,41.176471,24.522761,271.814969,506.131154,11192381060.000000,12411733160.000000,pimcomp,1.68
-resnet34,arch-c,throughput,8,batch,FAIL,FAIL,NA,NA,NA,NA,NA,NA,NA,NA,NA,NA
-googlenet,arch-a,latency,1,element,PASS,PASS,NA,NA,13.032305,62.923369,465.072002,231.194088,6060960174.919998,14547510894.240002,raptor,4.83
-googlenet,arch-b,latency,1,element,PASS,PASS,NA,NA,16.086822,37.935747,378.951815,242.039077,6096130396.919978,9181933205.239973,raptor,2.36
-googlenet,arch-c,latency,1,element,FAIL,PASS,NA,NA,NA,NA,NA,NA,NA,NA,NA,NA
-googlenet,arch-a,throughput,2,batch,PASS,FAIL,55.000000,66.200000,18.181818,15.094340,346.696726,434.959232,6303576840.000000,6565422375.000000,pimcomp,1.20
-googlenet,arch-b,throughput,2,batch,PASS,FAIL,55.000000,72.000000,18.181818,13.888889,314.645370,417.992095,5720824912.000000,5805445765.000000,pimcomp,1.31
-googlenet,arch-c,throughput,2,batch,FAIL,FAIL,NA,NA,NA,NA,NA,NA,NA,NA,NA,NA
-googlenet,arch-a,throughput,4,batch,PASS,FAIL,111.000000,66.200000,9.012016,15.094340,373.195227,434.959232,3363241363.000000,6565422375.000000,raptor,1.68
-googlenet,arch-c,throughput,4,batch,FAIL,FAIL,NA,NA,NA,NA,NA,NA,NA,NA,NA,NA
-googlenet,arch-a,throughput,8,batch,PASS,FAIL,61.300000,66.200000,16.310680,15.094340,341.592923,434.959232,5571612719.000000,6565422375.000000,pimcomp,1.08
-googlenet,arch-c,throughput,8,batch,FAIL,FAIL,NA,NA,NA,NA,NA,NA,NA,NA,NA,NA

validation/networks/pimcomp_models/validation_results.csv

@@ -1,6 +0,0 @@
-Operation,Result,Compile,Host mem,Cores mem,Cores,Xbars,Latency,Power,Energy
-vgg8-mnist-reconstructed,PASS,1.009 s,1.37 MiB,3.14 MiB,141,761,1.465778 ms,325.627854 mW,477298145.040001 pJ
-resnet18-v1-7,PASS,11.548 s,9.89 MiB,40.24 MiB,168,7676,28.099952 ms,312.513408 mW,8781611766.119984 pJ
-resnet34-v1-7,PASS,28.495 s,9.90 MiB,48.89 MiB,168,15292,45.781486 ms,326.833870 mW,14962940227.679951 pJ
-googlenet-12-pimsim-nn,PASS,6.573 s,10.74 MiB,22.41 MiB,168,7176,13.371204 ms,457.538139 mW,6117835798.919991 pJ
-yolo11n-pimsim-nn,FAIL,58.572 s,82.55 MiB,185.68 MiB,168,6484,885.264931 ms,189.218985 mW,167508931321.001465 pJ

validation/operations/README.md

@@ -1,7 +1,7 @@
-# Operation Validation Suite
+# Operation validation suite
 
 This directory contains the ONNX models used by `validation/validate.py` to
-validate individual operations through compilation, PIM simulation, and
+validate individual operations through compilation, Pim simulation, and
 comparison with the ONNX-MLIR reference runtime.
 
 ## Naming

validation/operations/validation_results.csv

@@ -1,178 +0,0 @@
-Operation,Arch,Result (l),Result (t),Compile (l),Host mem (l),Cores mem (l),Cores (l),Xbars (l),Latency (l),Power (l),Energy (l),Compile (t),Host mem (t),Cores mem (t),Cores (t),Xbars (t),Avg latency (t),Throughput (t),Avg power (t),Avg energy (t)
-add/after_gemm,arch-a,PASS,PASS,0.058 s,0.01 MiB,0.01 MiB,5,4,0.01 ms,104.70 mW,815012.96 pJ,0.057 s,0.01 MiB,0.01 MiB,6,4,145000.00 samples/s,0.01 ms,31.45 mW,216167.21 pJ/it
-add/basic,arch-a,PASS,PASS,0.048 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.22 mW,25266.00 pJ,0.050 s,0.00 MiB,0.00 MiB,1,0,3120000.00 samples/s,0.00 ms,2.23 mW,658.67 pJ/it
-add/broadcast_row,arch-a,PASS,PASS,0.048 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.22 mW,25266.00 pJ,0.051 s,0.00 MiB,0.00 MiB,1,0,3120000.00 samples/s,0.00 ms,2.23 mW,658.67 pJ/it
-add/channel_broadcast_1024,arch-a,PASS,PASS,0.049 s,0.02 MiB,0.01 MiB,1,0,0.01 ms,78.12 mW,540030.00 pJ,0.051 s,0.02 MiB,0.01 MiB,1,0,145000.00 samples/s,0.01 ms,2.11 mW,13388.67 pJ/it
-add/leading_dimension_broadcast,arch-a,PASS,PASS,0.051 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.22 mW,25266.00 pJ,0.049 s,0.00 MiB,0.00 MiB,1,0,3120000.00 samples/s,0.00 ms,2.23 mW,658.67 pJ/it
-concat/channel_axis,arch-a,PASS,PASS,0.048 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.16 mW,35718.00 pJ,0.050 s,0.00 MiB,0.00 MiB,1,0,2200000.00 samples/s,0.00 ms,2.16 mW,934.67 pJ/it
-concat/negative_axis,arch-a,PASS,PASS,0.050 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.09 mW,81450.00 pJ,0.051 s,0.00 MiB,0.00 MiB,1,0,961000.00 samples/s,0.00 ms,2.09 mW,2108.00 pJ/it
-concat/three_inputs_channel_axis,arch-a,PASS,PASS,0.048 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.15 mW,50328.00 pJ,0.047 s,0.00 MiB,0.00 MiB,1,0,1560000.00 samples/s,0.00 ms,2.15 mW,1332.67 pJ/it
-conv/batch_2,arch-a,PASS,PASS,0.059 s,0.00 MiB,0.00 MiB,2,2,0.01 ms,82.62 mW,1131451.48 pJ,0.062 s,0.00 MiB,0.01 MiB,4,2,129000.00 samples/s,0.01 ms,51.25 mW,406238.48 pJ/it
-conv/batch_4_pointwise,arch-a,PASS,PASS,0.058 s,0.00 MiB,0.01 MiB,5,4,0.00 ms,116.08 mW,456420.96 pJ,0.061 s,0.00 MiB,0.01 MiB,5,4,243000.00 samples/s,0.00 ms,44.13 mW,180813.46 pJ/it
-conv/depthwise_1024_channels,arch-a,PASS,PASS,0.080 s,0.19 MiB,0.38 MiB,129,128,0.22 ms,178.45 mW,39393966.72 pJ,0.141 s,0.36 MiB,0.48 MiB,87,128,3620.00 samples/s,0.28 ms,131.43 mW,37256350.26 pJ/it
-conv/depthwise_grouped,arch-a,PASS,PASS,0.056 s,0.01 MiB,0.00 MiB,5,4,0.01 ms,107.78 mW,671878.96 pJ,0.061 s,0.01 MiB,0.00 MiB,7,4,235000.00 samples/s,0.00 ms,53.10 mW,227356.96 pJ/it
-conv/dilated_3x3,arch-a,PASS,PASS,0.061 s,0.01 MiB,0.01 MiB,10,9,0.01 ms,118.77 mW,1034819.16 pJ,0.071 s,0.01 MiB,0.01 MiB,12,9,119000.00 samples/s,0.01 ms,61.00 mW,511357.16 pJ/it
-conv/dynamic,arch-a,PASS,PASS,0.056 s,0.00 MiB,0.00 MiB,5,0,0.00 ms,92.28 mW,169336.00 pJ,0.057 s,0.00 MiB,0.00 MiB,6,0,784000.00 samples/s,0.00 ms,18.61 mW,26517.00 pJ/it
-conv/explicit_padding,arch-a,PASS,PASS,0.060 s,0.01 MiB,0.02 MiB,17,16,0.01 ms,145.34 mW,1454397.84 pJ,0.064 s,0.01 MiB,0.02 MiB,19,16,153000.00 samples/s,0.01 ms,109.61 mW,715669.59 pJ/it
-conv/grouped_many_groups,arch-a,PASS,PASS,0.498 s,0.05 MiB,0.09 MiB,65,64,0.18 ms,142.21 mW,25867112.36 pJ,0.547 s,0.11 MiB,0.79 MiB,127,64,3750.00 samples/s,0.27 ms,141.11 mW,43353235.67 pJ/it
-conv/grouped_two_groups,arch-a,PASS,PASS,0.064 s,0.00 MiB,0.00 MiB,3,2,0.01 ms,101.46 mW,543914.48 pJ,0.066 s,0.00 MiB,0.01 MiB,9,2,146000.00 samples/s,0.01 ms,108.34 mW,741101.98 pJ/it
-conv/huge_pointwise_1024,arch-a,PASS,PASS,0.166 s,0.01 MiB,0.11 MiB,73,64,0.02 ms,249.55 mW,3896647.36 pJ,0.182 s,0.04 MiB,0.11 MiB,74,64,33300.00 samples/s,0.03 ms,133.89 mW,4052259.07 pJ/it
-conv/huge_pointwise_1024_dynamic,arch-a,PASS,PASS,0.084 s,8.04 MiB,12.61 MiB,168,0,2.63 ms,169.52 mW,445489032.00 pJ,0.263 s,11.49 MiB,10.61 MiB,127,0,213.00 samples/s,4.70 ms,164.24 mW,811591564.70 pJ/it
-conv/input_224_7x7_stride2,arch-a,PASS,PASS,0.775 s,24.14 MiB,61.87 MiB,168,169,38.41 ms,185.26 mW,7116544212.12 pJ,1.142 s,46.43 MiB,73.41 MiB,126,153,27.30 samples/s,36.66 ms,177.05 mW,6915042527.00 pJ/it
-conv/kernel_2x2,arch-a,PASS,PASS,0.056 s,0.00 MiB,0.00 MiB,1,1,0.00 ms,83.83 mW,360568.24 pJ,0.055 s,0.00 MiB,0.00 MiB,3,1,334000.00 samples/s,0.00 ms,51.45 mW,171905.91 pJ/it
-conv/kernel_3x3,arch-a,PASS,PASS,0.060 s,0.01 MiB,0.01 MiB,10,9,0.01 ms,123.80 mW,889640.16 pJ,0.063 s,0.01 MiB,0.01 MiB,12,9,219000.00 samples/s,0.00 ms,83.71 mW,382318.91 pJ/it
-conv/kernel_equals_input_spatial,arch-a,PASS,PASS,0.054 s,0.00 MiB,0.00 MiB,2,2,0.00 ms,89.61 mW,415689.48 pJ,0.057 s,0.00 MiB,0.00 MiB,4,2,293000.00 samples/s,0.00 ms,59.39 mW,204713.48 pJ/it
-conv/large_input_channels_1x1,arch-a,PASS,PASS,0.096 s,0.01 MiB,0.02 MiB,9,8,0.01 ms,117.82 mW,901121.92 pJ,0.092 s,0.01 MiB,0.02 MiB,10,8,132000.00 samples/s,0.01 ms,59.24 mW,447909.92 pJ/it
-conv/large_output_channels_1x1,arch-a,PASS,PASS,0.089 s,0.01 MiB,0.02 MiB,17,8,0.01 ms,128.44 mW,1139415.92 pJ,0.095 s,0.01 MiB,0.02 MiB,18,8,123000.00 samples/s,0.01 ms,43.92 mW,355735.17 pJ/it
-conv/large_spatial,arch-a,PASS,PASS,0.059 s,0.01 MiB,0.04 MiB,37,36,0.02 ms,172.07 mW,2928344.64 pJ,0.078 s,0.01 MiB,0.04 MiB,39,36,88500.00 samples/s,0.01 ms,169.91 mW,1920027.89 pJ/it
-conv/multi_channel,arch-a,PASS,PASS,0.057 s,0.00 MiB,0.00 MiB,4,3,0.01 ms,105.68 mW,685040.72 pJ,0.060 s,0.00 MiB,0.00 MiB,4,3,146000.00 samples/s,0.01 ms,30.09 mW,205787.97 pJ/it
-conv/non_square_kernel_1x3,arch-a,PASS,PASS,0.055 s,0.00 MiB,0.00 MiB,3,2,0.01 ms,99.35 mW,679752.48 pJ,0.059 s,0.00 MiB,0.00 MiB,3,2,141000.00 samples/s,0.01 ms,12.12 mW,85739.48 pJ/it
-conv/non_square_kernel_3x1,arch-a,PASS,PASS,0.058 s,0.00 MiB,0.00 MiB,3,2,0.01 ms,95.89 mW,1292976.48 pJ,0.061 s,0.00 MiB,0.00 MiB,3,2,72900.00 samples/s,0.01 ms,8.83 mW,121109.48 pJ/it
-conv/non_uniform_stride,arch-a,PASS,PASS,0.061 s,0.00 MiB,0.00 MiB,4,3,0.01 ms,104.05 mW,790874.72 pJ,0.059 s,0.00 MiB,0.00 MiB,4,3,131000.00 samples/s,0.01 ms,29.05 mW,221084.97 pJ/it
-conv/output_channel_grouping_minimal,arch-a,PASS,PASS,0.089 s,0.10 MiB,0.34 MiB,131,128,0.26 ms,170.73 mW,44125916.72 pJ,0.181 s,0.18 MiB,0.33 MiB,131,128,3910.00 samples/s,0.26 ms,181.50 mW,48146979.72 pJ/it
-conv/pointwise_1x1,arch-a,PASS,PASS,0.071 s,0.00 MiB,0.00 MiB,1,1,0.01 ms,80.24 mW,987244.24 pJ,0.084 s,0.00 MiB,0.00 MiB,3,1,131000.00 samples/s,0.01 ms,47.08 mW,380210.74 pJ/it
-conv/pointwise_tiled_chain,arch-a,PASS,PASS,0.819 s,0.01 MiB,0.04 MiB,20,80,0.04 ms,153.88 mW,6445455.20 pJ,0.777 s,0.05 MiB,0.08 MiB,22,80,12500.00 samples/s,0.08 ms,69.67 mW,5573378.45 pJ/it
-conv/real_asymmetric_padding,arch-a,PASS,PASS,0.074 s,0.01 MiB,0.03 MiB,29,28,0.01 ms,153.67 mW,2221606.72 pJ,0.087 s,0.00 MiB,0.03 MiB,31,28,104000.00 samples/s,0.01 ms,135.38 mW,1295814.97 pJ/it
-conv/relu_conv_store,arch-a,PASS,PASS,0.102 s,0.16 MiB,0.67 MiB,168,184,0.56 ms,183.08 mW,103057892.80 pJ,0.291 s,0.32 MiB,0.67 MiB,168,166,1640.00 samples/s,0.61 ms,182.39 mW,113644022.20 pJ/it
-conv/same_lower_3x3,arch-a,PASS,PASS,0.069 s,0.01 MiB,0.02 MiB,26,25,0.01 ms,166.15 mW,2215009.00 pJ,0.088 s,0.01 MiB,0.03 MiB,28,25,114000.00 samples/s,0.01 ms,134.46 mW,1180460.00 pJ/it
-conv/same_padding_3x3,arch-a,PASS,PASS,0.062 s,0.01 MiB,0.02 MiB,26,25,0.01 ms,166.15 mW,2215009.00 pJ,0.083 s,0.01 MiB,0.03 MiB,28,25,114000.00 samples/s,0.01 ms,134.46 mW,1180460.00 pJ/it
-conv/strategy_depthwise_16,arch-a,PASS,PASS,0.093 s,0.06 MiB,0.35 MiB,168,168,0.34 ms,197.94 mW,66331479.08 pJ,0.298 s,0.15 MiB,0.37 MiB,168,168,2890.00 samples/s,0.35 ms,196.87 mW,70672344.81 pJ/it
-conv/strategy_input_k_tiled,arch-a,PASS,PASS,0.079 s,0.08 MiB,0.27 MiB,109,108,0.35 ms,170.81 mW,60422605.92 pJ,0.120 s,0.16 MiB,0.30 MiB,85,101,3520.00 samples/s,0.28 ms,138.29 mW,40167697.42 pJ/it
-conv/strategy_output_channel_tiled,arch-a,PASS,PASS,0.079 s,0.03 MiB,0.16 MiB,74,72,0.09 ms,155.74 mW,14244739.28 pJ,0.146 s,0.08 MiB,0.25 MiB,111,72,12000.00 samples/s,0.08 ms,137.73 mW,12695085.91 pJ/it
-conv/strategy_streamed_packed,arch-a,PASS,PASS,0.168 s,3.34 MiB,7.89 MiB,168,168,9.35 ms,179.86 mW,1682364509.56 pJ,0.453 s,5.38 MiB,7.87 MiB,127,126,119.00 samples/s,8.39 ms,175.52 mW,1616768905.00 pJ/it
-conv/strategy_streamed_patch,arch-a,PASS,PASS,0.110 s,0.34 MiB,1.32 MiB,168,168,1.90 ms,181.91 mW,346476645.64 pJ,0.416 s,0.84 MiB,1.29 MiB,127,126,525.00 samples/s,1.90 ms,176.18 mW,359355537.30 pJ/it
-conv/strategy_tiled_2d,arch-a,PASS,PASS,0.170 s,0.11 MiB,0.44 MiB,168,168,0.42 ms,182.13 mW,75690907.84 pJ,0.235 s,0.28 MiB,0.45 MiB,130,168,3010.00 samples/s,0.33 ms,178.45 mW,62153061.01 pJ/it
-conv/stride_2,arch-a,PASS,PASS,0.060 s,0.01 MiB,0.00 MiB,5,4,0.01 ms,110.78 mW,580154.96 pJ,0.061 s,0.01 MiB,0.00 MiB,7,4,297000.00 samples/s,0.00 ms,48.26 mW,163092.63 pJ/it
-conv/with_bias_3x3,arch-a,PASS,PASS,0.069 s,0.00 MiB,0.01 MiB,4,3,0.01 ms,104.16 mW,776220.72 pJ,0.066 s,0.00 MiB,0.01 MiB,4,3,128000.00 samples/s,0.01 ms,28.71 mW,224217.97 pJ/it
-conv/with_constant,arch-a,PASS,PASS,0.058 s,0.00 MiB,0.00 MiB,1,1,0.01 ms,81.74 mW,541270.24 pJ,0.067 s,0.00 MiB,0.00 MiB,4,1,138000.00 samples/s,0.01 ms,90.41 mW,664255.74 pJ/it
-conv/without_kernel_shape_attr,arch-a,PASS,PASS,0.064 s,0.01 MiB,0.01 MiB,10,9,0.01 ms,123.80 mW,889640.16 pJ,0.066 s,0.01 MiB,0.01 MiB,12,9,219000.00 samples/s,0.00 ms,83.71 mW,382318.91 pJ/it
-conv/yolo11n_depthwise_head,arch-a,PASS,PASS,2.447 s,8.66 MiB,34.24 MiB,168,255,42.70 ms,200.52 mW,8562449708.00 pJ,3.011 s,22.90 MiB,34.20 MiB,168,216,19.40 samples/s,51.59 ms,195.15 mW,10205156420.00 pJ/it
-conv/yolo11n_heavy,arch-a,PASS,PASS,0.585 s,4.82 MiB,19.10 MiB,161,800,8.54 ms,350.86 mW,2994764012.00 pJ,1.897 s,10.40 MiB,20.59 MiB,161,800,83.80 samples/s,11.93 ms,299.23 mW,3739084612.00 pJ/it
-conv/yolo11n_stem,arch-a,PASS,PASS,0.996 s,12.86 MiB,37.59 MiB,168,488,14.24 ms,301.23 mW,4289558753.00 pJ,1.726 s,22.34 MiB,32.79 MiB,168,362,23.80 samples/s,42.04 ms,214.78 mW,9030156087.00 pJ/it
-div/after_gemm,arch-a,PASS,PASS,0.065 s,0.01 MiB,0.01 MiB,5,4,0.01 ms,104.70 mW,815012.96 pJ,0.071 s,0.01 MiB,0.01 MiB,6,4,145000.00 samples/s,0.01 ms,31.45 mW,216167.21 pJ/it
-div/basic,arch-a,PASS,PASS,0.057 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.22 mW,25266.00 pJ,0.057 s,0.00 MiB,0.00 MiB,1,0,3120000.00 samples/s,0.00 ms,2.23 mW,658.67 pJ/it
-div/channel_broadcast_1024,arch-a,PASS,PASS,0.060 s,0.02 MiB,0.01 MiB,1,0,0.01 ms,78.12 mW,540030.00 pJ,0.056 s,0.02 MiB,0.01 MiB,1,0,145000.00 samples/s,0.01 ms,2.11 mW,13388.67 pJ/it
-div/leading_dimension_broadcast,arch-a,PASS,PASS,0.067 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.22 mW,25266.00 pJ,0.060 s,0.00 MiB,0.00 MiB,1,0,3120000.00 samples/s,0.00 ms,2.23 mW,658.67 pJ/it
-div/runtime_scalar_rhs,arch-a,PASS,PASS,0.057 s,0.02 MiB,0.01 MiB,1,0,0.01 ms,78.12 mW,540030.00 pJ,0.055 s,0.02 MiB,0.01 MiB,1,0,145000.00 samples/s,0.01 ms,2.11 mW,13388.67 pJ/it
-div/scalar_constant,arch-a,PASS,PASS,0.054 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.22 mW,25266.00 pJ,0.055 s,0.00 MiB,0.00 MiB,1,0,3120000.00 samples/s,0.00 ms,2.23 mW,658.67 pJ/it
-gather/3d_input_axis1,arch-a,PASS,PASS,0.061 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.08 mW,45990.00 pJ,0.056 s,0.00 MiB,0.00 MiB,1,0,1700000.00 samples/s,0.00 ms,2.08 mW,1174.67 pJ/it
-gather/axis0_matrix_indices,arch-a,PASS,PASS,0.083 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.07 mW,54414.00 pJ,0.072 s,0.00 MiB,0.00 MiB,1,0,1440000.00 samples/s,0.00 ms,2.07 mW,1390.67 pJ/it
-gather/axis1,arch-a,PASS,PASS,0.064 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.06 mW,62526.00 pJ,0.066 s,0.00 MiB,0.00 MiB,1,0,1250000.00 samples/s,0.00 ms,2.06 mW,1598.67 pJ/it
-gather/negative_axis,arch-a,PASS,PASS,0.078 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.03 mW,112134.00 pJ,0.064 s,0.00 MiB,0.00 MiB,1,0,697000.00 samples/s,0.00 ms,2.03 mW,2870.67 pJ/it
-gather/negative_indices,arch-a,PASS,PASS,0.062 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.13 mW,29376.00 pJ,0.062 s,0.00 MiB,0.00 MiB,1,0,2670000.00 samples/s,0.00 ms,2.12 mW,748.67 pJ/it
-gemm/alpha_beta,arch-a,PASS,PASS,0.068 s,0.01 MiB,0.01 MiB,5,4,0.01 ms,105.27 mW,784908.96 pJ,0.070 s,0.01 MiB,0.01 MiB,6,4,153000.00 samples/s,0.01 ms,32.18 mW,210663.21 pJ/it
-gemm/bias_rank2_broadcast,arch-a,PASS,PASS,0.063 s,0.00 MiB,0.01 MiB,5,4,0.01 ms,105.98 mW,749484.96 pJ,0.063 s,0.01 MiB,0.01 MiB,6,4,168000.00 samples/s,0.01 ms,33.68 mW,200469.21 pJ/it
-gemm/dynamic,arch-a,PASS,PASS,0.065 s,0.00 MiB,0.00 MiB,5,0,0.00 ms,91.48 mW,221475.00 pJ,0.062 s,0.00 MiB,0.00 MiB,5,0,471000.00 samples/s,0.00 ms,20.30 mW,43105.75 pJ/it
-gemm/dynamic_alpha,arch-a,PASS,PASS,0.064 s,0.00 MiB,0.00 MiB,5,0,0.00 ms,91.42 mW,298198.00 pJ,0.063 s,0.00 MiB,0.00 MiB,5,0,337000.00 samples/s,0.00 ms,20.28 mW,60117.75 pJ/it
-gemm/dynamic_beta,arch-a,PASS,PASS,0.070 s,0.00 MiB,0.00 MiB,5,0,0.00 ms,91.32 mW,398595.00 pJ,0.059 s,0.00 MiB,0.00 MiB,5,0,246000.00 samples/s,0.00 ms,20.21 mW,82201.75 pJ/it
-gemm/dynamic_bias,arch-a,PASS,PASS,0.066 s,0.00 MiB,0.00 MiB,5,0,0.00 ms,91.45 mW,243703.00 pJ,0.063 s,0.00 MiB,0.00 MiB,5,0,422000.00 samples/s,0.00 ms,20.28 mW,48009.75 pJ/it
-gemm/dynamic_bias_alpha_beta,arch-a,PASS,PASS,0.061 s,0.00 MiB,0.00 MiB,5,0,0.01 ms,91.28 mW,513811.00 pJ,0.077 s,0.00 MiB,0.00 MiB,5,0,188000.00 samples/s,0.01 ms,20.20 mW,107673.75 pJ/it
-gemm/dynamic_transpose_b,arch-a,PASS,PASS,0.064 s,0.00 MiB,0.00 MiB,5,0,0.00 ms,91.38 mW,118883.00 pJ,0.065 s,0.00 MiB,0.00 MiB,5,0,781000.00 samples/s,0.00 ms,20.51 mW,26151.50 pJ/it
-gemm/huge_1024,arch-a,PASS,PASS,0.182 s,0.01 MiB,0.10 MiB,73,64,0.02 ms,215.04 mW,3767885.36 pJ,0.220 s,0.03 MiB,0.10 MiB,73,64,36900.00 samples/s,0.03 ms,148.63 mW,4053069.50 pJ/it
-gemm/large,arch-a,PASS,PASS,0.074 s,0.02 MiB,0.03 MiB,17,16,0.01 ms,140.15 mW,1573768.84 pJ,0.082 s,0.02 MiB,0.03 MiB,17,16,88800.00 samples/s,0.01 ms,84.59 mW,942235.51 pJ/it
-gemm/large_k_small_n,arch-a,PASS,PASS,0.142 s,0.01 MiB,0.01 MiB,9,8,0.00 ms,133.48 mW,633769.92 pJ,0.119 s,0.01 MiB,0.01 MiB,9,8,194000.00 samples/s,0.01 ms,76.91 mW,390598.09 pJ/it
-gemm/non_square,arch-a,PASS,PASS,0.069 s,0.00 MiB,0.01 MiB,5,4,0.00 ms,118.96 mW,419565.96 pJ,0.069 s,0.00 MiB,0.01 MiB,5,4,270000.00 samples/s,0.00 ms,46.78 mW,172713.46 pJ/it
-gemm/scalar_bias,arch-a,PASS,PASS,0.062 s,0.00 MiB,0.01 MiB,5,4,0.01 ms,105.98 mW,749484.96 pJ,0.066 s,0.01 MiB,0.01 MiB,6,4,168000.00 samples/s,0.01 ms,33.68 mW,200469.21 pJ/it
-gemm/small,arch-a,PASS,PASS,0.058 s,0.00 MiB,0.00 MiB,2,2,0.00 ms,90.14 mW,398436.48 pJ,0.059 s,0.00 MiB,0.00 MiB,4,2,327000.00 samples/s,0.00 ms,61.13 mW,188023.48 pJ/it
-gemm/small_k_large_n,arch-a,PASS,PASS,0.112 s,0.01 MiB,0.02 MiB,17,8,0.01 ms,131.01 mW,1043061.92 pJ,0.100 s,0.01 MiB,0.02 MiB,18,8,141000.00 samples/s,0.01 ms,47.48 mW,336507.17 pJ/it
-gemm/square_weights,arch-a,PASS,PASS,0.080 s,0.03 MiB,0.08 MiB,42,40,0.02 ms,151.77 mW,3284393.60 pJ,0.100 s,0.03 MiB,0.09 MiB,44,40,51800.00 samples/s,0.02 ms,115.71 mW,2278356.60 pJ/it
-gemm/transpose_a,arch-a,PASS,PASS,0.063 s,0.00 MiB,0.01 MiB,5,4,0.01 ms,109.14 mW,628868.96 pJ,0.062 s,0.00 MiB,0.01 MiB,6,4,212000.00 samples/s,0.00 ms,38.03 mW,179501.21 pJ/it
-gemm/transpose_a_and_b,arch-a,PASS,PASS,0.070 s,0.00 MiB,0.01 MiB,5,4,0.01 ms,109.14 mW,628868.96 pJ,0.080 s,0.00 MiB,0.01 MiB,6,4,212000.00 samples/s,0.00 ms,38.03 mW,179501.21 pJ/it
-gemm/transpose_b,arch-a,PASS,PASS,0.065 s,0.00 MiB,0.01 MiB,5,4,0.00 ms,118.96 mW,419565.96 pJ,0.069 s,0.00 MiB,0.01 MiB,5,4,270000.00 samples/s,0.00 ms,46.78 mW,172713.46 pJ/it
-gemm/transpose_b_with_bias,arch-a,PASS,PASS,0.064 s,0.01 MiB,0.01 MiB,5,4,0.01 ms,110.55 mW,557818.96 pJ,0.071 s,0.01 MiB,0.01 MiB,5,4,191000.00 samples/s,0.01 ms,38.98 mW,203117.46 pJ/it
-gemm/with_bias,arch-a,PASS,PASS,0.064 s,0.01 MiB,0.01 MiB,5,4,0.01 ms,108.77 mW,604966.96 pJ,0.062 s,0.01 MiB,0.01 MiB,5,4,175000.00 samples/s,0.01 ms,37.33 mW,213443.71 pJ/it
-gemv/all_constant,arch-a,PASS,PASS,0.071 s,0.00 MiB,0.00 MiB,0,0,0.00 ms,2.00 mW,0.00 pJ,0.061 s,0.00 MiB,0.00 MiB,0,0,0.00 samples/s,0.00 ms,2.00 mW,0.00 pJ/it
-gemv/constant_weight,arch-a,PASS,PASS,0.100 s,0.00 MiB,0.01 MiB,6,4,0.01 ms,111.15 mW,573535.96 pJ,0.079 s,0.00 MiB,0.01 MiB,8,4,235000.00 samples/s,0.00 ms,68.14 mW,293181.96 pJ/it
-gemv/non_uniform_bias,arch-a,PASS,PASS,0.080 s,0.00 MiB,0.01 MiB,6,4,0.01 ms,109.82 mW,609371.96 pJ,0.081 s,0.00 MiB,0.01 MiB,8,4,215000.00 samples/s,0.00 ms,66.23 mW,310779.96 pJ/it
-gemv/scalar_bias,arch-a,PASS,PASS,0.092 s,0.00 MiB,0.01 MiB,6,4,0.01 ms,109.82 mW,609371.96 pJ,0.095 s,0.00 MiB,0.01 MiB,8,4,215000.00 samples/s,0.00 ms,66.23 mW,310779.96 pJ/it
-gemv/uniform_bias,arch-a,PASS,PASS,0.090 s,0.00 MiB,0.01 MiB,6,4,0.01 ms,109.82 mW,609371.96 pJ,0.149 s,0.00 MiB,0.01 MiB,8,4,215000.00 samples/s,0.00 ms,66.23 mW,310779.96 pJ/it
-matmul/basic,arch-a,PASS,PASS,0.089 s,0.00 MiB,0.00 MiB,2,2,0.00 ms,90.14 mW,398436.48 pJ,0.071 s,0.00 MiB,0.00 MiB,4,2,327000.00 samples/s,0.00 ms,61.13 mW,188023.48 pJ/it
-matmul/batched_3d,arch-a,PASS,PASS,0.099 s,0.00 MiB,0.01 MiB,5,4,0.01 ms,108.59 mW,646972.96 pJ,0.086 s,0.00 MiB,0.01 MiB,6,4,207000.00 samples/s,0.00 ms,37.52 mW,181507.21 pJ/it
-matmul/batched_3d_dynamic,arch-a,PASS,PASS,0.065 s,0.00 MiB,0.00 MiB,4,0,0.00 ms,92.19 mW,167975.00 pJ,0.065 s,0.00 MiB,0.00 MiB,5,0,736000.00 samples/s,0.00 ms,17.42 mW,23971.67 pJ/it
-matmul/batched_left_constant,arch-a,PASS,PASS,0.069 s,0.00 MiB,0.02 MiB,9,8,0.01 ms,114.39 mW,1009105.92 pJ,0.070 s,0.01 MiB,0.02 MiB,11,8,133000.00 samples/s,0.01 ms,58.19 mW,441494.75 pJ/it
-matmul/batched_lhs_broadcast,arch-a,PASS,PASS,0.065 s,0.00 MiB,0.01 MiB,5,4,0.01 ms,109.39 mW,621440.96 pJ,0.071 s,0.00 MiB,0.01 MiB,6,4,217000.00 samples/s,0.00 ms,38.52 mW,177665.21 pJ/it
-matmul/batched_rhs_broadcast,arch-a,PASS,PASS,0.065 s,0.00 MiB,0.01 MiB,5,4,0.01 ms,108.59 mW,646972.96 pJ,0.065 s,0.00 MiB,0.01 MiB,6,4,207000.00 samples/s,0.00 ms,37.52 mW,181507.21 pJ/it
-matmul/dynamic,arch-a,PASS,PASS,0.061 s,0.00 MiB,0.00 MiB,5,0,0.00 ms,91.42 mW,148195.00 pJ,0.080 s,0.00 MiB,0.00 MiB,5,0,628000.00 samples/s,0.00 ms,20.41 mW,32505.75 pJ/it
-matmul/huge_1024,arch-a,PASS,PASS,0.188 s,0.01 MiB,0.10 MiB,73,64,0.02 ms,215.04 mW,3767885.36 pJ,0.224 s,0.03 MiB,0.10 MiB,73,64,36900.00 samples/s,0.03 ms,148.63 mW,4053069.50 pJ/it
-matmul/left_constant,arch-a,PASS,PASS,0.076 s,0.00 MiB,0.01 MiB,5,4,0.01 ms,108.86 mW,637168.96 pJ,0.068 s,0.00 MiB,0.01 MiB,6,4,208000.00 samples/s,0.00 ms,37.62 mW,180976.21 pJ/it
-matmul/matrix_vector,arch-a,PASS,PASS,0.120 s,0.52 MiB,0.78 MiB,168,173,0.38 ms,202.13 mW,77751814.88 pJ,0.384 s,0.97 MiB,0.72 MiB,127,173,2250.00 samples/s,0.44 ms,193.72 mW,92630594.79 pJ/it
-matmul/vector_matrix,arch-a,PASS,PASS,0.099 s,0.01 MiB,0.01 MiB,9,8,0.01 ms,118.68 mW,879301.92 pJ,0.104 s,0.01 MiB,0.01 MiB,9,8,132000.00 samples/s,0.01 ms,45.10 mW,342617.42 pJ/it
-matmul/yolo_attention,arch-a,PASS,PASS,0.526 s,1.02 MiB,43.44 MiB,168,0,8.15 ms,170.00 mW,1385775865.00 pJ,0.796 s,13.76 MiB,43.56 MiB,136,0,65.40 samples/s,15.29 ms,166.46 mW,2545338467.00 pJ/it
-mul/after_conv,arch-a,PASS,PASS,0.072 s,0.00 MiB,0.00 MiB,4,3,0.01 ms,107.64 mW,586955.72 pJ,0.072 s,0.00 MiB,0.00 MiB,4,3,183000.00 samples/s,0.01 ms,32.66 mW,178132.97 pJ/it
-mul/after_conv_scalar_constant,arch-a,PASS,PASS,0.120 s,0.00 MiB,0.00 MiB,4,3,0.01 ms,107.64 mW,586955.72 pJ,0.160 s,0.00 MiB,0.00 MiB,4,3,183000.00 samples/s,0.01 ms,32.66 mW,178132.97 pJ/it
-mul/basic,arch-a,PASS,PASS,0.061 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.22 mW,25266.00 pJ,0.086 s,0.00 MiB,0.00 MiB,1,0,3120000.00 samples/s,0.00 ms,2.23 mW,658.67 pJ/it
-mul/channel_broadcast_1024,arch-a,PASS,PASS,0.063 s,0.02 MiB,0.01 MiB,1,0,0.01 ms,78.12 mW,540030.00 pJ,0.061 s,0.02 MiB,0.01 MiB,1,0,145000.00 samples/s,0.01 ms,2.11 mW,13388.67 pJ/it
-mul/leading_dimension_broadcast,arch-a,PASS,PASS,0.064 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.22 mW,25266.00 pJ,0.059 s,0.00 MiB,0.00 MiB,1,0,3120000.00 samples/s,0.00 ms,2.23 mW,658.67 pJ/it
-mul/scalar_constant,arch-a,PASS,PASS,0.065 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.22 mW,25266.00 pJ,0.066 s,0.00 MiB,0.00 MiB,1,0,3120000.00 samples/s,0.00 ms,2.23 mW,658.67 pJ/it
-pool/avg_basic,arch-a,PASS,PASS,0.062 s,0.00 MiB,0.00 MiB,1,0,0.01 ms,78.02 mW,931506.00 pJ,0.063 s,0.00 MiB,0.00 MiB,1,0,84000.00 samples/s,0.01 ms,2.02 mW,24067.00 pJ/it
-pool/avg_ceil_mode,arch-a,PASS,PASS,0.062 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.03 mW,340146.00 pJ,0.057 s,0.00 MiB,0.00 MiB,1,0,230000.00 samples/s,0.00 ms,2.03 mW,8810.67 pJ/it
-pool/avg_explicit_padding,arch-a,PASS,PASS,0.125 s,0.00 MiB,0.00 MiB,1,0,0.01 ms,78.03 mW,688356.00 pJ,0.077 s,0.00 MiB,0.00 MiB,1,0,114000.00 samples/s,0.01 ms,2.03 mW,17809.00 pJ/it
-pool/avg_include_pad,arch-a,PASS,PASS,0.064 s,0.00 MiB,0.00 MiB,1,0,0.01 ms,78.02 mW,663612.00 pJ,0.059 s,0.00 MiB,0.00 MiB,1,0,118000.00 samples/s,0.01 ms,2.02 mW,17081.00 pJ/it
-pool/avg_large_channels,arch-a,PASS,PASS,0.069 s,0.04 MiB,0.02 MiB,1,0,0.24 ms,78.00 mW,18399156.00 pJ,0.067 s,0.04 MiB,0.02 MiB,1,0,4250.00 samples/s,0.24 ms,2.00 mW,471428.00 pJ/it
-pool/avg_non_uniform_stride,arch-a,PASS,PASS,0.060 s,0.00 MiB,0.00 MiB,1,0,0.01 ms,78.02 mW,1132254.00 pJ,0.060 s,0.00 MiB,0.00 MiB,1,0,69100.00 samples/s,0.01 ms,2.02 mW,29191.00 pJ/it
-pool/avg_real_asymmetric_padding,arch-a,PASS,PASS,0.070 s,0.00 MiB,0.00 MiB,1,0,0.03 ms,78.02 mW,1966692.00 pJ,0.073 s,0.00 MiB,0.00 MiB,1,0,39700.00 samples/s,0.03 ms,2.02 mW,50961.00 pJ/it
-pool/max_after_conv,arch-a,PASS,PASS,0.069 s,0.00 MiB,0.00 MiB,5,4,0.01 ms,99.12 mW,1210689.96 pJ,0.074 s,0.00 MiB,0.00 MiB,5,4,81600.00 samples/s,0.01 ms,28.10 mW,344619.71 pJ/it
-pool/max_basic,arch-a,PASS,PASS,0.057 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.06 mW,324744.00 pJ,0.055 s,0.00 MiB,0.00 MiB,1,0,241000.00 samples/s,0.00 ms,2.06 mW,8532.67 pJ/it
-pool/max_ceil_mode,arch-a,PASS,PASS,0.061 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.07 mW,151464.00 pJ,0.059 s,0.00 MiB,0.00 MiB,1,0,516000.00 samples/s,0.00 ms,2.07 mW,3972.67 pJ/it
-pool/max_global_style_kernel_equals_input,arch-a,PASS,PASS,0.066 s,0.00 MiB,0.00 MiB,1,0,0.01 ms,78.01 mW,658626.00 pJ,0.063 s,0.00 MiB,0.00 MiB,1,0,119000.00 samples/s,0.01 ms,2.01 mW,16871.00 pJ/it
-pool/max_non_square_kernel,arch-a,PASS,PASS,0.070 s,0.00 MiB,0.00 MiB,1,0,0.01 ms,78.02 mW,1063068.00 pJ,0.063 s,0.00 MiB,0.00 MiB,1,0,73600.00 samples/s,0.01 ms,2.02 mW,27417.00 pJ/it
-pool/max_real_asymmetric_padding,arch-a,PASS,PASS,0.060 s,0.00 MiB,0.00 MiB,1,0,0.01 ms,78.03 mW,814992.00 pJ,0.060 s,0.00 MiB,0.00 MiB,1,0,96100.00 samples/s,0.01 ms,2.03 mW,21173.00 pJ/it
-pool/max_same_upper,arch-a,PASS,PASS,0.066 s,0.00 MiB,0.00 MiB,1,0,0.01 ms,78.04 mW,625068.00 pJ,0.067 s,0.00 MiB,0.00 MiB,1,0,125000.00 samples/s,0.01 ms,2.04 mW,16233.00 pJ/it
-pool/max_stride2_multichannel,arch-a,PASS,PASS,0.067 s,0.00 MiB,0.00 MiB,1,0,0.02 ms,78.02 mW,1247274.00 pJ,0.074 s,0.00 MiB,0.00 MiB,1,0,62700.00 samples/s,0.02 ms,2.02 mW,32153.00 pJ/it
-reduce_mean/4d_spatial,arch-a,PASS,PASS,0.068 s,0.00 MiB,0.00 MiB,3,0,0.00 ms,92.45 mW,29676.00 pJ,0.068 s,0.00 MiB,0.00 MiB,3,0,2310000.00 samples/s,0.00 ms,4.54 mW,1959.17 pJ/it
-reduce_mean/4d_spatial_keepdims_0,arch-a,PASS,PASS,0.072 s,0.00 MiB,0.00 MiB,4,0,0.00 ms,94.35 mW,61801.00 pJ,0.070 s,0.00 MiB,0.00 MiB,4,0,1210000.00 samples/s,0.00 ms,19.43 mW,16020.25 pJ/it
-reduce_mean/after_conv,arch-a,PASS,PASS,0.071 s,0.00 MiB,0.00 MiB,5,3,0.01 ms,106.95 mW,571332.72 pJ,0.075 s,0.00 MiB,0.00 MiB,5,3,183000.00 samples/s,0.01 ms,19.71 mW,107526.72 pJ/it
-reduce_mean/all_axes_keepdims_0,arch-a,PASS,PASS,0.059 s,0.00 MiB,0.00 MiB,2,0,0.00 ms,79.24 mW,30982.00 pJ,0.063 s,0.00 MiB,0.00 MiB,2,0,2530000.00 samples/s,0.00 ms,3.31 mW,1260.00 pJ/it
-reduce_mean/all_axes_keepdims_1,arch-a,PASS,PASS,0.058 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.22 mW,17286.00 pJ,0.067 s,0.00 MiB,0.00 MiB,1,0,4570000.00 samples/s,0.00 ms,2.22 mW,437.33 pJ/it
-reduce_mean/basic,arch-a,PASS,PASS,0.068 s,0.00 MiB,0.00 MiB,4,0,0.00 ms,93.51 mW,34881.00 pJ,0.061 s,0.00 MiB,0.00 MiB,4,0,2600000.00 samples/s,0.00 ms,5.85 mW,2235.67 pJ/it
-reduce_mean/channel_axis_nchw,arch-a,PASS,PASS,0.061 s,0.03 MiB,0.02 MiB,4,0,0.16 ms,93.60 mW,15436518.00 pJ,0.062 s,0.03 MiB,0.08 MiB,4,0,12900.00 samples/s,0.08 ms,5.00 mW,388853.50 pJ/it
-reduce_mean/keepdims_0,arch-a,PASS,PASS,0.067 s,0.00 MiB,0.00 MiB,5,0,0.00 ms,91.40 mW,68368.00 pJ,0.056 s,0.00 MiB,0.00 MiB,5,0,1300000.00 samples/s,0.00 ms,20.71 mW,16115.50 pJ/it
-reduce_mean/large_dimension_1024,arch-a,PASS,PASS,0.057 s,0.01 MiB,0.00 MiB,1,0,0.00 ms,78.02 mW,217278.00 pJ,0.056 s,0.01 MiB,0.00 MiB,1,0,359000.00 samples/s,0.00 ms,2.02 mW,5274.00 pJ/it
-reduce_mean/legacy_axes_1_2_keepdims_1,arch-a,PASS,PASS,0.055 s,0.00 MiB,0.00 MiB,2,0,0.00 ms,79.35 mW,21505.00 pJ,0.058 s,0.00 MiB,0.00 MiB,2,0,3620000.00 samples/s,0.00 ms,3.45 mW,898.00 pJ/it
-reduce_mean/legacy_axis1_keepdims_0,arch-a,PASS,PASS,0.062 s,0.00 MiB,0.00 MiB,9,0,0.00 ms,92.50 mW,183708.00 pJ,0.065 s,0.00 MiB,0.00 MiB,9,0,679000.00 samples/s,0.00 ms,38.84 mW,57998.17 pJ/it
-reduce_mean/legacy_axis1_keepdims_1,arch-a,PASS,PASS,0.060 s,0.00 MiB,0.00 MiB,8,0,0.00 ms,94.56 mW,129830.00 pJ,0.066 s,0.00 MiB,0.00 MiB,8,0,1340000.00 samples/s,0.00 ms,10.15 mW,7594.50 pJ/it
-reduce_mean/legacy_empty_axes_noop,arch-a,PASS,PASS,0.051 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.22 mW,17286.00 pJ,0.055 s,0.00 MiB,0.00 MiB,1,0,4570000.00 samples/s,0.00 ms,2.22 mW,437.33 pJ/it
-reduce_mean/legacy_nchw_spatial,arch-a,PASS,PASS,0.058 s,0.00 MiB,0.00 MiB,3,0,0.00 ms,92.45 mW,29676.00 pJ,0.058 s,0.00 MiB,0.00 MiB,3,0,1720000.00 samples/s,0.00 ms,4.40 mW,2552.75 pJ/it
-reduce_mean/legacy_negative_axis,arch-a,PASS,PASS,0.057 s,0.00 MiB,0.00 MiB,6,0,0.00 ms,93.52 mW,51717.00 pJ,0.060 s,0.00 MiB,0.00 MiB,6,0,1760000.00 samples/s,0.00 ms,8.07 mW,4588.50 pJ/it
-reduce_mean/legacy_reduce_all_keepdims_1,arch-a,PASS,PASS,0.055 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.22 mW,17286.00 pJ,0.054 s,0.00 MiB,0.00 MiB,1,0,4570000.00 samples/s,0.00 ms,2.22 mW,437.33 pJ/it
-reduce_mean/negative_axis,arch-a,PASS,PASS,0.059 s,0.00 MiB,0.00 MiB,6,0,0.00 ms,93.52 mW,51717.00 pJ,0.058 s,0.00 MiB,0.00 MiB,6,0,1760000.00 samples/s,0.00 ms,8.07 mW,4588.50 pJ/it
-relu/4d,arch-a,PASS,PASS,0.056 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.18 mW,40734.00 pJ,0.053 s,0.00 MiB,0.00 MiB,1,0,1930000.00 samples/s,0.00 ms,2.18 mW,1014.00 pJ/it
-relu/after_conv,arch-a,PASS,PASS,0.063 s,0.00 MiB,0.00 MiB,4,3,0.01 ms,107.89 mW,577437.72 pJ,0.067 s,0.00 MiB,0.00 MiB,4,3,187000.00 samples/s,0.01 ms,32.91 mW,176189.97 pJ/it
-relu/after_gemm,arch-a,PASS,PASS,0.069 s,0.01 MiB,0.01 MiB,5,4,0.01 ms,105.16 mW,790056.96 pJ,0.074 s,0.01 MiB,0.01 MiB,6,4,151000.00 samples/s,0.01 ms,32.04 mW,211536.21 pJ/it
-relu/basic,arch-a,PASS,PASS,0.060 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.22 mW,17286.00 pJ,0.057 s,0.00 MiB,0.00 MiB,1,0,4570000.00 samples/s,0.00 ms,2.22 mW,437.33 pJ/it
-reshape/4d_to_2d_flatten,arch-a,PASS,PASS,0.061 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.28 mW,20196.00 pJ,0.061 s,0.00 MiB,0.00 MiB,1,0,3910000.00 samples/s,0.00 ms,2.28 mW,488.00 pJ/it
-reshape/infer_dim_minus_one,arch-a,PASS,PASS,0.055 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.30 mW,12684.00 pJ,0.058 s,0.00 MiB,0.00 MiB,1,0,6250000.00 samples/s,0.00 ms,2.30 mW,308.00 pJ/it
-reshape/same_rank,arch-a,PASS,PASS,0.060 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.30 mW,12684.00 pJ,0.058 s,0.00 MiB,0.00 MiB,1,0,6250000.00 samples/s,0.00 ms,2.30 mW,308.00 pJ/it
-reshape/zero_copies_input_dim,arch-a,PASS,PASS,0.059 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.30 mW,12684.00 pJ,0.056 s,0.00 MiB,0.00 MiB,1,0,6250000.00 samples/s,0.00 ms,2.30 mW,308.00 pJ/it
-resize/height_only,arch-a,PASS,PASS,0.060 s,0.00 MiB,0.00 MiB,4,0,0.00 ms,93.55 mW,64833.00 pJ,0.063 s,0.00 MiB,0.00 MiB,4,0,1880000.00 samples/s,0.00 ms,5.60 mW,2986.00 pJ/it
-resize/nearest_2x,arch-a,PASS,PASS,0.058 s,0.00 MiB,0.00 MiB,4,0,0.00 ms,93.57 mW,109761.00 pJ,0.059 s,0.00 MiB,0.00 MiB,4,0,1450000.00 samples/s,0.00 ms,5.46 mW,3776.00 pJ/it
-resize/nearest_downsample,arch-a,PASS,PASS,0.059 s,0.00 MiB,0.00 MiB,2,0,0.00 ms,79.45 mW,33925.00 pJ,0.059 s,0.00 MiB,0.00 MiB,2,0,2330000.00 samples/s,0.00 ms,3.28 mW,1360.50 pJ/it
-resize/non_uniform_scales,arch-a,PASS,PASS,0.063 s,0.00 MiB,0.00 MiB,6,0,0.00 ms,93.58 mW,164037.00 pJ,0.069 s,0.00 MiB,0.00 MiB,6,0,1250000.00 samples/s,0.00 ms,7.76 mW,6207.25 pJ/it
-resize/width_only,arch-a,PASS,PASS,0.062 s,0.00 MiB,0.00 MiB,2,0,0.00 ms,79.50 mW,53029.00 pJ,0.059 s,0.00 MiB,0.00 MiB,2,0,1700000.00 samples/s,0.00 ms,3.20 mW,1833.50 pJ/it
-resize/with_sizes,arch-a,PASS,PASS,0.061 s,0.00 MiB,0.00 MiB,3,0,0.00 ms,92.54 mW,73756.00 pJ,0.068 s,0.00 MiB,0.00 MiB,3,0,1700000.00 samples/s,0.00 ms,4.39 mW,2586.75 pJ/it
-sigmoid/4d,arch-a,PASS,PASS,0.057 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.18 mW,40734.00 pJ,0.060 s,0.00 MiB,0.00 MiB,1,0,1930000.00 samples/s,0.00 ms,2.18 mW,1014.00 pJ/it
-sigmoid/after_gemm,arch-a,PASS,PASS,0.063 s,0.01 MiB,0.01 MiB,5,4,0.01 ms,105.16 mW,790056.96 pJ,0.062 s,0.01 MiB,0.01 MiB,6,4,151000.00 samples/s,0.01 ms,32.04 mW,211536.21 pJ/it
-sigmoid/basic,arch-a,PASS,PASS,0.055 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.22 mW,17286.00 pJ,0.060 s,0.00 MiB,0.00 MiB,1,0,4570000.00 samples/s,0.00 ms,2.22 mW,437.33 pJ/it
-slice/2d_basic,arch-a,PASS,PASS,0.056 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.30 mW,18948.00 pJ,0.055 s,0.00 MiB,0.00 MiB,1,0,4170000.00 samples/s,0.00 ms,2.30 mW,491.67 pJ/it
-slice/after_conv,arch-a,PASS,PASS,0.067 s,0.00 MiB,0.01 MiB,7,6,0.01 ms,118.19 mW,1335082.88 pJ,0.074 s,0.00 MiB,0.01 MiB,7,6,87400.00 samples/s,0.01 ms,47.90 mW,547806.13 pJ/it
-slice/default_axes,arch-a,PASS,PASS,0.055 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.30 mW,18948.00 pJ,0.054 s,0.00 MiB,0.00 MiB,1,0,4170000.00 samples/s,0.00 ms,2.30 mW,491.67 pJ/it
-slice/large_channel_1024,arch-a,PASS,PASS,0.059 s,0.01 MiB,0.00 MiB,1,0,0.00 ms,78.14 mW,221304.00 pJ,0.052 s,0.01 MiB,0.00 MiB,1,0,353000.00 samples/s,0.00 ms,2.14 mW,5058.00 pJ/it
-slice/nchw_spatial_crop,arch-a,PASS,PASS,0.062 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.24 mW,101868.00 pJ,0.060 s,0.00 MiB,0.00 MiB,1,0,769000.00 samples/s,0.00 ms,2.24 mW,2851.67 pJ/it
-slice/negative_axis,arch-a,PASS,PASS,0.054 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.30 mW,44004.00 pJ,0.055 s,0.00 MiB,0.00 MiB,1,0,1790000.00 samples/s,0.00 ms,2.30 mW,1227.67 pJ/it
-slice/negative_indices,arch-a,PASS,PASS,0.056 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.30 mW,25212.00 pJ,0.054 s,0.00 MiB,0.00 MiB,1,0,3120000.00 samples/s,0.00 ms,2.30 mW,675.67 pJ/it
-slice/step2,arch-a,PASS,PASS,0.054 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.29 mW,159876.00 pJ,0.053 s,0.00 MiB,0.00 MiB,1,0,490000.00 samples/s,0.00 ms,2.29 mW,4619.67 pJ/it
-softmax/3d_last_axis,arch-a,PASS,PASS,0.056 s,0.00 MiB,0.00 MiB,1,0,UNSUPPORTED,UNSUPPORTED,UNSUPPORTED,0.056 s,0.00 MiB,0.00 MiB,1,0,UNSUPPORTED,UNSUPPORTED,UNSUPPORTED,UNSUPPORTED
-softmax/basic,arch-a,PASS,PASS,0.062 s,0.00 MiB,0.00 MiB,1,0,UNSUPPORTED,UNSUPPORTED,UNSUPPORTED,0.058 s,0.00 MiB,0.00 MiB,1,0,UNSUPPORTED,UNSUPPORTED,UNSUPPORTED,UNSUPPORTED
-softmax/channel_axis,arch-a,PASS,PASS,0.060 s,0.00 MiB,0.00 MiB,1,0,UNSUPPORTED,UNSUPPORTED,UNSUPPORTED,0.067 s,0.00 MiB,0.00 MiB,3,0,UNSUPPORTED,UNSUPPORTED,UNSUPPORTED,UNSUPPORTED
-softmax/large_dimension_1024,arch-a,PASS,PASS,0.059 s,0.01 MiB,0.01 MiB,1,0,UNSUPPORTED,UNSUPPORTED,UNSUPPORTED,0.054 s,0.01 MiB,0.01 MiB,1,0,UNSUPPORTED,UNSUPPORTED,UNSUPPORTED,UNSUPPORTED
-softmax/negative_axis,arch-a,PASS,PASS,0.059 s,0.00 MiB,0.00 MiB,1,0,UNSUPPORTED,UNSUPPORTED,UNSUPPORTED,0.060 s,0.00 MiB,0.00 MiB,1,0,UNSUPPORTED,UNSUPPORTED,UNSUPPORTED,UNSUPPORTED
-split/basic,arch-a,PASS,PASS,0.114 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.30 mW,31554.00 pJ,0.067 s,0.00 MiB,0.00 MiB,1,0,2490000.00 samples/s,0.00 ms,2.30 mW,861.67 pJ/it
-split/equal_three_way,arch-a,PASS,PASS,0.060 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.30 mW,44160.00 pJ,0.073 s,0.00 MiB,0.00 MiB,1,0,1780000.00 samples/s,0.00 ms,2.30 mW,1231.67 pJ/it
-split/negative_axis,arch-a,PASS,PASS,0.061 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.29 mW,84786.00 pJ,0.059 s,0.00 MiB,0.00 MiB,1,0,925000.00 samples/s,0.00 ms,2.29 mW,2413.67 pJ/it
-split/uneven_channel_axis_4d,arch-a,PASS,PASS,0.058 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.30 mW,18948.00 pJ,0.060 s,0.00 MiB,0.00 MiB,1,0,4170000.00 samples/s,0.00 ms,2.30 mW,491.67 pJ/it
-sub/after_gemm,arch-a,PASS,PASS,0.065 s,0.01 MiB,0.01 MiB,5,4,0.01 ms,104.70 mW,815012.96 pJ,0.065 s,0.01 MiB,0.01 MiB,6,4,145000.00 samples/s,0.01 ms,31.45 mW,216167.21 pJ/it
-sub/basic,arch-a,PASS,PASS,0.062 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.22 mW,25266.00 pJ,0.055 s,0.00 MiB,0.00 MiB,1,0,3120000.00 samples/s,0.00 ms,2.23 mW,658.67 pJ/it
-sub/broadcast_row,arch-a,PASS,PASS,0.059 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.22 mW,25266.00 pJ,0.077 s,0.00 MiB,0.00 MiB,1,0,3120000.00 samples/s,0.00 ms,2.23 mW,658.67 pJ/it
-sub/channel_broadcast_1024,arch-a,PASS,PASS,0.064 s,0.02 MiB,0.01 MiB,1,0,0.01 ms,78.12 mW,540030.00 pJ,0.058 s,0.02 MiB,0.01 MiB,1,0,145000.00 samples/s,0.01 ms,2.11 mW,13388.67 pJ/it
-sub/constant_lhs_broadcast,arch-a,PASS,PASS,0.057 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.22 mW,25188.00 pJ,0.061 s,0.00 MiB,0.00 MiB,1,0,3120000.00 samples/s,0.00 ms,2.23 mW,656.67 pJ/it
-sub/leading_dimension_broadcast,arch-a,PASS,PASS,0.056 s,0.00 MiB,0.00 MiB,1,0,0.00 ms,78.22 mW,25266.00 pJ,0.057 s,0.00 MiB,0.00 MiB,1,0,3120000.00 samples/s,0.00 ms,2.23 mW,658.67 pJ/it

validation/raptor_validation/artifacts.py

@@ -0,0 +1,33 @@
+from pathlib import Path
+
+
+ARTIFACTS_DIRNAME = "artifacts"
+
+
+def artifacts_dir(workspace_dir: str | Path) -> Path:
+    return Path(workspace_dir) / ARTIFACTS_DIRNAME
+
+
+def runner_uses_library(runner_path: str | Path, library_path: str | Path) -> bool:
+    runner_path = Path(runner_path)
+    library_path = Path(library_path)
+    try:
+        return (
+            runner_path.is_file()
+            and library_path.is_file()
+            and str(library_path.resolve()).encode() in runner_path.read_bytes()
+        )
+    except OSError:
+        return False
+
+
+def remove_lock_files(root: str | Path) -> int:
+    root = Path(root)
+    if not root.exists():
+        return 0
+    removed = 0
+    for path in root.rglob("*.lock"):
+        if path.is_file() or path.is_symlink():
+            path.unlink(missing_ok=True)
+            removed += 1
+    return removed

validation/raptor_validation/pimcomp_models.py

@@ -0,0 +1,39 @@
+from __future__ import annotations
+
+import argparse
+from pathlib import Path
+
+
+REPO = Path(__file__).resolve().parents[2]
+SUITE = REPO / "validation/networks/pimcomp_models"
+MODELS = {
+    "vgg8": SUITE / "vgg8/vgg8-mnist-reconstructed.onnx",
+    "resnet18": SUITE / "resnet18/resnet18-v1-7.onnx",
+    "resnet34": SUITE / "resnet34/resnet34-v1-7.onnx",
+    "googlenet": SUITE / "googlenet/googlenet-12-pimsim-nn.onnx",
+    "yolo11n": SUITE / "yolo11n/yolo11n-pimsim-nn.onnx",
+}
+FUNCTIONAL_MODELS = {
+    **MODELS,
+    "yolo11n": REPO / "validation/networks/yolo11n/depth_51/yolo11n_depth_51.onnx",
+}
+MODEL_NAMES = tuple(MODELS)
+DEFAULT_MODELS = MODEL_NAMES
+ABLATION_DEFAULT_MODELS = ("vgg8", "resnet18", "resnet34", "googlenet")
+
+
+def add_models_argument(
+    parser: argparse.ArgumentParser,
+    default: tuple[str, ...] = DEFAULT_MODELS,
+) -> None:
+    help_text = "Models to run (default: " + ", ".join(default) + ")."
+    if "yolo11n" not in default:
+        help_text += " Select yolo11n explicitly when needed."
+    parser.add_argument(
+        "--models",
+        nargs="+",
+        choices=MODEL_NAMES,
+        default=list(default),
+        metavar="MODEL",
+        help=help_text,
+    )

validation/raptor_validation/raptor.py

@@ -8,13 +8,13 @@ from .subprocess_utils import run_command_with_reporter
 
 PIM_PASS_LABELS = (
     ("ONNXToSpatialPass", "ONNX to Spatial"),
-    ("MergeComputeNodesPass", "Merge Compute Nodes"),
-    ("SpatialToPimPass", "Spatial to PIM"),
-    ("PimBufferizationPass", "Bufferize PIM"),
-    ("HostConstantFoldingPass", "Fold Host Constants"),
-    ("PimLocalMemoryPlanningPass", "Plan Local Memory"),
-    ("VerificationPass", "Verify PIM"),
-    ("EmitPimCodePass", "Emit PIM Code"),
+    ("MergeComputeNodesPass", "Merge compute nodes"),
+    ("SpatialToPimPass", "Spatial to Pim"),
+    ("PimBufferizationPass", "Bufferize Pim"),
+    ("HostConstantFoldingPass", "Fold host constants"),
+    ("PimLocalMemoryPlanningPass", "Plan local memory"),
+    ("VerificationPass", "Verify Pim"),
+    ("EmitPimCodePass", "Emit Pim code"),
 )
 PIM_PASS_LABEL_BY_SUFFIX = dict(PIM_PASS_LABELS)
 TIMING_LINE_RE = re.compile(r"^\s*([0-9]+\.[0-9]+)\s+\(\s*[0-9.]+%\)\s+(.+?)\s*$")

validation/raptor_validation/validate_one.py

@@ -9,6 +9,7 @@ import numpy as np
 from dataclasses import dataclass, field
 from pathlib import Path
 from colorama import Style, Fore
+from .artifacts import artifacts_dir, runner_uses_library
 from .gen_network_runner import gen_network_runner
 from .onnx_utils import (
     _ONNX_TO_NP,
@@ -26,13 +27,13 @@ from .subprocess_utils import run_command_with_reporter
 
 STAGE_TITLES = (
     "Compile ONNX",
-    "Build Runner",
-    "Generate Inputs",
-    "Run Reference",
-    "Compile PIM",
-    "Run Functional Simulation",
-    "Compare Outputs",
-    "Run Non-functional Simulation",
+    "Build runner",
+    "Generate inputs",
+    "Run reference",
+    "Compile Pim",
+    "Run functional simulation",
+    "Compare outputs",
+    "Run non-functional simulation",
 )
 STAGE_COLORS = {
     STAGE_TITLES[0]: Fore.BLUE,
@@ -46,8 +47,7 @@ STAGE_COLORS = {
 }
 STAGE_COUNT = len(STAGE_TITLES)
 GENERATED_DIR_NAMES = (
-    "inputs", "outputs", "pimcomp", "raptor", "runner", "simulation",
-    "throughput_validation",
+    "artifacts",
 )
 
 MODE_FULL = "full"
@@ -58,15 +58,15 @@ MODE_STAGE_TITLES = {
     MODE_FULL: STAGE_TITLES,
     MODE_COMPILE_ONLY: (
         "Compile ONNX",
-        "Build Runner",
-        "Compile PIM",
+        "Build runner",
+        "Compile Pim",
     ),
     MODE_RUN_ONLY: (
-        "Generate Inputs",
-        "Run Reference",
-        "Run Functional Simulation",
-        "Compare Outputs",
-        "Run Non-functional Simulation",
+        "Generate inputs",
+        "Run reference",
+        "Run functional simulation",
+        "Compare outputs",
+        "Run non-functional simulation",
     ),
 }
 
@@ -75,7 +75,7 @@ PIMSIM_FAILED = "ERROR"
 PIMSIM_UNSUPPORTED = "UNSUPPORTED"
 PIMSIM_SKIPPED = "SKIP"
 PIMSIM_NOT_RUN = "-"
-PIMSIM_UNSUPPORTED_VSOFTMAX = "pimsim-nn does not support opcode vsoftmax"
+PIMSIM_UNSUPPORTED_VSOFTMAX = "does not support opcode vsoftmax"
 
 
 class PimSimUnsupportedError(RuntimeError):
@@ -338,7 +338,7 @@ def run_pimsim_nn(
         else ("throughput", "average_latency_ms", "average_power_mw", "average_energy_pj")
     )
     if any(name not in metrics for name in required):
-        raise RuntimeError(f"pimsim-nn output did not contain required {execution_mode} metrics")
+        raise RuntimeError(f"Pimsim output did not contain required {execution_mode} metrics")
     return metrics
 
 
@@ -361,6 +361,11 @@ def clean_workspace_artifacts(workspace_dir, model_stem):
     for suffix in (".onnx.mlir", ".so", ".tmp"):
         remove_path(workspace_dir / f"{model_stem}{suffix}")
 
+    for path in workspace_dir.rglob("*.lock"):
+        if path.is_file() or path.is_symlink():
+            path.unlink(missing_ok=True)
+            removed_paths.append(path)
+
     return removed_paths
 
 
@@ -407,32 +412,34 @@ def build_dump_ranges(config_path, outputs_descriptor):
 
 
 def build_pim_simulator_command(
-        pim_dir, output_bin_path, dump_ranges, input_paths, mode="latency",
+        pim_dir, output_bin_path, dump_ranges, input_dir, batch_size, mode="latency",
         batch_output_dir=None):
     if mode not in ("latency", "throughput"):
         raise ValueError(f"unknown simulator mode: {mode}")
-    if not input_paths:
+    if batch_size < 1:
         raise ValueError("simulator requires at least one input")
+    if input_dir is None:
+        raise ValueError("simulator requires an input directory")
     command = [
         "cargo", "run", "--no-default-features", "--release", "--package", "pim-simulator", "--bin", "pim-simulator",
         "--", "-f", str(pim_dir), "-o", str(output_bin_path), "-d", dump_ranges,
-        "--mode", mode, "--batch-size", str(len(input_paths)),
+        "--mode", mode, "--batch-size", str(batch_size),
+        "--input-dir", str(input_dir),
     ]
     if batch_output_dir is not None:
         command += ["--batch-output-dir", str(batch_output_dir)]
-    for path in input_paths:
-        command += ["--input", str(path)]
     return command
 
 
 def run_pim_simulator(
         simulator_dir, pim_dir, output_bin_path, dump_ranges, reporter=None,
-        timeout_sec=None, input_paths=(), mode="latency", batch_output_dir=None):
+        timeout_sec=None, input_dir=None, batch_size=1, mode="latency", batch_output_dir=None):
     command = build_pim_simulator_command(
         pim_dir,
         output_bin_path,
         dump_ranges,
-        input_paths,
+        input_dir,
+        batch_size,
         mode=mode,
         batch_output_dir=batch_output_dir,
     )
@@ -516,7 +523,7 @@ def validate_execution(
         try:
             print_stage(
                 reporter, model_index, model_total, model_name,
-                "Run Functional Simulation", name,
+                "Run functional simulation", name,
             )
             write_inputs_to_memory_bin(
                 pim_dir / "memory.bin", pim_dir / "config.json", input_batch[0])
@@ -526,13 +533,13 @@ def validate_execution(
             run_pim_simulator(
                 simulator_dir, pim_dir, simulation_dir / "out.bin", dump_ranges,
                 reporter=reporter, timeout_sec=command_timeout_seconds,
-                input_paths=input_paths[:batch_size], mode=name,
+                input_dir=input_paths[0].parent, batch_size=batch_size, mode=name,
                 batch_output_dir=output_dir)
             reporter.advance()
 
             print_stage(
                 reporter, model_index, model_total, model_name,
-                "Compare Outputs", name,
+                "Compare outputs", name,
             )
             reporter.suspend()
             try:
@@ -553,10 +560,13 @@ def validate_execution(
 
     print_stage(
         reporter, model_index, model_total, model_name,
-        "Run Non-functional Simulation", name,
+        "Run non-functional simulation", name,
     )
     config_path = execution["pimsim_config"]
-    if state["compiled"] and pimsim_nn_build_dir is not None and config_path is not None:
+    if state["compiled"] and state["resource_metrics"].get("used_core_count") == 0:
+        state["pimsim_status"] = PIMSIM_SKIPPED
+        print_info(reporter, "Pimsim non-functional simulation skipped: no active cores")
+    elif state["compiled"] and pimsim_nn_build_dir is not None and config_path is not None:
         try:
             state["metrics"] = run_pimsim_nn(
                 pimsim_nn_build_dir, pim_dir, config_path, name,
@@ -580,7 +590,7 @@ def validate_execution(
     elif not state["compiled"]:
         state["pimsim_status"] = PIMSIM_NOT_RUN
     else:
-        print_info(reporter, "pimsim-nn non-functional simulation skipped")
+        print_info(reporter, "Pimsim non-functional simulation skipped")
     reporter.advance()
 
 
@@ -609,12 +619,12 @@ def validate_network(network_onnx_path, raptor_path, onnx_include_dir,
     owns_reporter = reporter is None
     reporter = reporter or ProgressReporter(model_total, stages_per_model=len(MODE_STAGE_TITLES[mode]), verbose=verbose)
 
-    workspace_dir = network_onnx_path.parent
+    workspace_dir = artifacts_dir(network_onnx_path.parent)
     raptor_dir = workspace_dir / "raptor"
     runner_dir = workspace_dir / "runner"
     runner_build_dir = runner_dir / "build"
     if mode != MODE_RUN_ONLY:
-        clean_workspace_artifacts(workspace_dir, network_onnx_path.stem)
+        clean_workspace_artifacts(network_onnx_path.parent, network_onnx_path.stem)
         Path.mkdir(raptor_dir, parents=True, exist_ok=True)
         Path.mkdir(runner_build_dir, parents=True, exist_ok=True)
 
@@ -669,7 +679,7 @@ def validate_network(network_onnx_path, raptor_path, onnx_include_dir,
                 print_info(reporter, f"Shared library saved to {network_so_path}")
                 reporter.advance()
 
-                print_stage(reporter, model_index, model_total, network_onnx_path.name, "Build Runner")
+                print_stage(reporter, model_index, model_total, network_onnx_path.name, "Build runner")
                 gen_network_runner(
                     network_onnx_path, network_so_path, onnx_include_dir,
                     entry="run_main_graph", out=runner_dir / "runner.c", verbose=False)
@@ -683,7 +693,10 @@ def validate_network(network_onnx_path, raptor_path, onnx_include_dir,
                 report_validation_failure(reporter, "reference", "compilation", exc)
         else:
             required_paths = (network_so_path, network_mlir_path, runner_path)
-            reference_ready = all(path.exists() for path in required_paths)
+            reference_ready = (
+                all(path.exists() for path in required_paths)
+                and runner_uses_library(runner_path, network_so_path)
+            )
             if not reference_ready:
                 report_validation_failure(reporter, "reference", "artifact lookup", FileNotFoundError(
                     "run-only mode requires the compiled shared library, ONNX MLIR, and runner"))
@@ -696,7 +709,7 @@ def validate_network(network_onnx_path, raptor_path, onnx_include_dir,
                 states[name]["compiled"] = (pim_dir / "config.json").exists()
                 if not states[name]["compiled"]:
                     report_validation_failure(reporter, name, "artifact lookup", FileNotFoundError(
-                        f"run-only mode requires compiled PIM artifacts at {pim_dir}"))
+                        f"run-only mode requires compiled Pim artifacts at {pim_dir}"))
                 else:
                     states[name]["resource_metrics"] = collect_pim_resource_metrics(pim_dir)
                     if name == "latency":
@@ -705,7 +718,7 @@ def validate_network(network_onnx_path, raptor_path, onnx_include_dir,
             try:
                 print_stage(
                     reporter, model_index, model_total, network_onnx_path.name,
-                    "Compile PIM", name,
+                        "Compile Pim", name,
                 )
                 root.mkdir(parents=True, exist_ok=True)
                 started = time.perf_counter()
@@ -724,7 +737,7 @@ def validate_network(network_onnx_path, raptor_path, onnx_include_dir,
                 states[name]["resource_metrics"] = collect_pim_resource_metrics(pim_dir)
                 if name == "latency":
                     resource_metrics = states[name]["resource_metrics"]
-                print_info(reporter, f"PIM artifacts saved to {pim_dir}")
+                print_info(reporter, f"Pim artifacts saved to {pim_dir}")
             except Exception as exc:
                 report_validation_failure(reporter, name, "compilation", exc)
             reporter.advance()
@@ -735,7 +748,7 @@ def validate_network(network_onnx_path, raptor_path, onnx_include_dir,
         else:
             input_batch = input_paths = reference_dirs = outputs_descriptor = None
             try:
-                print_stage(reporter, model_index, model_total, network_onnx_path.name, "Generate Inputs")
+                print_stage(reporter, model_index, model_total, network_onnx_path.name, "Generate inputs")
                 inputs_descriptor, outputs_descriptor = onnx_io(network_onnx_path)
                 first_inputs, _ = gen_random_inputs(inputs_descriptor, seed=seed)
                 input_batch = generate_input_batch(
@@ -754,7 +767,7 @@ def validate_network(network_onnx_path, raptor_path, onnx_include_dir,
 
                 if not reference_ready:
                     raise FileNotFoundError("reference runner is unavailable")
-                print_stage(reporter, model_index, model_total, network_onnx_path.name, "Run Reference")
+                print_stage(reporter, model_index, model_total, network_onnx_path.name, "Run reference")
                 reference_dirs = []
                 for index, flags in enumerate(input_flags):
                     reference_dir = workspace_dir / "outputs" / f"{index:06d}"

validation/tools/classification_local_image_validation.py

@@ -16,6 +16,7 @@ if str(VALIDATION_DIR) not in sys.path:
     sys.path.insert(0, str(VALIDATION_DIR))
 
 from raptor_validation.onnx_utils import _ONNX_TO_NP, onnx_io, write_inputs_binary, write_inputs_to_memory_bin
+from raptor_validation.artifacts import artifacts_dir
 from raptor_validation.validate_one import (
     MODE_COMPILE_ONLY,
     build_dump_ranges,
@@ -75,10 +76,11 @@ def ensure_local_artifacts(args, model_path: Path):
 
 
 def ensure_existing_artifacts(model_dir: Path):
+    artifact_root = artifacts_dir(model_dir)
     required_paths = [
-        model_dir / "runner" / "build" / "runner",
-        model_dir / "raptor" / "pim" / "config.json",
-        model_dir / "raptor" / "pim" / "memory.bin",
+        artifact_root / "runner" / "build" / "runner",
+        artifact_root / "raptor" / "pim" / "config.json",
+        artifact_root / "raptor" / "pim" / "memory.bin",
     ]
     missing = [str(path) for path in required_paths if not path.exists()]
     if missing:
@@ -185,13 +187,13 @@ def draw_classification_panel(image: Image.Image, results, output_path: Path):
 
 
 def run_reference_and_simulator(args, model_path: Path, tensor: np.ndarray):
-    model_dir = model_path.parent
-    runner_build_dir = model_dir / "runner" / "build"
+    artifact_root = artifacts_dir(model_path.parent)
+    runner_build_dir = artifact_root / "runner" / "build"
     runner_path = runner_build_dir / "runner"
-    pim_dir = model_dir / "raptor" / "pim"
-    simulation_dir = model_dir / "classification_demo" / "simulation"
-    reference_dir = model_dir / "classification_demo" / "reference"
-    inputs_dir = model_dir / "classification_demo" / "inputs"
+    pim_dir = artifact_root / "raptor" / "pim"
+    simulation_dir = artifact_root / "classification_demo" / "simulation"
+    reference_dir = artifact_root / "classification_demo" / "reference"
+    inputs_dir = artifact_root / "classification_demo" / "inputs"
 
     simulation_dir.mkdir(parents=True, exist_ok=True)
     reference_dir.mkdir(parents=True, exist_ok=True)
@@ -222,7 +224,7 @@ def run_reference_and_simulator(args, model_path: Path, tensor: np.ndarray):
     subprocess.run(runner_cmd, cwd=runner_build_dir, check=True)
 
     write_inputs_to_memory_bin(pim_dir / "memory.bin", pim_dir / "config.json", [tensor])
-    input_bin_path = simulation_dir / "input.bin"
+    input_bin_path = simulation_dir / "input_0.bin"
     write_inputs_binary(input_bin_path, [tensor])
     dump_ranges = build_dump_ranges(pim_dir / "config.json", output_descriptors)
     output_bin_path = simulation_dir / "out.bin"
@@ -232,7 +234,8 @@ def run_reference_and_simulator(args, model_path: Path, tensor: np.ndarray):
         output_bin_path,
         dump_ranges,
         timeout_sec=args.command_timeout_seconds,
-        input_paths=[input_bin_path],
+        input_dir=input_bin_path.parent,
+        batch_size=1,
     )
 
     output_index, output_name, output_dtype_code, output_shape = output_descriptors[0]

validation/tools/conv_lowering_tiles_drawio.py

@@ -37,55 +37,55 @@ class Tile:
 
 
 TILES = (
-    Tile("classic-reference", "REFERENCE", "Classic Conv + exact weight unfolding",
+    Tile("classic-reference", "Reference", "Classic Conv + exact weight unfolding",
          "Original OIHW weights; the two implementations choose different K orders.",
          "reference", "Y[p,o] = Σc,kh,kw Xpatch[p,c,kh,kw] · W[o,c,kh,kw]",
          "At every output position, multiply the patch by one filter and add every product."),
-    Tile("pimcomp-element", "PIMCOMP", "Element pipeline",
+    Tile("pimcomp-element", "Pimcomp", "Element pipeline",
          "One patch vector per input cycle; mapped weights stay fixed.",
          "pimcomp_element", "patchPIM[1×K] · WflatPIM[K×O] → Yp[1×O]",
          "Keep Wflat in the arrays; stream one patch each cycle to produce all O outputs."),
-    Tile("pimcomp-batch", "PIMCOMP", "Batch / replicated pipeline",
+    Tile("pimcomp-batch", "Pimcomp", "Batch / replicated pipeline",
          "Complete Wflat copies divide patches or input samples.",
          "pimcomp_batch", "for replica r: Yr = patchr[1×K] · WflatPIM[K×O]",
          "Copy all weights R times and send different patches to the copies in parallel."),
-    Tile("raptor-legacy-im2col", "RAPTOR", "Legacy explicit im2col",
+    Tile("raptor-legacy-im2col", "Raptor", "Legacy explicit im2col",
          "Every patch becomes one row of a global P×K matrix.",
          "legacy", "Y[P×O] = im2col(X)[P×K] · WflatR[K×O]",
          "Write every image patch as one matrix row, then multiply the two large matrices."),
-    Tile("raptor-packed-im2col", "RAPTOR", "Packed im2col",
+    Tile("raptor-packed-im2col", "Raptor", "Packed im2col",
          "Pack q patch rows and repeat Wflat on a block diagonal.",
          "packed", "packedY[1×qO] = [patch0|…|patchq−1] · diag(WflatR,…,WflatR)",
          "Join q patches and use diagonal weight copies so one multiply computes q independent outputs."),
-    Tile("raptor-streamed-patch", "RAPTOR", "Streamed patch",
+    Tile("raptor-streamed-patch", "Raptor", "Streamed patch",
          "Gather one patch into bounded scratch; avoid global im2col.",
          "streamed_patch", "Yp[1×O] = scratchPatchp[1×K] · WflatR[K×O]",
          "Gather one patch, multiply it, write its output, and reuse scratch for the next patch."),
-    Tile("raptor-streamed-packed", "RAPTOR", "Streamed packed",
+    Tile("raptor-streamed-packed", "Raptor", "Streamed packed",
          "Gather q patch rows in bounded scratch, then block-diagonal pack them.",
          "streamed_packed", "packedY = packedScratch[1×qK] · diag(WflatR×q)[qK×qO]",
          "Gather q patches in small scratch, join them, multiply by diagonal weights, then unpack q outputs."),
-    Tile("raptor-depthwise", "RAPTOR", "Depthwise special case",
+    Tile("raptor-depthwise", "Raptor", "Depthwise special case",
          "Each channel owns one row-major 3×3 kernel; channels never reduce together.",
          "depthwise", "Y[p,c] = Σkh,kw Xpatch[p,c,kh,kw] · W[c,kh,kw]",
          "For each channel separately, multiply its nine patch values by its nine weights and add."),
-    Tile("raptor-output-channel-tiled", "RAPTOR", "Output-channel tiled",
+    Tile("raptor-output-channel-tiled", "Raptor", "Output-channel tiled",
          "Every O tile retains all channel-major K rows and selects output columns.",
          "c_tiled", "Y[:,Oj] = patch[1×K] · Wflat[:,Oj][K×|Oj|]; concat j",
          "Reuse the full patch for each output-filter group, then join the output groups."),
-    Tile("raptor-input-k-tiled", "RAPTOR", "Input-K tiled",
+    Tile("raptor-input-k-tiled", "Raptor", "Input-K tiled",
          "Split matching K ranges; add their partial output vectors.",
          "k_tiled", "Y[1×O] = Σi patch[Ki] · Wflat[Ki,:]",
          "Multiply matching K slices independently, then add their partial output vectors."),
-    Tile("raptor-tiled-2d", "RAPTOR", "Two-dimensional tiled",
+    Tile("raptor-tiled-2d", "Raptor", "Two-dimensional tiled",
          "Partition both K rows and output-filter columns.",
          "tiled_2d", "Y[:,Oj] = Σi patch[Ki] · Wflat[Ki,Oj]; concat j",
          "Split both directions: add results down K and join results across output groups."),
-    Tile("raptor-row-strip", "RAPTOR", "Pixel-major row-strip",
+    Tile("raptor-row-strip", "Raptor", "Pixel-major row-strip",
          "A lane forms patches across one output row and slices K.",
          "row_strip", "for x: Y[r,x,:] = Σi patch[r,x,Ki] · Wflat[Ki,:]",
          "Move across one output row; at each x form a patch, multiply its K slices, and add."),
-    Tile("raptor-row-strip-c-tiled", "RAPTOR", "Row-strip + output tiling",
+    Tile("raptor-row-strip-c-tiled", "Raptor", "Row-strip + output tiling",
          "Each row lane is duplicated across disjoint output-column tiles.",
          "row_strip_c", "for x,j: Y[r,x,Oj] = patch[r,x,:] · Wflat[:,Oj]",
          "Give each output-filter group a copy of the row lane, then join their output columns."),
@@ -473,7 +473,7 @@ def operation_scene(d: Drawio, parent: str, scene: str) -> None:
 def reference_body(d: Drawio, parent: str) -> None:
     add_box(d, parent, 24, 252, 712, 158, "", fill="#ffffff", stroke=PIMCOMP)
     add_text(d, parent, 40, 260, 680, 22,
-             "PIMCOMP: spatial-major, row-wise positions; C interleaved", 11,
+             "Pimcomp: Spatial-major, row-wise positions; C interleaved", 11,
              color=PIMCOMP, align="center", bold=True)
     order_vector(d, parent, 306, "pimcomp", weight=False, label="Input patch")
     order_vector(d, parent, 366, "pimcomp", weight=True, label="Matching W")
@@ -483,7 +483,7 @@ def reference_body(d: Drawio, parent: str) -> None:
 
     add_box(d, parent, 24, 424, 712, 158, "", fill="#ffffff", stroke=RAPTOR)
     add_text(d, parent, 40, 432, 680, 22,
-             "RAPTOR: channel-major; each 3×3 plane is row-major", 11,
+             "Raptor: channel-major; each 3×3 plane is row-major", 11,
              color=RAPTOR, align="center", bold=True)
     order_vector(d, parent, 478, "raptor", weight=False, label="Input patch")
     order_vector(d, parent, 538, "raptor", weight=True, label="Matching W")
@@ -500,22 +500,22 @@ def reference_body(d: Drawio, parent: str) -> None:
 def layout_reference(d: Drawio, parent: str, tile: Tile, accent: str) -> None:
     if tile.scene == "depthwise":
         layout = "independent row-major Kc=9 per channel"
-    elif tile.owner == "PIMCOMP":
-        layout = "PIMCOMP spatial-major K order"
+    elif tile.owner == "Pimcomp":
+        layout = "Pimcomp Spatial-major K order"
     else:
-        layout = "RAPTOR channel-major K order"
+        layout = "Raptor channel-major K order"
     add_box(d, parent, 24, 140, 712, 42, "", fill=PALE, stroke=accent)
     add_text(d, parent, 38, 146, 684, 30,
-             f"LAYOUT → see REFERENCE tile: {layout}", 10,
+             f"Layout → see Reference tile: {layout}", 10,
              color=accent, align="center", bold=True)
 
 
 def algorithm_card(d: Drawio, parent: str, tile: Tile, accent: str) -> None:
     add_box(d, parent, 24, 638, 712, 102, "", fill="#ffffff", stroke=accent)
-    add_text(d, parent, 40, 646, 90, 34, "ALGORITHM", 9,
+    add_text(d, parent, 40, 646, 90, 34, "Algorithm", 9,
              color=accent, bold=True)
     add_text(d, parent, 132, 644, 588, 38, tile.algorithm, 10)
-    add_text(d, parent, 40, 690, 90, 34, "MATH", 9,
+    add_text(d, parent, 40, 690, 90, 34, "Math", 9,
              color=accent, bold=True)
     add_text(d, parent, 132, 686, 588, 42, tile.formula, 10,
              color=accent, bold=True)
@@ -524,7 +524,7 @@ def algorithm_card(d: Drawio, parent: str, tile: Tile, accent: str) -> None:
 def render_tile(d: Drawio, tile: Tile, index: int) -> None:
     col, row = index % COLS, index // COLS
     parent = d.group(col * (TILE + GAP), row * (TILE + GAP), tile.slug)
-    accent = {"REFERENCE": REFERENCE, "PIMCOMP": PIMCOMP, "RAPTOR": RAPTOR}[tile.owner]
+    accent = {"Reference": REFERENCE, "Pimcomp": PIMCOMP, "Raptor": RAPTOR}[tile.owner]
     add_box(d, parent, 0, 0, TILE, TILE, "", fill="#fbfcff", stroke=accent,
             stroke_width=3)
     add_box(d, parent, 24, 20, 106, 28, tile.owner, fill=accent, stroke=accent,

validation/tools/pim/ablation/README.md

@@ -0,0 +1,66 @@
+# Raptor compiler ablation
+
+`run_ablation.py` performs the complete synchronization/Spatial-planning
+ablation study on the Pimcomp model suite. By default it runs `vgg8`,
+`resnet18`, `resnet34`, and `googlenet` across `arch-a` and `arch-b`, latency,
+and throughput pipeline 4. `arch-c` and `yolo11n` are run only when selected
+explicitly.
+Latency is pipeline 1; the wrapper invokes the suite runner separately for
+latency and throughput pipeline 4.
+
+```bash
+.venv/bin/python validation/tools/pim/ablation/run_ablation.py \
+  --jobs 4
+```
+
+## Variants
+
+| Variant | Raptor options |
+|---|---|
+| `baseline` | None. |
+| `no-sync` | `--pim-disable-synchronization` |
+| `no-plan` | `--pim-disable-spatial-planning` |
+| `no-sync-no-plan` | Both options. |
+
+Every variant runs Raptor only. Pimcomp is not compiled, validated, or
+simulated. Reference inputs and outputs are generated once under the shared
+common-artifact root and reused by every variant. Ctrl+C terminates the active
+variant and all of its worker jobs.
+
+The percentage baseline is `baseline`, so its available values are `+0.00%`.
+Every other
+variant reports the signed percentage difference of its Raptor metrics from
+that reference for the same model, architecture, mode, and pipeline. Positive
+values mean the metric is higher; negative values mean it is lower.
+
+## Options
+
+| Option | Description and default |
+|---|---|
+| `-h`, `--help` | Show help and exit. |
+| `--out-dir PATH` | Suite root. Default: `validation/networks/pimcomp_models`. Model artifacts are stored below `<out-dir>/<model>/artifacts`; disabled variants use `<arch>/<mode>[/pipelineN]/ablation/<variant>`. Variant summaries remain under `<out-dir>/ablation/<variant>/`. |
+| `--models MODEL [...]` | Models to run. Default: `vgg8 resnet18 resnet34 googlenet`; include `yolo11n` explicitly when needed. |
+| `--variant NAME` | Select a variant from the table above; repeat for multiple variants. Default: all variants. The feature-full baseline and percentage reference are added automatically when needed. |
+| `--dry-run` | Print the suite-runner commands that would run without modifying files. Default: off. |
+
+All options of
+[`run_pimcomp_models.py`](../pimcomp/compare/README.md), including
+`--archs`, `--mode`, `--pipeline`, `--no-fast`, and
+`--raptor-extra-arg=ARG`, are forwarded to each variant. `--out-dir`,
+`--variant`, and `--dry-run` belong to this wrapper; `--only` is reserved for
+the suite runner's comparison mode and is rejected by this wrapper, which
+always invokes `--raptor-only`. If neither `--mode` nor `--pipeline` is
+forwarded, the wrapper uses its default latency and throughput/pipeline-4
+case set. Supplying either option overrides that default and is passed through
+as one suite-runner invocation per variant.
+
+The feature-full `baseline` variant uses the normal `run_pimcomp_models.py`
+artifact paths and is rerun with the same selected cases and forwarded options
+as the disabled variants. The three disabled variants are stored below each model's
+`artifacts/<arch>/<mode>[/pipelineN]/ablation/` directory. Shared reference
+artifacts remain under each model's `artifacts/common` directory. Transient
+per-variant comparison summaries are removed after aggregation. The combined
+table is written to `<out-dir>/results_ablation.csv`; it contains only the variant, case
+identifiers, and signed percentage `latency`, `throughput`, `power`, and
+`energy` columns. These are Raptor metrics;
+Pimcomp metrics are omitted because the ablation invokes Raptor only.

validation/tools/pim/ablation/run_ablation.py

@@ -0,0 +1,327 @@
+#!/usr/bin/env python3
+"""Run the synchronization and Spatial-planning ablation matrix on Pimcomp models."""
+
+from __future__ import annotations
+
+import argparse
+import csv
+import math
+import os
+import shlex
+import signal
+import subprocess
+import sys
+from dataclasses import dataclass
+from pathlib import Path
+
+
+REPO = Path(__file__).resolve().parents[4]
+RUNNER = REPO / "validation/tools/pim/pimcomp/compare/run_pimcomp_models.py"
+DEFAULT_OUT_DIR = REPO / "validation/networks/pimcomp_models"
+sys.path.insert(0, str(REPO / "validation"))
+
+from raptor_validation.pimcomp_models import (
+    ABLATION_DEFAULT_MODELS,
+    add_models_argument,
+)
+from raptor_validation.artifacts import remove_lock_files
+
+
+@dataclass(frozen=True)
+class Variant:
+    name: str
+    raptor_args: tuple[str, ...]
+
+
+VARIANTS = (
+    Variant("baseline", ()),
+    Variant("no-sync", ("--pim-disable-synchronization",)),
+    Variant("no-plan", ("--pim-disable-spatial-planning",)),
+    Variant(
+        "no-sync-no-plan",
+        ("--pim-disable-synchronization", "--pim-disable-spatial-planning"),
+    ),
+)
+VARIANT_BY_NAME = {variant.name: variant for variant in VARIANTS}
+COMPARISON_RESULTS_FILENAME = "results_comparison.csv"
+ABLATION_RESULTS_FILENAME = "results_ablation.csv"
+CASE_FIELDS = ("arch", "model", "mode", "raptor_pipeline")
+PERCENTAGE_FIELDS = (
+    ("raptor_latency_ms", "latency"),
+    ("raptor_throughput_samples_s", "throughput"),
+    ("raptor_power_mw", "power"),
+    ("raptor_energy_pj", "energy"),
+)
+RESULT_FIELDS = (*CASE_FIELDS, *(target for _, target in PERCENTAGE_FIELDS))
+DEFAULT_CASE_ARGUMENTS = (
+    ("latency", ("--mode", "latency")),
+    ("throughput/pipeline4", ("--mode", "throughput", "--pipeline", "4")),
+)
+DEFAULT_CASE_KEYS = frozenset({("latency", "1"), ("throughput", "4")})
+
+
+def parse_args(argv: list[str] | None = None) -> tuple[argparse.Namespace, list[str]]:
+    parser = argparse.ArgumentParser(
+        description="Run the complete compiler ablation matrix on the Pimcomp model suite.",
+        allow_abbrev=False,
+    )
+    parser.add_argument(
+        "--out-dir",
+        type=Path,
+        default=DEFAULT_OUT_DIR,
+        help="Artifact root (default: validation/networks/pimcomp_models).",
+    )
+    add_models_argument(parser, ABLATION_DEFAULT_MODELS)
+    parser.add_argument(
+        "--variant",
+        choices=tuple(VARIANT_BY_NAME),
+        action="append",
+        dest="variants",
+        help="Run only this variant; baseline is added when needed. Repeat as needed.",
+    )
+    parser.add_argument("--dry-run", action="store_true", help="Print runner commands without modifying files.")
+    args, forwarded = parser.parse_known_args(argv)
+    if any(option == "--only" or option.startswith("--only=") for option in forwarded):
+        parser.error("--only is not supported; the ablation wrapper always runs Raptor only")
+    return args, forwarded
+
+
+def selected_variants(names: list[str] | None) -> list[Variant]:
+    requested = set(names or VARIANT_BY_NAME)
+    requested.add("baseline")
+    return [variant for variant in VARIANTS if variant.name in requested]
+
+
+def has_option(arguments: list[str], option: str) -> bool:
+    return any(argument == option or argument.startswith(f"{option}=") for argument in arguments)
+
+
+def runner_argument_sets(forwarded: list[str]) -> tuple[tuple[str, list[str]], ...]:
+    if has_option(forwarded, "--mode") or has_option(forwarded, "--pipeline"):
+        return (("requested", forwarded),)
+    return tuple(
+        (label, [*forwarded, *case_arguments])
+        for label, case_arguments in DEFAULT_CASE_ARGUMENTS
+    )
+
+
+def variant_output_dir(out_dir: Path, variant: Variant) -> Path:
+    return out_dir if variant.name == "baseline" else out_dir / "ablation" / variant.name
+
+
+def remove_variant_summaries(out_dir: Path) -> None:
+    summary_root = out_dir / "ablation"
+    if not summary_root.is_dir():
+        return
+    for summary in summary_root.glob("*/results_comparison.csv"):
+        summary.unlink()
+    for variant_dir in summary_root.iterdir():
+        if variant_dir.is_dir() and not any(variant_dir.iterdir()):
+            variant_dir.rmdir()
+    if not any(summary_root.iterdir()):
+        summary_root.rmdir()
+
+
+def runner_command(
+    out_dir: Path,
+    variant: Variant,
+    models: list[str],
+    forwarded: list[str],
+    common_root: Path,
+    dry_run: bool,
+) -> list[str]:
+    command = [
+        sys.executable,
+        str(RUNNER),
+        "--out-dir",
+        str(out_dir),
+        "--raptor-only",
+        "--models",
+        *models,
+    ]
+    if variant.name != "baseline":
+        command.extend(("--ablation-variant", variant.name))
+    if not has_option(forwarded, "--common-dir"):
+        command.extend(("--common-dir", str(common_root)))
+    command.extend(forwarded)
+    command.extend(f"--raptor-extra-arg={arg}" for arg in variant.raptor_args)
+    if dry_run:
+        command.append("--dry-run")
+    return command
+
+
+def terminate_process_group(process: subprocess.Popen[bytes]) -> None:
+    if process.poll() is not None:
+        return
+    try:
+        os.killpg(process.pid, signal.SIGTERM)
+    except ProcessLookupError:
+        return
+    try:
+        process.wait(timeout=5)
+    except subprocess.TimeoutExpired:
+        try:
+            os.killpg(process.pid, signal.SIGKILL)
+        except ProcessLookupError:
+            return
+        process.wait()
+
+
+def percentage_difference(value: str | None, reference: str | None) -> str:
+    try:
+        current = float(value) if value is not None else math.nan
+        baseline = float(reference) if reference is not None else math.nan
+    except ValueError:
+        return "NA"
+    if not math.isfinite(current) or not math.isfinite(baseline) or baseline == 0:
+        return "NA"
+    return f"{(current - baseline) / baseline * 100:+.2f}%"
+
+
+def aggregate_results(
+    out_dir: Path,
+    variants: list[Variant],
+    selected_cases: frozenset[tuple[str, str]] | None = None,
+    selected_models: frozenset[str] | None = None,
+) -> tuple[Path | None, list[str]]:
+    fields: list[str] | None = None
+    rows_by_variant: dict[str, list[dict[str, str]]] = {}
+    failures = []
+    for variant in variants:
+        results_path = variant_output_dir(out_dir, variant) / COMPARISON_RESULTS_FILENAME
+        if not results_path.is_file():
+            failures.append(f"{variant.name}: missing {results_path}")
+            continue
+        with results_path.open(newline="", encoding="utf-8") as stream:
+            reader = csv.DictReader(stream)
+            if reader.fieldnames is None:
+                failures.append(f"{variant.name}: empty {results_path}")
+                continue
+            if fields is None:
+                fields = reader.fieldnames
+            elif reader.fieldnames != fields:
+                failures.append(f"{variant.name}: inconsistent columns in {results_path}")
+                continue
+            selected_rows = rows_by_variant.setdefault(variant.name, [])
+            for row in reader:
+                if selected_models is not None and row.get("model") not in selected_models:
+                    continue
+                if selected_cases is not None and (
+                    row.get("mode"), row.get("raptor_pipeline")
+                ) not in selected_cases:
+                    continue
+                selected_rows.append(row)
+    if fields is None:
+        return None, failures
+    baseline_rows = {
+        tuple(row.get(field, "") for field in CASE_FIELDS): row
+        for row in rows_by_variant.get("baseline", [])
+    }
+    if "baseline" not in rows_by_variant:
+        failures.append("baseline: missing percentage reference results")
+    result_rows = []
+    variant_order = {variant.name: index for index, variant in enumerate(VARIANTS)}
+    for variant in variants:
+        for row in rows_by_variant.get(variant.name, []):
+            reference = baseline_rows.get(tuple(row.get(field, "") for field in CASE_FIELDS))
+            result_rows.append(
+                {
+                    "variant": variant.name,
+                    **{field: row.get(field, "") for field in CASE_FIELDS},
+                    **{
+                        target: percentage_difference(
+                            row.get(source), reference.get(source) if reference else None
+                        )
+                        for source, target in PERCENTAGE_FIELDS
+                    },
+                }
+            )
+    result_rows.sort(
+        key=lambda row: (
+            row["arch"],
+            row["model"],
+            row["mode"],
+            variant_order[row["variant"]],
+            int(row["raptor_pipeline"]),
+        )
+    )
+    output = out_dir / ABLATION_RESULTS_FILENAME
+    with output.open("w", newline="", encoding="utf-8") as stream:
+        writer = csv.DictWriter(stream, fieldnames=("variant", *RESULT_FIELDS), lineterminator="\n")
+        writer.writeheader()
+        writer.writerows(result_rows)
+    return output, failures
+
+
+def main(argv: list[str] | None = None) -> int:
+    args, forwarded = parse_args(argv)
+    if not RUNNER.is_file():
+        print(f"Missing Pimcomp runner: {RUNNER}", file=sys.stderr)
+        return 1
+    variants = selected_variants(args.variants)
+    out_dir = args.out_dir.resolve()
+    case_sets = runner_argument_sets(forwarded)
+    if args.dry_run:
+        for variant in variants:
+            for _, case_forwarded in case_sets:
+                print(
+                    shlex.join(
+                        runner_command(
+                            out_dir,
+                            variant,
+                            args.models,
+                            case_forwarded,
+                            out_dir,
+                            True,
+                        )
+                    )
+                )
+        return 0
+
+    out_dir.mkdir(parents=True, exist_ok=True)
+    failed = []
+    current_process = None
+    try:
+        for variant in variants:
+            for case_label, case_forwarded in case_sets:
+                command = runner_command(
+                    out_dir, variant, args.models, case_forwarded, out_dir, False
+                )
+                print(f"[{variant.name}/{case_label}] {shlex.join(command)}")
+                current_process = subprocess.Popen(
+                    command,
+                    cwd=REPO,
+                    start_new_session=True,
+                )
+                returncode = current_process.wait()
+                current_process = None
+                if returncode:
+                    failed.append(f"{variant.name}/{case_label}: runner exited with {returncode}")
+    except KeyboardInterrupt:
+        if current_process is not None:
+            terminate_process_group(current_process)
+        remove_lock_files(out_dir)
+        remove_variant_summaries(out_dir)
+        print("Interrupted; terminated the active ablation job.", file=sys.stderr)
+        return 130
+
+    remove_lock_files(out_dir)
+    selected_cases = DEFAULT_CASE_KEYS if len(case_sets) > 1 else None
+    output, aggregation_failures = aggregate_results(
+        out_dir,
+        variants,
+        selected_cases,
+        frozenset(args.models),
+    )
+    remove_variant_summaries(out_dir)
+    failed.extend(aggregation_failures)
+    if output is not None:
+        print(f"Ablation results: {output}")
+    if failed:
+        print("Failed: " + "; ".join(failed), file=sys.stderr)
+        return 1
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

validation/tools/pim/pimcomp/compare/README.md

@@ -0,0 +1,62 @@
+# Pimcomp model comparison
+
+These scripts compare Raptor with Pimcomp on the models in
+`validation/networks/pimcomp_models`.
+
+`run_pimcomp_models.py` runs the supported model suite across the configured
+architectures and simulation modes. `compare_raptor_pimcomp_model.py` is the
+lower-level one-model comparison used by the suite runner.
+
+## Suite runner
+
+Run all five models, Arch-A and Arch-B, latency, and throughput:
+
+```bash
+.venv/bin/python validation/tools/pim/pimcomp/compare/run_pimcomp_models.py
+```
+
+### Options
+
+| Option | Description and default |
+|---|---|
+| `-h`, `--help` | Show help and exit. |
+| `--out-dir PATH` | Suite root; artifacts are placed below `<out-dir>/<model>/artifacts/`. Default: `validation/networks/pimcomp_models`. |
+| `--common-dir PATH` | Shared root for per-model reference artifacts. Default: `<out-dir>/<model>/artifacts/common`. |
+| `--models MODEL [...]` | Models to run: `vgg8`, `resnet18`, `resnet34`, `googlenet`, `yolo11n`. Default: all five; pass a subset to select specific models. |
+| `--archs ARCH [...]` | Pimcomp hardware profiles to run. Default: `arch-a arch-b`. |
+| `--mode {latency,throughput} [...]` | Simulation modes. Default: both. |
+| `--only {raptor,pimcomp}` | Re-run only one compiler side and reuse the other side's existing report. Default: run both sides. |
+| `--raptor-only` | Run Raptor without compiling, validating, or simulating Pimcomp. Default: off. |
+| `--pipeline {1,2,4,8}` | Select one Raptor pipeline. Default: latency pipeline 1 and throughput pipelines 2, 4, and 8. |
+| `--pimsim-time-ms INT` | Throughput convergence deadline. Default: `1000`. |
+| `--batch-size INT` | Functional throughput batch size. Default: `128`. |
+| `--timeout-seconds FLOAT` | Per-stage timeout; `0` means unlimited. Default: `0`. |
+| `-j INT`, `--jobs INT` | Parallel comparison workers. Default: `4`. |
+| `--clean` | Remove generated comparison artifacts and summaries, then exit. Default: off. |
+| `--ablation-variant NAME` | Put generated artifacts below `<mode>[/pipelineN]/ablation/NAME` and write the summary below `ablation/NAME`. Default: none. |
+| `--dry-run` | Print commands without modifying files. Default: off. |
+| `--no-fast` | Disable fast throughput convergence. Default: off. |
+| `--raptor-extra-arg=ARG` | Extra Raptor compiler argument; repeat for multiple arguments. Default: none. |
+
+Arguments beginning with `--` must use the equals form when passed through:
+
+```bash
+.venv/bin/python validation/tools/pim/pimcomp/compare/run_pimcomp_models.py \
+  --models vgg8 \
+  --raptor-extra-arg=--pim-disable-synchronization
+```
+
+The runner writes `results_comparison.csv` under the selected result root. It reuses
+shared model inputs, outputs, and reference runners, and reuses Pimcomp
+artifacts between pipelines in the same model/architecture/mode group. With
+`--raptor-only`, only Raptor results are generated and Pimcomp is not invoked.
+All generated files are kept below each model's `artifacts/` directory;
+`--clean` also removes any `.lock` files left by an interrupted reference
+generation.
+
+## One-model comparator
+
+`compare_raptor_pimcomp_model.py` compares one ONNX model. Its required
+arguments are `--model PATH` and `--out-dir PATH`; use `--help` for the full
+lower-level interface. The suite runner supplies the model, hardware profile,
+simulation mode, and reuse paths automatically.

validation/tools/pim/pimcomp/compare/run_pimcomp_models.py

@@ -19,35 +19,37 @@ from colorama import Fore, Style
 
 
 REPO = Path(__file__).resolve().parents[5]
-SUITE = REPO / "validation/networks/pimcomp_models"
 sys.path.insert(0, str(REPO / "validation"))
 
-from raptor_validation.pimsim_nn import parse_pimsim_nn_metrics  # noqa: E402
-from raptor_validation.validate_one import STAGE_COLORS  # noqa: E402
+from raptor_validation.pimcomp_models import (
+    FUNCTIONAL_MODELS,
+    MODELS,
+    SUITE,
+    add_models_argument,
+)
+from raptor_validation.artifacts import artifacts_dir, remove_lock_files
+from raptor_validation.pimsim_nn import parse_pimsim_nn_metrics
+from raptor_validation.validate_one import STAGE_COLORS
 
 PIMCOMP_SOURCE = REPO / "third_party/PIMCOMP-NN"
 PIMCOMP_CONFIGS = REPO / "validation/pimsim_configs/pimcomp"
-COMPARE = Path(__file__).resolve().with_name("compare_raptor_pimcomp.py")
+COMPARE = Path(__file__).resolve().with_name("compare_raptor_pimcomp_model.py")
 ARCHES = tuple(sorted(path.name for path in PIMCOMP_CONFIGS.iterdir() if path.is_dir()))
-MODELS = {
-    "vgg8": SUITE / "vgg8/vgg8-mnist-reconstructed.onnx",
-    "resnet18": SUITE / "resnet18/resnet18-v1-7.onnx",
-    "resnet34": SUITE / "resnet34/resnet34-v1-7.onnx",
-    "googlenet": SUITE / "googlenet/googlenet-12-pimsim-nn.onnx",
-    "yolo11n": SUITE / "yolo11n/yolo11n-pimsim-nn.onnx",
-}
+DEFAULT_ARCHES = ("arch-a", "arch-b")
 COMPARISONS = (
     ("latency", 1, "element"),
     ("throughput", 2, "batch"),
     ("throughput", 4, "batch"),
     ("throughput", 8, "batch"),
 )
+RESULTS_FILENAME = "results_comparison.csv"
 
 
 @dataclass(frozen=True)
 class ComparisonSpec:
     label: str
     model: Path
+    functional_model: Path
     output_dir: Path
     common_dir: Path
     config: Path
@@ -62,34 +64,60 @@ def model_dir(root: Path | None, name: str) -> Path:
     return root / name if root is not None else MODELS[name].parent
 
 
-def result_dir(root: Path | None, name: str, arch: str, mode: str, pipeline: int) -> Path:
-    base = model_dir(root, name)
+def result_dir(
+    root: Path | None,
+    name: str,
+    arch: str,
+    mode: str,
+    pipeline: int,
+    ablation_variant: str | None = None,
+) -> Path:
+    base = artifacts_dir(model_dir(root, name))
     suffix = "latency" if mode == "latency" else f"throughput/pipeline{pipeline}"
-    return base / arch / suffix
+    result = base / arch / suffix
+    if ablation_variant is not None:
+        result /= Path("ablation") / ablation_variant
+    return result
 
 
-def clean_artifacts(root: Path | None, models: list[str], arches: list[str]) -> int:
+def common_dir(root: Path | None, name: str, common_root: Path | None = None) -> Path:
+    suffix = "common" if FUNCTIONAL_MODELS[name] == MODELS[name] else "common-functional"
+    base = common_root / name if common_root is not None else model_dir(root, name)
+    return artifacts_dir(base) / suffix
+
+
+def clean_artifacts(
+    root: Path | None,
+    models: list[str],
+    common_root: Path | None = None,
+) -> int:
     removed = 0
     for name in models:
         base = model_dir(root, name)
-        arch_dirs = {base / arch for arch in arches}
-        arch_dirs.update(path for path in base.glob("arch-*") if path.is_dir() and not path.is_symlink())
-        for path in arch_dirs:
+        for path in (artifacts_dir(base),):
             if path.is_dir() and not path.is_symlink():
                 shutil.rmtree(path)
                 removed += 1
-        common = base / "common"
-        if common.is_dir() and not common.is_symlink():
-            shutil.rmtree(common)
-            removed += 1
     for path in (
-        (root or SUITE) / "results.csv",
-        (root or SUITE) / "results_latency.csv",
-        (root or SUITE) / "results_throughput.csv",
+        (root or SUITE) / RESULTS_FILENAME,
+        (root or SUITE) / "results_ablation.csv",
     ):
         if path.is_file() or path.is_symlink():
             path.unlink()
             removed += 1
+    summary_root = root or SUITE
+    if (summary_root / "ablation").is_dir():
+        for path in (summary_root / "ablation").glob("*/results_comparison.csv"):
+            path.unlink(missing_ok=True)
+            removed += 1
+    removed += remove_lock_files(summary_root)
+    if common_root is not None:
+        for name in models:
+            path = artifacts_dir(common_root / name)
+            if path.is_dir() and not path.is_symlink():
+                shutil.rmtree(path)
+                removed += 1
+        removed += remove_lock_files(common_root)
     return removed
 
 
@@ -98,8 +126,11 @@ def write_results_csv(
     arch: str,
     models: list[str],
     comparisons: tuple[tuple[str, int, str], ...] = COMPARISONS,
+    ablation_variant: str | None = None,
 ) -> Path:
-    output = (root or SUITE) / "results.csv"
+    output = (root or SUITE) / RESULTS_FILENAME
+    if ablation_variant is not None:
+        output = output.parent / "ablation" / ablation_variant / RESULTS_FILENAME
     fields = (
         "arch",
         "model",
@@ -142,7 +173,9 @@ def write_results_csv(
                     rows.append({field: row.get(field, "NA") for field in fields})
     for name in models:
         for comparison_mode, pipeline, pimcomp_pipeline in comparisons:
-            report_path = result_dir(root, name, arch, comparison_mode, pipeline) / "pimcomp/comparison_report.json"
+            report_path = result_dir(
+                root, name, arch, comparison_mode, pipeline, ablation_variant
+            ) / "pimcomp/comparison_report.json"
             row = {
                 "model": name,
                 "arch": arch,
@@ -205,6 +238,7 @@ def write_results_csv(
             int(row["raptor_pipeline"]),
         )
     )
+    output.parent.mkdir(parents=True, exist_ok=True)
     with open(output, "w", newline="", encoding="utf-8") as f:
         writer = csv.DictWriter(f, fieldnames=fields, lineterminator="\n")
         writer.writeheader()
@@ -227,23 +261,9 @@ def performance_values(performance: dict) -> dict[str, float | None]:
     }
 
 
-def comparison_passed(report: dict, compiler: str | None = None) -> bool:
-    other_compiler = "PIMCOMP" if compiler == "raptor" else "RAPTOR"
-    if any(
-        compiler is None or other_compiler not in failure.get("stage", "").upper()
-        for failure in report.get("failures", [])
-    ):
-        return False
-    compilers = (compiler,) if compiler is not None else ("raptor", "pimcomp")
-    for name in compilers:
-        result = report.get(f"{name}_validation") or {}
-        if result.get("status") != "done" or not result.get("passed"):
-            return False
-    for name in compilers:
-        performance = report.get(f"{name}_performance") or {}
-        if performance.get("error") or performance.get("skipped"):
-            return False
-    return True
+def comparison_passed(report: dict) -> bool:
+    result = report.get("raptor_validation") or {}
+    return result.get("status") == "done" and bool(result.get("passed"))
 
 
 def functional_validation_status(result: dict | None) -> str:
@@ -265,11 +285,11 @@ def format_value(value: float | None) -> str:
 
 
 def print_stage(title: str, color: str) -> None:
-    print("\n" + Style.BRIGHT + color + f"[{title}]" + Style.RESET_ALL, flush=True)
+    print(Style.BRIGHT + color + f"[{title}]" + Style.RESET_ALL, flush=True)
 
 
 def print_completed(label: str, output: str = "") -> None:
-    print("\n" + Style.BRIGHT + Fore.CYAN + f"[Completed {label}]" + Style.RESET_ALL)
+    print(Style.BRIGHT + Fore.CYAN + f"[Completed {label}]" + Style.RESET_ALL)
     if output:
         print(output, end="" if output.endswith("\n") else "\n")
     print("=" * 72, flush=True)
@@ -311,11 +331,12 @@ def validate_pimcomp_source() -> None:
     source = header.read_text(encoding="utf-8")
     for setting in ("int population_num = 200;", "int max_iteration = 1000;"):
         if setting not in source:
-            raise RuntimeError(f"PIMCOMP paper setting is missing: {setting}")
+            raise RuntimeError(f"Pimcomp paper setting is missing: {setting}")
 
 
 def comparison_command(
     model: Path,
+    functional_model: Path,
     result_dir: Path,
     common_dir: Path,
     config: Path,
@@ -326,6 +347,8 @@ def comparison_command(
     batch_size: int,
     timeout: float,
     fast: bool,
+    raptor_extra_args: list[str] | tuple[str, ...] = (),
+    raptor_only: bool = False,
     reuse_raptor_report: Path | None = None,
     reuse_pimcomp_dir: Path | None = None,
     reuse_pimcomp_report: Path | None = None,
@@ -343,6 +366,8 @@ def comparison_command(
         str(COMPARE),
         "--model",
         str(model),
+        "--functional-model",
+        str(functional_model),
         "--out-dir",
         str(result_dir),
         "--common-dir",
@@ -360,16 +385,23 @@ def comparison_command(
         pimcomp_pipeline,
         "--pimcomp-replication",
         "GA",
+        *(["--raptor-only"] if raptor_only else []),
         f"--raptor-extra-arg=--pipeline={pipeline}",
+        *[f"--raptor-extra-arg={arg}" for arg in raptor_extra_args],
         "--timeout-seconds",
         str(timeout),
-        "--fail-on-error",
         *([] if fast else ["--no-fast"]),
         *reuse_args,
     ]
 
 
-def prepare_common_command(model: Path, common_dir: Path, timeout: float) -> list[str]:
+def prepare_common_command(
+    model: Path,
+    common_dir: Path,
+    timeout: float,
+    batch_size: int,
+    jobs: int,
+) -> list[str]:
     return [
         sys.executable,
         str(COMPARE),
@@ -380,6 +412,10 @@ def prepare_common_command(model: Path, common_dir: Path, timeout: float) -> lis
         "--common-dir",
         str(common_dir),
         "--prepare-common",
+        "--batch-size",
+        str(batch_size),
+        "--jobs",
+        str(jobs),
         "--timeout-seconds",
         str(timeout),
     ]
@@ -394,6 +430,7 @@ def comparison_command_for(
     report = spec.output_dir / "pimcomp/comparison_report.json"
     return comparison_command(
         spec.model,
+        spec.functional_model,
         spec.output_dir,
         spec.common_dir,
         spec.config,
@@ -404,13 +441,19 @@ def comparison_command_for(
         args.batch_size,
         args.timeout_seconds,
         not args.no_fast,
+        args.raptor_extra_args,
+        args.raptor_only,
         reuse_raptor_report=(
             report
             if args.only == "pimcomp"
             else None
         ),
-        reuse_pimcomp_dir=spec.shared_pimcomp_dir if reuse_shared_pimcomp and args.only != "raptor" else None,
-        reuse_pimcomp_report=report if args.only == "raptor" else None,
+        reuse_pimcomp_dir=(
+            spec.shared_pimcomp_dir
+            if reuse_shared_pimcomp and args.only != "raptor" and not args.raptor_only
+            else None
+        ),
+        reuse_pimcomp_report=report if args.only == "raptor" and not args.raptor_only else None,
     )
 
 
@@ -483,20 +526,29 @@ def config_path(arch: str, mode: str, sim_time_ms: int, *, write: bool) -> Path:
 
 def main() -> int:
     parser = argparse.ArgumentParser(
-        description="Compare supported PIMCOMP models with Raptor latency and throughput schedules."
+        description="Compare supported Pimcomp models with Raptor latency and throughput schedules."
     )
     parser.add_argument(
         "--out-dir",
         type=Path,
-        help="Result root (default: artifacts beside each model under validation/).",
+        help="Suite root; generated artifacts go below each model's artifacts/ directory (default: validation/networks/pimcomp_models).",
     )
-    parser.add_argument("--models", nargs="+", choices=MODELS, default=list(MODELS))
     parser.add_argument(
-        "--arch",
+        "--common-dir",
+        type=Path,
+        help="Shared root for per-model reference artifacts. Default: inside --out-dir.",
+    )
+    parser.add_argument(
+        "--ablation-variant",
+        help="Place Raptor artifacts below <mode>/ablation/<variant> and write the comparison summary below ablation/<variant>.",
+    )
+    add_models_argument(parser)
+    parser.add_argument(
+        "--archs",
         nargs="+",
         choices=ARCHES,
-        default=list(ARCHES),
-        help="PIM architectures to run (default: all architectures).",
+        default=list(DEFAULT_ARCHES),
+        help=f"Pim architectures to run (default: {', '.join(DEFAULT_ARCHES)}).",
     )
     parser.add_argument(
         "--mode",
@@ -510,6 +562,11 @@ def main() -> int:
         choices=("raptor", "pimcomp"),
         help="Re-run only this compiler's compile, validation, and simulation stages; preserve the other side from its report.",
     )
+    parser.add_argument(
+        "--raptor-only",
+        action="store_true",
+        help="Run only Raptor; do not compile, validate, or simulate Pimcomp. Default: off.",
+    )
     parser.add_argument(
         "--pipeline",
         type=int,
@@ -520,13 +577,13 @@ def main() -> int:
         "--pimsim-time-ms",
         type=int,
         default=1000,
-        help="throughput pimsim-nn convergence deadline in ms (default: 1000).",
+        help="Throughput Pimsim convergence deadline in ms (default: 1000).",
     )
     parser.add_argument(
         "--batch-size",
         type=int,
-        default=128,
-        help="functional throughput batch size (default: 128).",
+        default=64,
+        help="functional throughput and shared reference batch size (default: 64).",
     )
     parser.add_argument(
         "--timeout-seconds",
@@ -550,15 +607,32 @@ def main() -> int:
     parser.add_argument(
         "--no-fast",
         action="store_true",
-        help="Disable fast pimsim-nn throughput convergence for authoritative experiments.",
+        help="Disable fast Pimsim throughput convergence for authoritative experiments.",
+    )
+    parser.add_argument(
+        "--raptor-extra-arg",
+        action="append",
+        default=[],
+        dest="raptor_extra_args",
+        help="Additional argument to pass to Raptor; repeat as needed.",
     )
     args = parser.parse_args()
 
+    if args.ablation_variant and Path(args.ablation_variant).name != args.ablation_variant:
+        parser.error("--ablation-variant must be a single directory name")
+
+    if args.only is not None and args.raptor_only:
+        parser.error("--only cannot be combined with --raptor-only")
+
     out_dir = args.out_dir.resolve() if args.out_dir is not None else None
-    args.arch = list(dict.fromkeys(args.arch))
+    common_root = args.common_dir.resolve() if args.common_dir is not None else None
+    args.archs = list(dict.fromkeys(args.archs))
     args.mode = list(dict.fromkeys(args.mode))
     if args.clean:
-        print(f"Removed {clean_artifacts(out_dir, args.models, args.arch)} comparison artifact path(s).")
+        print(
+            f"Removed {clean_artifacts(out_dir, args.models, common_root)} "
+            "comparison artifact path(s)."
+        )
         return 0
 
     if args.jobs < 1:
@@ -571,7 +645,7 @@ def main() -> int:
         parser.error("--timeout-seconds must be non-negative")
     comparisons_by_arch: dict[str, tuple[tuple[str, int, str], ...]] = {}
     configs_by_arch: dict[str, dict[str, Path]] = {}
-    for arch in args.arch:
+    for arch in args.archs:
         comparisons = tuple(
             comparison for comparison in COMPARISONS
             if comparison[0] in args.mode
@@ -585,17 +659,29 @@ def main() -> int:
         }
         comparisons_by_arch[arch] = comparisons
         configs_by_arch[arch] = configs
+    reference_batch_size = max(
+        1 if mode == "latency" else args.batch_size
+        for comparisons in comparisons_by_arch.values()
+        for mode, _, _ in comparisons
+    )
 
-    missing = [str(MODELS[name]) for name in args.models if not MODELS[name].exists()]
+    missing = [
+        str(path)
+        for name in args.models
+        for path in (MODELS[name], FUNCTIONAL_MODELS[name])
+        if not path.exists()
+    ]
     if missing:
         parser.error(f"missing model(s): {', '.join(missing)}")
 
     if args.only is not None:
         missing_reuse = []
-        for arch in args.arch:
+        for arch in args.archs:
             for name in args.models:
                 for mode, pipeline, _ in comparisons_by_arch[arch]:
-                    comparison_dir = result_dir(out_dir, name, arch, mode, pipeline)
+                    comparison_dir = result_dir(
+                        out_dir, name, arch, mode, pipeline, args.ablation_variant
+                    )
                     required = comparison_dir / "pimcomp/comparison_report.json"
                     if not required.exists():
                         missing_reuse.append(str(required))
@@ -610,15 +696,17 @@ def main() -> int:
                 + ", ".join(missing_reuse)
             )
 
+    if not args.raptor_only:
         validate_pimcomp_source()
     if out_dir is not None and not args.dry_run:
         out_dir.mkdir(parents=True, exist_ok=True)
 
-    print(Style.BRIGHT + f"Found {len(args.models)} PIMCOMP model(s) to compare." + Style.RESET_ALL)
-    print(f"Architectures: {', '.join(args.arch)}")
+    print(Style.BRIGHT + f"Found {len(args.models)} Pimcomp model(s) to compare." + Style.RESET_ALL)
+    print(f"Architectures: {', '.join(args.archs)}")
     print(f"Modes: {', '.join(args.mode)}")
-    print(f"Throughput pimsim time: {args.pimsim_time_ms} ms")
+    print(f"Throughput Pimsim time: {args.pimsim_time_ms} ms")
     print(f"Max parallel jobs: {args.jobs}")
+    print(f"Shared reference batch: {reference_batch_size}")
     print(
         f"Comparison jobs: "
         f"{sum(len(args.models) * len(comparisons) for comparisons in comparisons_by_arch.values())}"
@@ -626,29 +714,38 @@ def main() -> int:
     print(f"Results root: {out_dir or SUITE}")
     print("=" * 72)
 
-    print_stage("Prepare shared artifacts", STAGE_COLORS["Build Runner"])
+    print_stage("Prepare shared artifacts", STAGE_COLORS["Build runner"])
+    try:
         for name in args.models:
             try:
                 run(
                     prepare_common_command(
-                    MODELS[name],
-                    model_dir(out_dir, name) / "common",
+                        FUNCTIONAL_MODELS[name],
+                        common_dir(out_dir, name, common_root),
                         args.timeout_seconds,
+                        reference_batch_size,
+                        args.jobs,
                     ),
                     dry_run=args.dry_run,
                 )
             finally:
                 print_completed(f"shared artifacts: {name}")
+    except KeyboardInterrupt:
+        remove_lock_files(out_dir or SUITE)
+        print("Interrupted; cleaned validation lock files.", file=sys.stderr)
+        return 130
 
     failed = []
     comparison_specs: list[ComparisonSpec] = []
     shared_pimcomp_by_group: dict[tuple[str, str, str], Path] = {}
-    for arch in args.arch:
+    for arch in args.archs:
         comparisons = comparisons_by_arch[arch]
         configs = configs_by_arch[arch]
         for name in args.models:
             for mode, pipeline, pimcomp_pipeline in comparisons:
-                model_result_dir = result_dir(out_dir, name, arch, mode, pipeline)
+                model_result_dir = result_dir(
+                    out_dir, name, arch, mode, pipeline, args.ablation_variant
+                )
                 label = f"{arch}/{name}/{mode}/pipeline{pipeline}"
                 group = (arch, name, mode)
                 shared_pimcomp_dir = shared_pimcomp_by_group.get(group)
@@ -660,8 +757,9 @@ def main() -> int:
                     ComparisonSpec(
                         label=label,
                         model=MODELS[name],
+                        functional_model=FUNCTIONAL_MODELS[name],
                         output_dir=model_result_dir,
-                        common_dir=model_dir(out_dir, name) / "common",
+                        common_dir=common_dir(out_dir, name, common_root),
                         config=configs[mode],
                         mode=mode,
                         pipeline=pipeline,
@@ -681,7 +779,8 @@ def main() -> int:
             if run(command, dry_run=True, check=False):
                 failed.append(spec.label)
     elif comparison_specs:
-        anchor_specs = comparison_specs if args.only == "raptor" else [
+        raptor_only_run = args.only == "raptor" or args.raptor_only
+        anchor_specs = comparison_specs if raptor_only_run else [
             spec for spec in comparison_specs if spec.anchor
         ]
         anchor_jobs = [
@@ -692,10 +791,11 @@ def main() -> int:
             )
             for spec in anchor_specs
         ]
-        dependent_specs = [] if args.only == "raptor" else [
+        dependent_specs = [] if raptor_only_run else [
             spec for spec in comparison_specs if not spec.anchor
         ]
         print_directly = min(args.jobs, len(comparison_specs)) == 1
+        try:
             with (nullcontext(None) if print_directly else TemporaryDirectory(prefix="raptor-pimcomp-")) as log_dir:
                 anchor_failed, log_offset = run_comparison_jobs(
                     anchor_jobs,
@@ -723,33 +823,41 @@ def main() -> int:
                     log_offset,
                 )
                 failed.extend(dependent_failed)
+        except KeyboardInterrupt:
+            remove_lock_files(out_dir or SUITE)
+            print("Interrupted; cleaned validation lock files.", file=sys.stderr)
+            return 130
 
     if args.dry_run:
         return 1 if failed else 0
 
+    remove_lock_files(out_dir or SUITE)
+
     results_path = None
-    for arch in args.arch:
+    for arch in args.archs:
         results_path = write_results_csv(
             out_dir,
             arch,
             args.models,
             comparisons_by_arch[arch],
+            args.ablation_variant,
         )
     assert results_path is not None
-    print_stage(results_path.name, STAGE_COLORS["Compare Outputs"])
+    print_stage(results_path.name, STAGE_COLORS["Compare outputs"])
     print(results_path.read_text(encoding="utf-8"), end="")
-    for arch in args.arch:
+    for arch in args.archs:
         for name in args.models:
             for mode, pipeline, _ in comparisons_by_arch[arch]:
                 label = f"{arch}/{name}/{mode}/pipeline{pipeline}"
-                report_path = result_dir(out_dir, name, arch, mode, pipeline) / "pimcomp/comparison_report.json"
+                report_path = result_dir(
+                    out_dir, name, arch, mode, pipeline, args.ablation_variant
+                ) / "pimcomp/comparison_report.json"
                 if not report_path.exists() or not comparison_passed(
                     json.loads(report_path.read_text(encoding="utf-8")),
-                    args.only,
                 ):
                     if label not in failed:
                         failed.append(label)
-    print("\n" + Style.BRIGHT + Fore.CYAN + "Summary" + Style.RESET_ALL)
+    print(Style.BRIGHT + Fore.CYAN + "Summary" + Style.RESET_ALL)
     total_jobs = sum(len(args.models) * len(comparisons) for comparisons in comparisons_by_arch.values())
     print(Style.BRIGHT + f"Passed: {total_jobs - len(failed)}" + Style.RESET_ALL)
     print(Style.BRIGHT + f"Failed: {len(failed)}" + Style.RESET_ALL)

validation/tools/pim/pimcomp/correctness/README.md

@@ -1,6 +1,49 @@
-# PIMCOMP batch correctness reproduction
+# Pimcomp correctness experiments
 
-PIMCOMP's batch scheduler currently emits an incomplete standalone program for
+## Synchronization ordering
+
+The synchronization experiment checks whether repeated PIMCOMP and Raptor
+programs preserve cross-core global-memory generations under legal execution
+schedules. It builds a small two-convolution model for Arch-A, Arch-B, and
+Arch-C, finds cross-core `ST`/`LD` dependencies, and compares greedy,
+bounded-stall, randomized, and adversarial runs using the Rust simulator's
+provenance trace. Byte-identical batch inputs isolate intermediate-memory
+ordering from the separate host-input lifetime issue.
+
+The experiment records static dependency evidence, dynamic provenance,
+functional output comparisons, architecture contract classifications, and
+per-artifact PIMCOMP/Raptor conclusions. Diagnostic scheduling changes only
+the functional simulator's execution order; it does not modify PIMCOMP or the
+`pimsim-nn` performance oracle.
+
+Prerequisites are the repository virtual environment, the release Raptor
+compiler, built PIMCOMP frontend/backend, and the existing `pimsim-nn` build.
+Run from the repository root with a new or empty output directory:
+
+```bash
+.venv/bin/python \
+  validation/tools/pim/pimcomp/correctness/synchronization/run_experiment.py \
+  --out-dir /tmp/pimcomp-adversarial-sync \
+  --batch-size 4 --seed 0 --self-check
+```
+
+The directory contains:
+
+- [`run_experiment.py`](synchronization/run_experiment.py): orchestration,
+  classification, reporting, and self-checks.
+- [`global_memory.py`](synchronization/global_memory.py): artifact compilation,
+  dependency analysis, simulator execution, and provenance helpers.
+- [`architecture_contract.py`](synchronization/architecture_contract.py) and
+  [`architecture_evidence.json`](synchronization/architecture_evidence.json):
+  conservative architecture-contract evidence and labels.
+
+The main outputs are `adversarial_memory_sync_report.json` and
+`adversarial_memory_sync_report.md`, with per-architecture evidence below the
+same output root.
+
+## Pimcomp batch prefill reproduction
+
+Pimcomp's batch scheduler currently emits an incomplete standalone program for
 models containing post operations. The generated `VerificationInfo.json` uses a
 negative `source_address` to identify the preceding node, but the batch
 verifier resolves that address by copying the provider tensor directly from
@@ -17,33 +60,69 @@ the missing computation visible in Rust functional validation.
 For the checked-in GoogLeNet throughput/pipeline2 artifact, 39 provider tensors
 are referenced by batch loads. Nineteen have generated stores; twenty are
 never written. Preloading the provider tensors with the same ONNX Runtime
-intermediates used by PIMCOMP's verifier makes the exported program pass. This
-reproduces the verifier's input contract; it does not repair PIMCOMP's batch
+intermediates used by Pimcomp's verifier makes the exported program pass. This
+reproduces the verifier's input contract; it does not repair Pimcomp's batch
 schedule.
 
 Run the default reproduction from the repository root:
 
 ```bash
-.venv/bin/python validation/tools/pim/pimcomp/correctness/run_prefill_experiment.py
+.venv/bin/python validation/tools/pim/pimcomp/correctness/prefill/run_experiment.py
 ```
 
-The launcher accepts an alternate comparison directory, model, and work
-work directory, and shared reference-artifact directory:
+The launcher accepts an alternate comparison directory, model, work directory,
+and shared reference-artifact directory:
 
 ```bash
-.venv/bin/python validation/tools/pim/pimcomp/correctness/run_prefill_experiment.py \
-  validation/networks/pimcomp_models/googlenet/arch-a/throughput/pipeline2 \
+.venv/bin/python validation/tools/pim/pimcomp/correctness/prefill/run_experiment.py \
+  validation/networks/pimcomp_models/googlenet/artifacts/arch-a/throughput/pipeline2 \
   validation/networks/pimcomp_models/googlenet/googlenet-12-pimsim-nn.onnx \
-  /tmp/pimcomp-prefill-googlenet \
-  validation/networks/pimcomp_models/googlenet/common
+  validation/networks/pimcomp_models/googlenet/artifacts/arch-a/throughput/pipeline2/correctness/prefill \
+  validation/networks/pimcomp_models/googlenet/artifacts/common
 ```
 
+Without the optional work-directory argument, the experiment uses the same
+`correctness/prefill/` directory below the comparison artifacts.
+
 It runs the exported artifact once with its original memory image and once
-with [`prefill_batch_memory.py`](prefill_batch_memory.py), then compares both
+with [`batch_memory.py`](prefill/batch_memory.py), then compares both
 outputs with the recorded native reference. The expected GoogLeNet result is a
 baseline maximum difference near `6.70705` and a prefilled maximum difference
 near `4.05e-6`.
 
-The issue is in PIMCOMP batch scheduling/validation semantics, not in the Rust
+The issue is in Pimcomp batch scheduling/validation semantics, not in the Rust
 simulator's vector-length interpretation. Vector lengths remain element counts
 as specified by the reference ISA.
+
+## ResNet BatchNorm correctness gap
+
+Pimcomp has a separate correctness limitation in its ResNet element pipeline.
+`BatchNormalization` is an ONNX operation. The frontend's
+[`fuse_operators()` pass](../../../../../third_party/PIMCOMP-NN/frontend/frontend.py#L535)
+marks it as fused and removes the node from the
+scheduled graph, but the released path does not fold the BatchNorm affine
+transform into the convolution weights and bias. The corresponding verifier
+workaround in [`verification.py`](../../../../../third_party/PIMCOMP-NN/verification/verification.py#L99)
+replaces BatchNorm parameters with identity values (scale and variance equal
+to one, bias and mean equal to zero) before running ONNX Runtime.
+
+Therefore Pimcomp's native verifier and exported element program agree with
+each other, but they do not implement the original ResNet ONNX model. On the
+current ResNet-18 Arch-A latency artifact, using the same input:
+
+| Comparison | Maximum absolute difference |
+|---|---:|
+| Pimcomp native verifier vs Rust export | `6.7e-8` |
+| Rust export vs original ONNX reference | `4.8294563` |
+
+The Pimcomp output ranges from approximately `-0.187` to `0.220`, while the
+original ONNX output ranges from `-3.572` to `4.834`; 462 of 1000 final
+elements differ by more than one. This is not a Rust simulator or Python
+exporter regression. It is a Pimcomp model-semantics mismatch caused by
+dropping BatchNorm numerics. The same issue affects the ResNet-34 latency
+artifact. VGG and GoogLeNet do not show this particular mismatch because they
+do not contain the same ResNet BatchNorm path.
+
+The latency comparison intentionally uses the original ONNX reference, so
+these Pimcomp rows must remain `FAIL` until Pimcomp folds BatchNorm correctly
+or the comparison explicitly uses a BatchNorm-neutralized reference.

validation/tools/pim/pimcomp/correctness/prefill/batch_memory.py

@@ -1,5 +1,5 @@
 #!/usr/bin/env python3
-"""Populate the host-side intermediate buffers expected by PIMCOMP batch mode."""
+"""Populate the host-side intermediate buffers expected by Pimcomp batch mode."""
 
 from __future__ import annotations
 
@@ -23,7 +23,7 @@ def flatten_reference(value: np.ndarray) -> np.ndarray:
     elif value.ndim == 2:
         value = value.transpose()
     else:
-        raise ValueError(f"PIMCOMP batch verification only flattens 2D/4D tensors, got {value.shape}")
+        raise ValueError(f"Pimcomp batch verification only flattens 2D/4D tensors, got {value.shape}")
     return value.astype(np.float32, copy=False).reshape(-1)
 
 
@@ -94,7 +94,7 @@ def prefill_batch_memory(
     session = ort.InferenceSession(runtime_model.SerializeToString(), providers=["CPUExecutionProvider"])
     session_inputs = session.get_inputs()
     if len(session_inputs) != 1:
-        raise ValueError("PIMCOMP export currently requires exactly one runtime input tensor")
+        raise ValueError("Pimcomp export currently requires exactly one runtime input tensor")
     input_meta = session_inputs[0]
     input_tensor = np.loadtxt(input_path, delimiter=",", dtype=np.float32).reshape(input_meta.shape)
     provider_names = [node_list[index]["name"] for index in providers]

validation/tools/pim/pimcomp/correctness/prefill/run_experiment.py

@@ -1,5 +1,5 @@
 #!/usr/bin/env python3
-"""Reproduce the PIMCOMP batch prefill correctness experiment."""
+"""Reproduce the Pimcomp batch prefill correctness experiment."""
 
 from __future__ import annotations
 
@@ -7,19 +7,18 @@ import argparse
 import json
 import subprocess
 import sys
-import tempfile
 from pathlib import Path
 
 import numpy as np
 
-from prefill_batch_memory import prefill_batch_memory
+from batch_memory import prefill_batch_memory
 
 
 SCRIPT_DIR = Path(__file__).resolve().parent
-REPO_ROOT = SCRIPT_DIR.parents[4]
-DEFAULT_COMPARISON_DIR = REPO_ROOT / "validation/networks/pimcomp_models/googlenet/arch-a/throughput/pipeline2"
+REPO_ROOT = SCRIPT_DIR.parents[5]
+DEFAULT_COMPARISON_DIR = REPO_ROOT / "validation/networks/pimcomp_models/googlenet/artifacts/arch-a/throughput/pipeline2"
 DEFAULT_MODEL = REPO_ROOT / "validation/networks/pimcomp_models/googlenet/googlenet-12-pimsim-nn.onnx"
-DEFAULT_COMMON_DIR = REPO_ROOT / "validation/networks/pimcomp_models/googlenet/common"
+DEFAULT_COMMON_DIR = REPO_ROOT / "validation/networks/pimcomp_models/googlenet/artifacts/common"
 SIMULATOR_MANIFEST = REPO_ROOT / "backend-simulators/pim/pim-simulator/Cargo.toml"
 
 
@@ -28,6 +27,7 @@ def run_simulator(
     memory: Path,
     output: Path,
     dump: str,
+    input_dir: Path,
 ) -> None:
     subprocess.run(
         [
@@ -44,6 +44,8 @@ def run_simulator(
             str(comparison_dir / "pimcomp/exported"),
             "--memory",
             str(memory),
+            "--input-dir",
+            str(input_dir),
             "-o",
             str(output),
             "-d",
@@ -86,14 +88,18 @@ def main() -> int:
     model = args.model.resolve()
     common_dir = args.common_dir.resolve()
     if args.work_dir is None:
-        work_dir = Path(tempfile.mkdtemp(prefix="pimcomp-prefill."))
+        work_dir = comparison_dir / "correctness/prefill"
+        work_dir.mkdir(parents=True, exist_ok=True)
     else:
         work_dir = args.work_dir.resolve()
         work_dir.mkdir(parents=True, exist_ok=True)
 
     input_path = common_dir / "inputs/in0.csv"
-    if not input_path.is_file():
-        input_path = comparison_dir / "inputs/in0.csv"
+    simulator_input_dir = work_dir / "inputs"
+    simulator_input_dir.mkdir(parents=True, exist_ok=True)
+    np.loadtxt(input_path, delimiter=",", dtype=np.float32).tofile(
+        simulator_input_dir / "input_0.bin"
+    )
 
     prefilled_memory = work_dir / "prefilled_memory.bin"
     metadata_path = work_dir / "metadata.json"
@@ -117,8 +123,15 @@ def main() -> int:
         comparison_dir / "pimcomp/exported/memory.bin",
         baseline_output,
         dump,
+        simulator_input_dir,
+    )
+    run_simulator(
+        comparison_dir,
+        prefilled_memory,
+        prefilled_output,
+        dump,
+        simulator_input_dir,
     )
-    run_simulator(comparison_dir, prefilled_memory, prefilled_output, dump)
     compare_outputs(baseline_output, prefilled_output, reference, work_dir)
     return 0
 

validation/tools/pim/pimcomp/correctness/synchronization/architecture_contract.py

@@ -0,0 +1,187 @@
+#!/usr/bin/env python3
+"""Small, offline architecture-evidence helpers used by the sync experiment.
+
+The adversarial experiment owns the dynamic investigation.  This module keeps
+the stable repository/configuration facts it needs in one place and exposes a
+deliberately small API so the experiment can also be imported as a library.
+"""
+
+from __future__ import annotations
+
+import hashlib
+import json
+import subprocess
+from pathlib import Path
+from typing import Any
+
+
+REPO = Path(__file__).resolve().parents[6]
+VALIDATION = REPO / "validation"
+PIMCOMP_ROOT = REPO / "third_party/PIMCOMP-NN"
+PIMSIM_ROOT = REPO / "backend-simulators/pim/pimsim-nn"
+RUST_ROOT = REPO / "backend-simulators/pim/pim-simulator"
+COMPARE_DIR = REPO / "validation/tools/pim/pimcomp/compare"
+EVIDENCE_PATH = Path(__file__).with_name("architecture_evidence.json")
+
+
+def _default_evidence() -> dict[str, Any]:
+    return {
+        "schema": 1,
+        "status": "documentary_evidence_plus_repository_checks",
+        "architectures": {
+            "arch-a": {
+                "pimcomp_identity": "ISAAC-like static/deterministic timing model",
+                "primary_classification": "STATIC_TIMING_CONTRACT_MAPPING_UNPROVEN",
+                "config": "validation/pimsim_configs/pimcomp/arch-a/throughput_config_1000ms.json",
+                "hardware_reference": "ISAAC (HPCA 2016), documentary mapping requires review of the cited paper/configuration.",
+            },
+            "arch-b": {
+                "pimcomp_identity": "PUMA-like architecture",
+                "primary_classification": "HARDWARE_SYNC_EXISTS_BUT_NOT_MODELED_BY_PIMSIM_NN",
+                "config": "validation/pimsim_configs/pimcomp/arch-b/throughput_config_1000ms.json",
+                "hardware_reference": "PUMA, documentary valid/count synchronization is not encoded in ordinary PIMCOMP LD/ST.",
+            },
+            "arch-c": {
+                "pimcomp_identity": "ISSCC 2023 ReRAM architecture row",
+                "primary_classification": "MAPPING_NOT_ESTABLISHED",
+                "config": "validation/pimsim_configs/pimcomp/arch-c/throughput_config_1000ms.json",
+                "hardware_reference": "ISSCC 2023 ReRAM reference; cross-system mapping is unresolved.",
+            },
+        },
+    }
+
+
+def load_evidence() -> dict[str, Any]:
+    if EVIDENCE_PATH.is_file():
+        return json.loads(EVIDENCE_PATH.read_text(encoding="utf-8"))
+    return _default_evidence()
+
+
+def source_contract() -> dict[str, Any]:
+    """Return the contract claims used for report labeling.
+
+    These labels are intentionally conservative: they are not a substitute
+    for a paper citation and never turn an unordered relation into a safe one.
+    """
+
+    return {
+        "ld_st": "ordinary timed global-memory accesses in the Rust model",
+        "send_recv": "explicit point-to-point synchronization modeled by the simulator",
+        "wait_sync": "instruction-level synchronization when emitted",
+        "valid_count": "not encoded by PIMCOMP exported LD/ST instructions",
+        "static_timing": "not proven as a PIMCOMP-to-ISAAC contract",
+    }
+
+
+def classify_contract(
+    architecture: str, _contract: dict[str, Any], manifest: dict[str, Any]
+) -> str:
+    return manifest["architectures"][architecture]["primary_classification"]
+
+
+def architecture_source_evidence(architecture: str, manifest: dict[str, Any]) -> dict[str, Any]:
+    return dict(manifest["architectures"][architecture])
+
+
+def git_identity(path: Path) -> dict[str, Any]:
+    result: dict[str, Any] = {"path": str(path), "commit": None, "worktree_status": []}
+    if not path.exists():
+        result["error"] = "missing"
+        return result
+    try:
+        result["commit"] = subprocess.run(
+            ["git", "-C", str(path), "rev-parse", "HEAD"],
+            check=True,
+            capture_output=True,
+            text=True,
+        ).stdout.strip()
+        status = subprocess.run(
+            ["git", "-C", str(path), "status", "--short"],
+            check=True,
+            capture_output=True,
+            text=True,
+        ).stdout.splitlines()
+        result["worktree_status"] = status
+    except (OSError, subprocess.CalledProcessError) as exc:
+        result["error"] = f"{type(exc).__name__}: {exc}"
+    return result
+
+
+def tree_hash(root: Path, pattern: str) -> dict[str, str]:
+    values: dict[str, str] = {}
+    for path in sorted(root.glob(pattern)):
+        if not path.is_file():
+            continue
+        digest = hashlib.sha256(path.read_bytes()).hexdigest()
+        values[str(path.relative_to(root))] = digest
+    return values
+
+
+def instruction_graph(artifact: Path) -> tuple[dict[tuple[int, int], list[tuple[int, int]]], dict[str, Any]]:
+    """Build same-core program-order edges from the exported JSON streams.
+
+    Cross-core synchronization is added only when an artifact explicitly
+    carries matching SEND/RECV metadata.  Ordinary global LD/ST creates no
+    edge by construction; that is the relation this experiment is testing.
+    """
+
+    graph: dict[tuple[int, int], list[tuple[int, int]]] = {}
+    streams: dict[int, list[dict[str, Any]]] = {}
+    for path in sorted(artifact.glob("core_*.json"), key=lambda item: int(item.stem.split("_")[1])):
+        try:
+            instructions = json.loads(path.read_text(encoding="utf-8"))
+        except (OSError, json.JSONDecodeError):
+            continue
+        core = int(path.stem.split("_")[1]) + 1
+        streams[core] = instructions if isinstance(instructions, list) else []
+        for index in range(len(streams[core]) - 1):
+            graph.setdefault((core, index), []).append((core, index + 1))
+
+    sends: dict[tuple[int, int], tuple[int, int]] = {}
+    recvs: dict[tuple[int, int], tuple[int, int]] = {}
+    for core, instructions in streams.items():
+        for index, instruction in enumerate(instructions):
+            op = str(instruction.get("op", instruction.get("operation", ""))).lower()
+            peer = instruction.get("core")
+            if peer is None:
+                continue
+            key = (core, int(peer) + 1)
+            if op == "send":
+                sends.setdefault(key, (core, index))
+            elif op == "recv":
+                recvs.setdefault(key, (core, index))
+    for key, send in sends.items():
+        recv = recvs.get((key[1], key[0]))
+        if recv is not None:
+            graph.setdefault(send, []).append(recv)
+    return graph, {"cores": sorted(streams), "explicit_send_recv_edges": len(sends)}
+
+
+def make_identical_inputs(model: Path, batch_size: int, out: Path) -> list[Path]:
+    import numpy as np
+
+    import sys
+
+    if str(COMPARE_DIR) not in sys.path:
+        sys.path.insert(0, str(COMPARE_DIR))
+    import compare_raptor_pimcomp_model as compare  # noqa: PLC0415
+
+    inputs, _ = compare.onnx_io(model)
+    arrays = []
+    for _index, _name, element_type, shape in inputs:
+        dtype = compare._ONNX_TO_NP[element_type]
+        arrays.append(np.full(shape, 1.0, dtype=dtype))
+    flattened = np.concatenate(
+        [compare.flatten_pimcomp_input(array) for array in arrays]
+    ) if arrays else np.empty(0, dtype=np.float32)
+    samples = [[flattened.copy()] for _ in range(batch_size)]
+    return [
+        Path(path)
+        for path in compare.write_input_batch_binaries(
+            samples, out / "inputs/pimcomp_isolated"
+        )
+    ]
+
+
+if __name__ == "__main__":
+    print(json.dumps(load_evidence(), indent=2, sort_keys=True))

validation/tools/pim/pimcomp/correctness/synchronization/architecture_evidence.json

@@ -0,0 +1,24 @@
+{
+  "schema": 1,
+  "status": "documentary_evidence_plus_repository_checks",
+  "architectures": {
+    "arch-a": {
+      "pimcomp_identity": "ISAAC-like static/deterministic timing model",
+      "primary_classification": "STATIC_TIMING_CONTRACT_MAPPING_UNPROVEN",
+      "config": "validation/pimsim_configs/pimcomp/arch-a/throughput_config_1000ms.json",
+      "hardware_reference": "ISAAC (HPCA 2016), documentary mapping requires review of the cited paper/configuration."
+    },
+    "arch-b": {
+      "pimcomp_identity": "PUMA-like architecture",
+      "primary_classification": "HARDWARE_SYNC_EXISTS_BUT_NOT_MODELED_BY_PIMSIM_NN",
+      "config": "validation/pimsim_configs/pimcomp/arch-b/throughput_config_1000ms.json",
+      "hardware_reference": "PUMA, documentary valid/count synchronization is not encoded in ordinary PIMCOMP LD/ST."
+    },
+    "arch-c": {
+      "pimcomp_identity": "ISSCC 2023 ReRAM architecture row",
+      "primary_classification": "MAPPING_NOT_ESTABLISHED",
+      "config": "validation/pimsim_configs/pimcomp/arch-c/throughput_config_1000ms.json",
+      "hardware_reference": "ISSCC 2023 ReRAM reference; cross-system mapping is unresolved."
+    }
+  }
+}

validation/tools/pim/pimcomp/correctness/synchronization/global_memory.py

@@ -0,0 +1,479 @@
+#!/usr/bin/env python3
+"""Reusable build, execution, and provenance helpers for PIMCOMP audits."""
+
+from __future__ import annotations
+
+import json
+import os
+import shutil
+import subprocess
+import sys
+from bisect import bisect_left
+from collections import defaultdict
+from pathlib import Path
+from typing import Any
+
+import numpy as np
+
+
+REPO = Path(__file__).resolve().parents[6]
+VALIDATION = REPO / "validation"
+CONFIG_ROOT = VALIDATION / "pimsim_configs/pimcomp"
+PIMCOMP_ROOT = REPO / "third_party/PIMCOMP-NN"
+PIMSIM_NN_ROOT = REPO / "backend-simulators/pim/pimsim-nn"
+RUST_ROOT = REPO / "backend-simulators/pim/pim-simulator"
+RUST_BINARY = RUST_ROOT / "target/release/pim-simulator"
+COMPARE_SCRIPT = REPO / "validation/tools/pim/pimcomp/compare/compare_raptor_pimcomp_model.py"
+PYTHON = REPO / ".venv/bin/python"
+
+sys.path.insert(0, str(COMPARE_SCRIPT.parent))
+import compare_raptor_pimcomp_model as compare  # noqa: E402
+
+
+class ExperimentError(RuntimeError):
+    pass
+
+
+def architecture_configs(architecture: str) -> tuple[Path, Path]:
+    root = CONFIG_ROOT / architecture
+    return root / "throughput_config_1000ms.json", root / "latency_config.json"
+
+
+def check_prerequisites(throughput: Path, latency: Path) -> None:
+    required = {
+        "PIMCOMP backend": PIMCOMP_ROOT / "build/PIMCOMP-NN",
+        "PIMCOMP frontend": PIMCOMP_ROOT / "frontend/frontend.py",
+        "Raptor compiler": REPO / "build_release/Release/bin/onnx-mlir",
+        "Rust simulator source": RUST_ROOT,
+        "pimsim-nn build": PIMSIM_NN_ROOT / "build",
+        "throughput config": throughput,
+        "latency config": latency,
+    }
+    missing = [f"{label}: {path}" for label, path in required.items() if not path.exists()]
+    if missing:
+        raise ExperimentError("missing prerequisites:\n" + "\n".join(missing))
+
+
+def _run(cmd: list[str], cwd: Path, log: Path, timeout: float = 0.0) -> subprocess.CompletedProcess[str]:
+    log.parent.mkdir(parents=True, exist_ok=True)
+    try:
+        result = subprocess.run(
+            [str(value) for value in cmd],
+            cwd=cwd,
+            stdout=subprocess.PIPE,
+            stderr=subprocess.STDOUT,
+            text=True,
+            timeout=None if timeout <= 0 else timeout,
+        )
+    except subprocess.TimeoutExpired as exc:
+        log.write_text((exc.stdout or "") + "\nTIMEOUT\n", encoding="utf-8")
+        raise ExperimentError(f"command timed out: {' '.join(map(str, cmd))}") from exc
+    log.write_text(result.stdout, encoding="utf-8")
+    if result.returncode:
+        raise ExperimentError(
+            f"command failed ({result.returncode}): {' '.join(map(str, cmd))}\n"
+            f"see {log}\n{result.stdout[-3000:]}"
+        )
+    return result
+
+
+def build_simulator(args: Any, out: Path) -> None:
+    _run(
+        [
+            "cargo", "build", "--release", "--no-default-features",
+            "--package", "pim-simulator", "--bin", "pim-simulator",
+        ],
+        RUST_ROOT,
+        out / "cargo_build.log",
+        float(getattr(args, "timeout", 0.0)),
+    )
+    if not RUST_BINARY.is_file():
+        raise ExperimentError(f"Rust simulator binary was not produced: {RUST_BINARY}")
+
+
+def make_model(path: Path) -> None:
+    import onnx
+    from onnx import TensorProto, helper, numpy_helper
+
+    shape = [1, 64, 8, 8]
+    weights = []
+    for name in ("w0", "w1"):
+        weight = np.zeros((64, 64, 3, 3), dtype=np.float32)
+        for channel in range(64):
+            weight[channel, channel, 1, 1] = 1.0
+        weights.append(numpy_helper.from_array(weight, name=name))
+    input_value = helper.make_tensor_value_info("input", TensorProto.FLOAT, shape)
+    output_value = helper.make_tensor_value_info("output", TensorProto.FLOAT, shape)
+    nodes = [
+        helper.make_node(
+            "Conv", ["input", "w0"], ["hidden"], name="conv0",
+            kernel_shape=[3, 3], strides=[1, 1], pads=[1, 1, 1, 1],
+            dilations=[1, 1], group=1,
+        ),
+        helper.make_node(
+            "Conv", ["hidden", "w1"], ["output"], name="conv1",
+            kernel_shape=[3, 3], strides=[1, 1], pads=[1, 1, 1, 1],
+            dilations=[1, 1], group=1,
+        ),
+    ]
+    graph = helper.make_graph(nodes, "pimcomp_sync_two_conv", [input_value], [output_value], weights)
+    model = helper.make_model(graph, opset_imports=[helper.make_operatorsetid("", 13)])
+    model.ir_version = min(model.ir_version, 8)
+    onnx.checker.check_model(model)
+    path.parent.mkdir(parents=True, exist_ok=True)
+    onnx.save(model, path)
+
+
+def make_inputs(
+    model: Path, batch_size: int, seed: int, out: Path
+) -> tuple[list[list[np.ndarray]], list[Path], list[Path], list[np.ndarray]]:
+    del seed  # deterministic values are intentional; the seed remains in the report.
+    inputs_desc, _ = compare.onnx_io(model)
+    if len(inputs_desc) != 1:
+        raise ExperimentError("the synchronization model must have exactly one input")
+    _index, _name, element_type, shape = inputs_desc[0]
+    dtype = compare._ONNX_TO_NP[element_type]
+    arrays: list[np.ndarray] = [
+        np.full(shape, float(10 ** index), dtype=dtype) for index in range(max(8, batch_size))
+    ]
+    input_batch = [[array] for array in arrays]
+    raptor_paths = [
+        Path(path)
+        for path in compare.write_input_batch_binaries(input_batch, out / "inputs/raptor")
+    ]
+    pimcomp_batch = [[compare.flatten_pimcomp_input(array)] for array in arrays]
+    pimcomp_paths = [
+        Path(path)
+        for path in compare.write_input_batch_binaries(pimcomp_batch, out / "inputs/pimcomp")
+    ]
+    return input_batch, raptor_paths, pimcomp_paths, arrays
+
+
+def make_references(
+    model: Path, input_batch: list[list[np.ndarray]], architecture_out: Path, _args: Any
+) -> list[Path]:
+    import onnxruntime as ort
+
+    input_desc, output_desc = compare.onnx_io(model)
+    session = ort.InferenceSession(str(model), providers=["CPUExecutionProvider"])
+    references: list[Path] = []
+    for index, sample in enumerate(input_batch):
+        values = {input_desc[item][1]: sample[item] for item in range(len(input_desc))}
+        outputs = session.run(None, values)
+        directory = architecture_out / "reference" / f"iteration_{index:06d}"
+        directory.mkdir(parents=True, exist_ok=True)
+        for output, descriptor in zip(outputs, output_desc):
+            output_index, name, _dtype, _shape = descriptor
+            filename = f"output{output_index}_{compare.sanitize_output_name(name)}.csv"
+            np.savetxt(directory / filename, np.asarray(output).reshape(-1), delimiter=",")
+        references.append(directory)
+    return references
+
+
+def compile_artifact(args: Any, model: Path, architecture_out: Path, throughput_config: Path) -> dict[str, Any]:
+    comparison = architecture_out / "comparison"
+    command = [
+        str(PYTHON), str(COMPARE_SCRIPT),
+        "--model", str(model),
+        "--out-dir", str(comparison),
+        "--common-dir", str(architecture_out / "common"),
+        "--pimcomp-config", str(throughput_config),
+        "--pimsim-mode", "throughput",
+        "--pimsim-time-ms", "1000",
+        "--batch-size", str(args.batch_size),
+        "--pimcomp-pipeline", "batch",
+        "--pimcomp-replication", "GA",
+        "--raptor-extra-arg=--pipeline=4",
+        "--seed", str(args.seed),
+        "--timeout-seconds", str(args.timeout),
+    ]
+    if args.no_fast:
+        command.append("--no-fast")
+    log = architecture_out / "comparison_compile.log"
+    try:
+        _run(command, REPO, log, args.timeout)
+    except ExperimentError as exc:
+        report = comparison / "pimcomp/comparison_report.json"
+        if not report.is_file():
+            raise
+        report_data = json.loads(report.read_text(encoding="utf-8"))
+        raptor_error = "; ".join(
+            str(item.get("error", ""))
+            for item in report_data.get("failures", [])
+            if "RAPTOR" in str(item.get("stage", "")).upper()
+        ) or str(exc)
+    else:
+        report = comparison / "pimcomp/comparison_report.json"
+        report_data = json.loads(report.read_text(encoding="utf-8"))
+        raptor_error = "; ".join(
+            str(item.get("error", ""))
+            for item in report_data.get("failures", [])
+            if "RAPTOR" in str(item.get("stage", "")).upper()
+        ) or None
+
+    paths = report_data.get("paths", {})
+    pimcomp = Path(paths["pimcomp_exported_pim"]) if paths.get("pimcomp_exported_pim") else comparison / "pimcomp/exported"
+    pimsim = Path(paths["pimcomp_pimsim_nn"]) if paths.get("pimcomp_pimsim_nn") else comparison / "pimcomp/pimsim_nn"
+    raptor = Path(paths["raptor_pim"]) if paths.get("raptor_pim") else comparison / "raptor/pim.missing"
+    if not pimcomp.is_dir():
+        raise ExperimentError(f"PIMCOMP Rust artifact missing; see {report}")
+    return {
+        "artifact": pimcomp,
+        "pimsim_artifact": pimsim if pimsim.is_dir() else None,
+        "raptor_artifact": raptor if raptor.is_dir() else Path(),
+        "raptor_error": raptor_error,
+        "comparison_report": report,
+    }
+
+
+def _instruction_op(instruction: dict[str, Any]) -> str:
+    return str(instruction.get("op", instruction.get("operation", ""))).lower()
+
+
+def _address(instruction: dict[str, Any], registers: dict[int, int], register: str) -> int | None:
+    try:
+        base = int(registers[int(instruction[register])])
+    except (KeyError, TypeError, ValueError):
+        return None
+    offset = instruction.get("offset") or {}
+    select = int(offset.get("offset_select", 0))
+    value = int(offset.get("offset_value", 0))
+    # LD's global operand is r1 (selector bit 2); ST's global operand is rd
+    # (selector bit 1).  The local simulator uses the same asymmetric ISA.
+    selector_bit = 1 if register == "rd" else 2
+    return base + value if select & selector_bit else base
+
+
+def _static_instruction(
+    core_file_index: int,
+    artifact_format: str,
+    instruction_index: int,
+    instruction: dict[str, Any],
+    address: int,
+    size: int,
+    artifact: Path,
+) -> dict[str, Any]:
+    core = core_file_index + 1
+    core_file = f"core_{core_file_index}.json"
+    binary_file = f"core_{core_file_index}.pim"
+    return {
+        "core": core,
+        # JSON PIMCOMP streams enter the Rust executor after an initial
+        # synthetic slot; Raptor's emitted binary/JSON streams do not.
+        "pc": instruction_index if artifact_format == "binary+json" else instruction_index - 1,
+        "artifact_pc": instruction_index,
+        "address": address,
+        "size": size,
+        "instruction_file": core_file,
+        "execution_file": binary_file if (artifact / binary_file).is_file() else core_file,
+        "artifact_format": artifact_format,
+    }
+
+
+def analyze_artifact(artifact: Path) -> dict[str, Any]:
+    core_paths = sorted(artifact.glob("core_*.json"), key=lambda path: int(path.stem.split("_")[1]))
+    if not core_paths:
+        raise ExperimentError(f"artifact has no core_*.json files: {artifact}")
+    artifact_format = "binary+json" if any(artifact.glob("core_*.pim")) else "json"
+    stores: list[dict[str, Any]] = []
+    loads: list[dict[str, Any]] = []
+    counts: dict[str, int] = defaultdict(int)
+    instruction_files: dict[str, str] = {}
+    participating = 0
+    for path in core_paths:
+        core_file_index = int(path.stem.split("_")[1])
+        instructions = json.loads(path.read_text(encoding="utf-8"))
+        if instructions:
+            participating += 1
+        instruction_files[f"core_{core_file_index}"] = str(path)
+        registers: dict[int, int] = {}
+        for index, instruction in enumerate(instructions):
+            op = _instruction_op(instruction)
+            counts[op] += 1
+            if op in {"sldi", "lldi"} and "rd" in instruction and "imm" in instruction:
+                registers[int(instruction["rd"])] = int(instruction["imm"])
+                continue
+            if op not in {"ld", "st"}:
+                continue
+            address = _address(instruction, registers, "rd" if op == "st" else "rs1")
+            if address is None:
+                continue
+            size = int(instruction.get("size", instruction.get("len", 0)))
+            if size <= 0:
+                continue
+            item = _static_instruction(
+                core_file_index, artifact_format, index, instruction, address, size, artifact
+            )
+            (stores if op == "st" else loads).append(item)
+
+    loads_by_address = sorted(loads, key=lambda item: int(item["address"]))
+    starts = [int(item["address"]) for item in loads_by_address]
+    dependencies: list[dict[str, Any]] = []
+    for store in stores:
+        begin = int(store["address"])
+        end = begin + int(store["size"])
+        first = bisect_left(starts, end)
+        for load in loads_by_address[:first]:
+            if load["core"] == store["core"]:
+                continue
+            load_begin = int(load["address"])
+            load_end = load_begin + int(load["size"])
+            if load_end <= begin:
+                continue
+            dependencies.append({
+                "overlap": {
+                    "address_begin": max(begin, load_begin),
+                    "address_end": min(end, load_end),
+                },
+                "writer": dict(store),
+                "reader": dict(load),
+                "explicit_sync_ordering_evidence": False,
+            })
+    dependencies.sort(key=lambda item: (
+        item["overlap"]["address_begin"], item["writer"]["core"],
+        item["writer"]["pc"], item["reader"]["core"], item["reader"]["pc"],
+    ))
+    return {
+        "artifact_format": artifact_format,
+        "instruction_files": instruction_files,
+        "stores": stores,
+        "loads": loads,
+        "cross_core_dependencies": dependencies,
+        "cross_core_dependency_count": len(dependencies),
+        "participating_core_count": participating,
+        "instruction_counts": dict(sorted(counts.items())),
+        "representative_dependency": dependencies[0] if dependencies else None,
+    }
+
+
+def trace_events(path: Path) -> list[dict[str, Any]]:
+    if not path.is_file():
+        return []
+    events = []
+    for line in path.read_text(encoding="utf-8").splitlines():
+        if line.strip():
+            events.append(json.loads(line))
+    return events
+
+
+def _prov(event: dict[str, Any]) -> tuple[int, ...]:
+    return tuple(int(value) for value in event.get("provenance", []))
+
+
+def _overlap(a: dict[str, Any], b: dict[str, Any]) -> bool:
+    return max(int(a["address"]), int(b["address"])) < min(
+        int(a["address"]) + int(a["size"]), int(b["address"]) + int(b["size"])
+    )
+
+
+def dynamic_analysis(events: list[dict[str, Any]], static: dict[str, Any], config: Path) -> dict[str, Any]:
+    stores = [event for event in events if event.get("event") == "global_store"]
+    loads = [event for event in events if event.get("event") == "global_load"]
+    input_stores = [event for event in events if event.get("event") == "external_input_store"]
+    event_keys = {
+        (int(event.get("core", -1)), int(event.get("pc", -1)), int(event.get("address", -1)), int(event.get("size", -1)))
+        for event in stores + loads
+    }
+    executed = sum(
+        1 for dependency in static["cross_core_dependencies"]
+        if (
+            int(dependency["writer"]["core"]), int(dependency["writer"]["pc"]),
+            int(dependency["writer"]["address"]), int(dependency["writer"]["size"]),
+        ) in event_keys
+        and (
+            int(dependency["reader"]["core"]), int(dependency["reader"]["pc"]),
+            int(dependency["reader"]["address"]), int(dependency["reader"]["size"]),
+        ) in event_keys
+    )
+    cross_links = []
+    for load in loads:
+        for writer in load.get("last_writers", []):
+            if int(writer.get("core", -1)) == 0 or not writer.get("provenance"):
+                continue
+            cross_links.append((load, writer))
+    host_races = []
+    for load in loads:
+        expected = (int(load.get("core_iteration", -1)),)
+        if any(_overlap(load, source) and _prov(load) and _prov(load) != expected for source in input_stores):
+            host_races.append(load)
+    reused_versions = {
+        int(version)
+        for store in stores
+        for version in store.get("overwritten_versions", [])
+    }
+    mixed = [event for event in events if event.get("event") == "cross_sample_data_mix"]
+    return {
+        "executed_cross_core_dependencies": executed,
+        "sample_dependent_cross_core_links": len(cross_links),
+        "host_input_lifetime_races": len(host_races),
+        "send_recv_generation_races": 0,
+        "mixed_sample_operations": len(mixed),
+        "global_memory_versions_reused": len(reused_versions),
+        "global_memory_generation_races": 0,
+        "host_input_race_contaminates_test": bool(host_races),
+        "config": str(config),
+    }
+
+
+def run_rust(
+    artifact: Path,
+    inputs: list[Path],
+    references: list[Path],
+    outputs_desc: list[tuple[int, str, int, list[int]]],
+    out: Path,
+    args: Any,
+    *,
+    schedule_policy: str = "greedy",
+    schedule_seed: int = 0,
+    schedule_target: str | None = None,
+    schedule_deferral_budget: int | None = None,
+    target_stall: str | None = None,
+    channel_last: bool = False,
+) -> dict[str, Any]:
+    if not inputs:
+        raise ExperimentError("Rust run requires at least one input")
+    out.mkdir(parents=True, exist_ok=True)
+    output = out / "output.bin"
+    batch_outputs = out / "iterations"
+    shutil.rmtree(batch_outputs, ignore_errors=True)
+    dump = compare.build_dump_ranges(artifact / "config.json", outputs_desc)
+    mode = "latency" if len(inputs) == 1 else "throughput"
+    command = [
+        str(RUST_BINARY), "--folder", str(artifact), "--output", str(output),
+        "--dump", dump, "--mode", mode, "--batch-size", str(len(inputs)),
+        "--input-dir", str(inputs[0].parent), "--batch-output-dir", str(batch_outputs),
+        "--provenance-trace", str(out / "provenance.jsonl"),
+        "--diagnostic-schedule-policy", schedule_policy,
+        "--diagnostic-schedule-seed", str(schedule_seed),
+    ]
+    if schedule_target:
+        command += ["--diagnostic-schedule-target", schedule_target]
+    if schedule_deferral_budget is not None:
+        command += ["--diagnostic-schedule-deferral-budget", str(schedule_deferral_budget)]
+    if target_stall:
+        command += ["--diagnostic-target-stall", target_stall]
+    log = out / "simulator.log"
+    try:
+        _run(command, RUST_ROOT, log, float(args.timeout))
+    except ExperimentError as exc:
+        return {
+            "passed": False, "max_diffs": {}, "error": str(exc), "completed": False,
+            "command": [str(value) for value in command],
+        }
+    failed: list[int] = []
+    max_diffs: dict[str, float] = {}
+    for index, reference in enumerate(references[: len(inputs)]):
+        result = compare.compare_simulator_outputs(
+            batch_outputs / f"output_{index:06d}.bin", outputs_desc, reference,
+            threshold=args.threshold, rtol=args.rtol, channel_last=channel_last,
+        )
+        if not result.passed:
+            failed.append(index)
+        for name, value in result.max_diffs.items():
+            max_diffs[name] = max(max_diffs.get(name, 0.0), value)
+    return {
+        "passed": not failed, "max_diffs": max_diffs,
+        "failed_iterations": failed, "completed": True,
+        "command": [str(value) for value in command],
+        "trace": str(out / "provenance.jsonl"),
+    }