compact memory contiguity with for loops

# Conflicts:
#	src/PIM/Dialect/Pim/Transforms/Bufferization/PimBufferizationPass.cpp

.gitignore

@@ -4,14 +4,11 @@
 
 .claude
 .codex
-AGENTS.md
 
 CMakeUserPresets.json
 
-build
-build_release
-cmake-build-debug
-cmake-build-release
+build_*
 compile.sh
+pimcomp_utils/*
 
 **/__*

AGENTS.md

@@ -0,0 +1,92 @@
+- Always read the full README.md before doing anything.
+- Build commands:
+    - `cmake --build ./build_release`
+    - `cmake --build ./build_debug`
+- Never use `ninja` directly: it bypasses cmake's configuration and invalidates the build cache.
+- Always tries the release version build first and ask before building with the debug version
+
+# Code changes
+
+- Keep changes minimal and localized to the relevant parts of the code.
+- Preserve the existing naming conventions and coding style used in the surrounding code.
+- Keep code easy to read, well organized, and suitable for future extensibility. A function must not be longer than
+  200/250 lines for readability and cognitive complexity.
+- Prefer clear naming and structure over comments. Add comments only when they materially improve clarity.
+- Do not rename symbols, move files, or restructure modules unless that is necessary for the requested change.
+
+# Working style
+
+- Infer style and conventions from the existing code before introducing new patterns.
+- When several implementation options are possible, prefer the simplest one that fits the current architecture and
+  minimizes churn.
+- Avoid broad refactors unless I explicitly ask for them.
+
+# Responses
+
+- When showing code in chat, make it easy to copy-paste into the codebase.
+- Keep outputs focused on the changed parts.
+- At the end of the response, briefly list any bad practices, mistakes, or cleaner alternatives you noticed, separate
+  from the main solution.
+
+# Guidelines
+
+## 1. Think Before Coding
+
+**Don't assume. Don't hide confusion. Surface tradeoffs.**
+
+Before implementing:
+
+- State your assumptions explicitly. If uncertain, ask.
+- If multiple interpretations exist, present them - don't pick silently.
+- If a simpler approach exists, say so. Push back when warranted.
+- If something is unclear, stop. Name what's confusing. Ask.
+
+## 2. Simplicity First
+
+**Minimum code that solves the problem. Nothing speculative.**
+
+- No features beyond what was asked.
+- No error handling for impossible scenarios.
+- If you write 200 lines and it could be 50, rewrite it.
+
+Ask yourself: "Would a senior engineer say this is overcomplicated?" If yes, simplify.
+
+## 3. Surgical Changes
+
+**Touch only what you must. Clean up only your own mess.**
+
+When editing existing code:
+
+- Don't "improve" adjacent code, comments, or formatting.
+- Don't refactor things that aren't broken.
+- Match existing style, even if you'd do it differently.
+- If you notice unrelated dead code, mention it - don't delete it.
+
+When your changes create orphans:
+
+- Remove imports/variables/functions that YOUR changes made unused.
+- Don't remove pre-existing dead code unless asked, but mention it.
+
+The test: Every changed line should trace directly to the user's request.
+
+## 4. Goal-Driven Execution
+
+**Define success criteria. Loop until verified.**
+
+Transform tasks into verifiable goals:
+
+- "Add validation" → "Write tests for invalid inputs, then make them pass"
+- "Fix the bug" → "Write a test that reproduces it, then make it pass"
+- "Refactor X" → "Ensure tests pass before and after"
+
+For multi-step tasks, state a brief plan:
+
+```
+1. [Step] → verify: [check]
+2. [Step] → verify: [check]
+3. [Step] → verify: [check]
+```
+
+Strong success criteria let you loop independently. Weak criteria ("make it work") require constant clarification.
+
+---

CMakeLists.txt

@@ -3,31 +3,99 @@ cmake_minimum_required(VERSION 3.20.0)
 
 project(raptor)
 
-# Add symlink to PIM as accelerator in onnx-mlir
-function(raptor_ensure_symlink link_path target_path)
-  get_filename_component(link_parent "${link_path}" DIRECTORY)
+# Materialize a CMake shim directory
+function(raptor_write_external_cmake_shim shim_dir external_source_dir description)
+  get_filename_component(real_external_source_dir "${external_source_dir}" REALPATH)
+  file(RELATIVE_PATH relative_external_source_dir "${shim_dir}" "${real_external_source_dir}")
 
-  if(NOT EXISTS "${link_parent}")
-    message(FATAL_ERROR "Directory not found: ${link_parent}")
-  endif()
+  if (NOT EXISTS "${real_external_source_dir}/CMakeLists.txt")
+    message(FATAL_ERROR
+      "External CMake source directory not found or missing CMakeLists.txt:\n"
+      "  ${real_external_source_dir}"
+    )
+  endif ()
+
+  if (IS_SYMLINK "${shim_dir}")
+    message(STATUS "Removing old full-directory symlink: ${shim_dir}")
+    file(REMOVE "${shim_dir}")
+  endif ()
+
+  if (EXISTS "${shim_dir}" AND NOT IS_DIRECTORY "${shim_dir}")
+    message(FATAL_ERROR "Expected directory or absent path, got file: ${shim_dir}")
+  endif ()
+
+  file(MAKE_DIRECTORY "${shim_dir}")
+
+  set(shim_file "${shim_dir}/CMakeLists.txt")
+  set(shim_contents
+    "get_filename_component(raptor_external_source_dir
+    \"\${CMAKE_CURRENT_LIST_DIR}/${relative_external_source_dir}\"
+    REALPATH
+)
+add_subdirectory(
+    \"\${raptor_external_source_dir}\"
+    \"\${CMAKE_CURRENT_BINARY_DIR}/raptor-external\"
+)
+if (DEFINED PIM_ENABLED)
+    set(PIM_ENABLED \"\${PIM_ENABLED}\" PARENT_SCOPE)
+endif ()
+"
+  )
+
+  if (EXISTS "${shim_file}")
+    file(READ "${shim_file}" old_contents)
+  else ()
+    set(old_contents "")
+  endif ()
+
+  if (NOT old_contents STREQUAL shim_contents)
+    file(WRITE "${shim_file}" "${shim_contents}")
+    message(STATUS "Wrote CMake shim for ${description}: ${shim_file}")
+  else ()
+    message(STATUS "CMake shim already up to date for ${description}")
+  endif ()
+
+  # Mirror the external tree's first-level entries into the shim directory
+  # so legacy includes like src/Accelerators/PIM/Compiler/... keep working.
+  file(GLOB children RELATIVE "${real_external_source_dir}" "${real_external_source_dir}/*")
+
+  foreach (child IN LISTS children)
+    if (child STREQUAL "CMakeLists.txt")
+      continue()
+    endif ()
+
+    set(real_child "${real_external_source_dir}/${child}")
+    set(shim_child "${shim_dir}/${child}")
+
+    if (IS_SYMLINK "${shim_child}")
+      file(READ_SYMLINK "${shim_child}" existing_link_target)
+      if (existing_link_target STREQUAL real_child)
+        continue()
+      endif ()
+      file(REMOVE_RECURSE "${shim_child}")
+    elseif (EXISTS "${shim_child}")
+      # Do not delete real files/directories. This protects the generated shim.
+      continue()
+    endif ()
 
-  if(NOT EXISTS "${link_path}")
-    message(STATUS "Creating symlink ${link_path} -> ${target_path}")
     file(CREATE_LINK
-        "${target_path}"
-        "${link_path}"
+      "${real_child}"
+      "${shim_child}"
       SYMBOLIC
     )
-  endif()
+  endforeach ()
 endfunction()
 
-raptor_ensure_symlink(
+raptor_write_external_cmake_shim(
   "${CMAKE_CURRENT_SOURCE_DIR}/onnx-mlir/src/Accelerators/PIM"
   "${CMAKE_CURRENT_SOURCE_DIR}/src/PIM"
+  "PIM accelerator"
 )
-raptor_ensure_symlink(
+
+raptor_write_external_cmake_shim(
   "${CMAKE_CURRENT_SOURCE_DIR}/onnx-mlir/test/accelerators/PIM"
   "${CMAKE_CURRENT_SOURCE_DIR}/test/PIM"
+  "PIM accelerator tests"
 )
 
 # Patch onnx-mlir sources for PIM accelerator support.
@@ -38,21 +106,21 @@ function(raptor_apply_patch file_path anchor replacement description)
 
   # Already applied – replacement text is present
   string(FIND "${contents}" "${replacement}" already_applied_pos)
-  if(NOT already_applied_pos EQUAL -1)
+  if (NOT already_applied_pos EQUAL -1)
     message(STATUS "Patch already applied: ${description}")
     return()
-  endif()
+  endif ()
 
   # Anchor must exist for the patch to be applicable
   string(FIND "${contents}" "${anchor}" anchor_pos)
-  if(anchor_pos EQUAL -1)
+  if (anchor_pos EQUAL -1)
     message(FATAL_ERROR
       "Patch anchor not found – onnx-mlir may have changed.\n"
       "  Patch : ${description}\n"
       "  File  : ${file_path}\n"
       "  Anchor: ${anchor}"
     )
-  endif()
+  endif ()
 
   string(REPLACE "${anchor}" "${replacement}" patched "${contents}")
   file(WRITE "${file_path}" "${patched}")

README.md

@@ -1,226 +1,321 @@
 # Raptor
 
-Raptor is a domain-specific MLIR compiler for neural networks (ONNX format)
-targeting in-memory computing / processing-in-memory (PIM) architectures.
-It progressively lowers ONNX-MLIR through a set of MLIR dialects down to
-target-specific artifacts (currently JSON code for the `pimsim-nn` simulator).
+Raptor is a domain-specific MLIR compiler for neural networks in ONNX format,
+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`.
+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`.
 
 ## Overview
 
-PIM architectures perform most of the computation directly in memory.
-Raptor's first supported target is `pimsim-nn`, which simulates a chip with:
-- a shared host memory,
-- a number of cores that do most of the computation directly in their memory
-  (vector ops, vmm/mvm on ReRAM crossbars),
-- no branching instructions (branchless architecture) and no hardware loop
-  support — any repeated work (e.g. convolutions) must be unrolled into
-  explicit per-iteration instructions.
+PIM architectures perform most computation directly in memory. The supported
+target models a chip with:
+- shared host memory,
+- multiple PIM cores,
+- ReRAM crossbars for vector-matrix / matrix-vector work,
+- explicit communication between cores,
+- no hardware branch or loop support in emitted simulator code.
 
-Because of this, the amount of emitted instructions explodes quickly and the
-compiler must optimize aggressively at every stage to keep compilation
-tractable.
-
-A second target, `PulPim`, is planned for an accelerator with RISC-V cores
-each carrying its own in-memory computing unit and crossbars. It will live in
-a dedicated dialect (future work).
+Because repeated work such as convolutions is eventually made explicit, emitted
+instruction counts can grow quickly. Most compiler work therefore focuses on
+lowering, scheduling, memory layout, and code-generation optimizations.
 
 ### Targets and simulators
 
-`pimsim-nn` (under `backend-simulators/pim/pimsim-nn`) is used for
-**performance** estimates (latency, energy), but does not functionally execute
-the JSON code it consumes. To validate the numerical correctness of the JSON
-code produced by Raptor (or, for comparison, by the `pimcomp` compiler), we use
-a Rust simulator we maintain in-tree at
-`backend-simulators/pim/pim-simulator`.
+- `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 performance simulator submodule.
+  The helper scripts in `pimcomp_utils/` are for comparison with PIMCOMP-NN and
+  contain local paths; treat them as local utilities, not portable workflows.
 
 ## Compilation pipeline
 
-The PIM-related sources live under `src/PIM` and the tests under `test/PIM`.
-When working on this codebase, most changes should stay confined to those
-trees (you only need to look outside, e.g. at `onnx-mlir` or `llvm`, for
-framework-level details).
+The PIM sources live under `src/PIM` and tests under `test/PIM`. CMake exposes
+them to ONNX-MLIR through generated shim directories under
+`onnx-mlir/src/Accelerators/PIM` and `onnx-mlir/test/accelerators/PIM`.
 
 High-level lowering flow:
 
 ```
-ONNX-MLIR ──► Spatial ──► Pim (tensor) ──► Pim (bufferized) ──► PIM JSON
+ONNX-MLIR -> Spatial -> Pim (tensor) -> Pim (bufferized) -> PIM artifacts
 ```
 
-1. **ONNX → Spatial** (`src/PIM/Conversion/ONNXToSpatial`).
-   Lowers ONNX ops into the `spat` dialect (`src/PIM/Dialect/Spatial`).
-   Spatial models a high-level spatial in-memory accelerator: vmm/mvm
-   operations are accelerated by storing a constant RHS matrix into a
-   crossbar. Crossbars cannot be re-programmed during execution, have a
-   limited fixed size, and there is a limited number of them per core.
-   Conversion patterns are split by op family under
-   `Conversion/ONNXToSpatial/Patterns/{Math,NN,Tensor}` (Conv, Gemm, MatMul,
-   Elementwise, ReduceMean, Pool, Relu, Sigmoid, Softmax, Concat, Gather,
-   Reshape, Resize, Split).
+1. **ONNX -> Spatial** (`src/PIM/Conversion/ONNXToSpatial`).
+   Lowers supported ONNX ops into the `spat` dialect
+   (`src/PIM/Dialect/Spatial`). Conversion patterns are split by op family under
+   `Patterns/{Math,NN,Tensor}` and currently cover Conv, Gemm, MatMul,
+   elementwise Add/Mul/Div, ReduceMean, pooling, Relu, Sigmoid, Softmax,
+   Concat, Gather, Reshape, Resize, and Split.
 
-2. **Spatial → Pim** (`src/PIM/Conversion/SpatialToPim`).
-   Lowers Spatial to the `pim` dialect (`src/PIM/Dialect/Pim`), which
-   materializes PIM cores (`pim.core`), inter-core communication
-   (`pim.send` / `pim.receive`), halts, and crossbar-level operations.
+2. **Merge compute nodes**
+   (`src/PIM/Dialect/Spatial/Transforms/MergeComputeNodes`).
+   Builds a compute graph, schedules it with the PEFT scheduler, and materializes
+   the merge schedule into Spatial IR. Supporting scheduling code lives under
+   `MergeComputeNodes/Scheduling`.
 
-3. **Merge compute nodes** (`src/PIM/Dialect/Spatial/Transforms/MergeComputeNodes`).
-   A DCP-inspired heuristic (Dynamic Critical Path — see the original
-   scheduling paper by Kwok & Ahmad,
-   [DCP-eScience2007](https://clouds.cis.unimelb.edu.au/papers/DCP-eScience2007.pdf))
-   that coarsens the virtual node graph and decides how to group compute
-   nodes onto cores. Our implementation is only DCP-*inspired*: it is a
-   heuristic with different assumptions from the paper (different cost
-   model, constraints from crossbar capacity / core resources, and a
-   windowed coarsening loop instead of full-graph reprioritization). The
-   `dcp-critical-window-size` option controls how many lowest-slack virtual
-   nodes each coarsening iteration considers (0 = legacy full-graph
-   analysis). Related sources: `DCPGraph/DCPAnalysis.cpp`, `Graph.cpp/.hpp`,
-   `MergeComputeNodesPass.cpp`.
+3. **Spatial -> Pim** (`src/PIM/Conversion/SpatialToPim`).
+   Lowers Spatial operations to the `pim` dialect (`src/PIM/Dialect/Pim`),
+   including `pim.core`, `pim.core_batch`, communication, tensor packing, global
+   tensor materialization, and return-path normalization.
 
 4. **Bufferization** (`src/PIM/Dialect/Pim/Transforms/Bufferization`).
-   Converts tensor-semantics PIM IR into memref-semantics PIM IR using the
-   standard MLIR `BufferizableOpInterface` machinery
-   (`OpBufferizationInterfaces.*`, `PimBufferization.td`).
+   Converts tensor-semantics PIM IR into memref-semantics PIM IR using MLIR's
+   bufferization interfaces.
 
-5. **Static memory coalescing** (`src/PIM/Dialect/Pim/Transforms/StaticMemoryCoalescing`).
-   Conservatively reuses same-typed local memref allocations inside PIM cores
-   after bufferization and before code generation.
+5. **Static memory coalescing**
+   (`src/PIM/Dialect/Pim/Transforms/StaticMemoryCoalescing`).
+   Reuses compatible local memref allocations inside PIM cores before codegen.
 
-6. **PIM code generation** (`src/PIM/Pass/PimCodegen`):
-   - `HostConstantFolding` — folds host-side constants.
-   - `MaterializeHostConstantsPass` — materializes the remaining host
-     constants for emission.
-   - `VerificationPass` — checks invariants before emission.
-   - `EmitPimJsonPass` — emits the final PIM JSON consumed by `pimsim-nn`
-     and `pim-simulator`.
+6. **PIM code generation** (`src/PIM/Pass/PimCodegen` and
+   `src/PIM/Compiler`).
+   Folds host constants, materializes remaining host constants, verifies PIM IR,
+   emits `.pim` core files, writes weights, and writes `memory.bin` /
+   `config.json`.
 
 Supporting pieces:
-- `src/PIM/Compiler` — PIM-specific compiler options (crossbar size/count,
-  core count, DCP window, experimental conv impl, concat error handling, …)
-  and `PimCodeGen` entry points.
-- `src/PIM/Common` — shared utilities (`PimCommon`, `LabeledList`).
-- `src/PIM/Pass` — auxiliary passes (`MessagePass`, `CountInstructionPass`)
-  and the `PIMPasses.h` registry used by `PimAccelerator`.
-- `src/PIM/PimAccelerator.{cpp,hpp}` — accelerator entry point: registers
-  dialects, passes, and plugs Raptor into the ONNX-MLIR driver.
+- `src/PIM/Common` - shared IR, filesystem, diagnostics, reports, and utility
+  helpers.
+- `src/PIM/Compiler` - PIM compiler options, memory/address planning, binary
+  instruction format, artifact writing, weight emission, and codegen entry
+  points.
+- `src/PIM/Conversion/SpatialToGraphviz` - optional Spatial graphviz conversion
+  pass.
+- `src/PIM/Pass` - pass registration and auxiliary passes.
+- `src/PIM/PimAccelerator.{cpp,hpp}` - ONNX-MLIR accelerator entry point.
 
 ## Key compiler options
 
-Pass these on the `onnx-mlir` command line when compiling for PIM:
+Pass these to `onnx-mlir` when compiling for PIM:
 
-- `--maccel=PIM` — select the PIM accelerator.
-- `--EmitSpatial` / `--EmitPim` / `--EmitPimBufferized` / `--EmitPimCodegen`
-  — stop the pipeline at the requested stage (default: `EmitPimCodegen`).
-- `--pim-only-codegen` — assume the input is already bufferized PIM IR and
-  run only the codegen tail.
-- `--crossbar-size=<N>` / `--crossbar-count=<N>` — crossbar dimensions and
-  per-core count.
-- `--core-count=<N>` — number of cores. Required for PIM compilation.
-- `--pim-merge-scheduler={peft,dcp}` — scheduler used by the Spatial
-  merge-compute-nodes pass (default: `peft`).
-- `--dcp-critical-window-size=<N>` — DCP coarsening window (0 = legacy).
-- `--use-experimental-conv-impl` — alternative convolution lowering.
-- `--ignore-concat-error` — soft-fail corner case in `ConcatOp`.
+- `--maccel=PIM` - select the 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 `2`.
+- `--crossbar-count=<N>` - crossbars per core. Default in code is `256`.
+- `--pim-merge-scheduler=peft` - merge scheduler. `peft` is the only accepted
+  value in the current code.
+- `--pim-only-codegen` - assume input is already bufferized PIM IR and only run
+  the codegen tail.
+- `--pim-emit-json` - also emit `core_*.json` instruction files alongside
+  `core_*.pim`.
+- `--use-experimental-conv-impl` - use the alternate convolution lowering.
+- `--ignore-concat-error` - soft-fail a ConcatOp corner case.
+
+Example:
+
+```bash
+./build_release/Release/bin/onnx-mlir model.onnx -o /tmp/raptor/model \
+  --maccel=PIM --EmitPimCodegen \
+  --crossbar-size=2048 --crossbar-count=256 --core-count=1000
+```
+
+This writes PIM artifacts under `/tmp/raptor/pim/`.
 
 ## Validation
 
-Functional validation lives in `validation/` and drives the Rust
-`pim-simulator` to compare Raptor's output against a reference.
+Functional validation lives in `validation/`. It compiles ONNX models, builds a
+native ONNX-MLIR reference runner, generates random inputs, runs Raptor, runs
+the Rust PIM simulator, and compares outputs.
 
-Per-operation validation (from `validation/`):
+Python dependencies used by the validation scripts are `numpy`, `onnx`, and
+`colorama`. The simulator requires the Rust toolchain.
 
-```
-validate.py \
-    --raptor-path ../cmake-build-release/Release/bin/onnx-mlir \
-    --onnx-include-dir ../onnx-mlir/include \
+Per-operation validation from the repository root:
+
+```bash
+python3 validation/validate.py \
+  --raptor-path build_release/Release/bin/onnx-mlir \
+  --onnx-include-dir onnx-mlir/include \
   --core-count 1000
 ```
 
-End-to-end network validation (example: first 4 layers of YOLOv11n):
+Validate one network or a subset by pointing `--operations-dir` at any directory
+containing `.onnx` files:
 
-```
-validate.py \
-    --raptor-path ../cmake-build-release/Release/bin/onnx-mlir \
-    --onnx-include-dir ../onnx-mlir/include \
-    --operations-dir ./networks/yolo11n/depth_04 \
+```bash
+python3 validation/validate.py \
+  --raptor-path build_release/Release/bin/onnx-mlir \
+  --onnx-include-dir onnx-mlir/include \
+  --operations-dir validation/networks/yolo11n/depth_04 \
   --crossbar-size 2048 --crossbar-count 256 --core-count 1000
 ```
 
-Available networks under `validation/networks/`: `vgg16`, `yolo11n`.
-Available operations under `validation/operations/`: `add`, `conv`, `div`,
-`gather`, `gemm`, `gemv`, `mul`, `pool`, `reduce_mean`, `relu`, `resize`,
-`sigmoid`, `softmax`, `split`.
+Useful validation options:
+- `--simulator-dir <path>` - override the auto-detected
+  `backend-simulators/pim/pim-simulator` path.
+- `--threshold <float>` - maximum allowed per-element output difference.
+- `--seed <int>` - RNG seed for generated inputs.
+- `--command-timeout-seconds <float>` - timeout for compiler, runner, and
+  simulator subprocesses.
+- `--verbose` - print subprocess logs and average PIM pass timings.
+- `--clean` - remove generated validation artifacts and exit.
 
-## Rebuilding
+Each validation run writes artifacts in the model workspace, for example under
+`validation/operations/gemm/small/`:
+- `inputs/` - generated input CSV files.
+- `outputs/` - native ONNX-MLIR reference outputs.
+- `raptor/` - compiler artifacts, including `*.onnx.mlir`, dialect dumps under
+  `dialects/`, reports under `reports/`, and final PIM artifacts under `pim/`.
+- `runner/` - generated reference runner source, build tree, and shared library.
+- `simulation/out.bin` - raw simulator output used for comparison.
 
-Release build (fast):
+The compiler currently dumps dialect snapshots such as `spatial0.mlir`,
+`spatial1_dcp_merged.mlir`, `pim0.mlir`, `pim1_buff.mlir`,
+`pim2_coalesced.mlir`, `pim3_folded.mlir`, and
+`pim4_materialized.mlir` when an output directory is available.
 
-```
-cmake --build /home/nico/raptor/raptor/cmake-build-release --target onnx-mlir -j 30
+To rerun the simulator manually with tracing after validation has produced a
+`raptor/pim/` 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 \
+  -d <addr0>,<size0>,<addr1>,<size1>,...
 ```
 
-A slower debug build is also available — configure it the same way but with
-`-DCMAKE_BUILD_TYPE=Debug` (see installation instructions below).
+With `--features tracing`, the simulator writes per-core traces as
+`TraceCore0`, `TraceCore1`, ... next to `out.bin`. The validator normally
+computes the `-d` ranges from `raptor/pim/config.json` and model output shapes.
+
+Available validation networks under `validation/networks/`: `vgg16`,
+`yolo11n`, `yolo11nv2`.
+
+Available operation suites under `validation/operations/`: `add`, `concat`,
+`conv`, `div`, `gather`, `gemm`, `gemv`, `matmul`, `mul`, `pool`,
+`reduce_mean`, `relu`, `reshape`, `resize`, `sigmoid`, `softmax`, `split`.
+
+Generated operation tests can be regenerated with:
+
+```bash
+python3 validation/operations/gen_tests.py
+```
 
 ## Build
 
+Initialize submodules first:
+
+```bash
+git submodule update --init --recursive
+```
+
+The project follows ONNX-MLIR's build requirements. The CI workflow documents
+the currently used versions and setup:
+- CMake 4.3.0 in CI,
+- LLVM/MLIR checked out under `onnx-mlir/llvm-project`,
+- Protobuf `v34.0`,
+- Rust stable for `pim-simulator`,
+- Python packages `numpy`, `onnx`, `colorama` for validation.
+
 ### Protobuf
 
-Use the following commands to install protobuf:
-```
+Install Protobuf if your system does not already provide a compatible version:
+
+```bash
 git clone --depth 1 --branch v34.0 https://github.com/protocolbuffers/protobuf
-cd protobuf
-mkdir build
-cd build
-cmake .. -G Ninja -DCMAKE_BUILD_TYPE=Release
-ninja
-sudo ninja install
+cmake -S protobuf -B protobuf/build -G Ninja \
+  -DCMAKE_BUILD_TYPE=Release \
+  -Dprotobuf_BUILD_TESTS=OFF
+cmake --build protobuf/build
+sudo cmake --install protobuf/build
 ```
 
-You can now remove the protobuf repo directory with:
-```
-cd ../..
+You can then remove the temporary checkout:
+
+```bash
 rm -rf protobuf
 ```
 
-### Mlir
+### MLIR
 
-Follow the first part of instructions [here](onnx-mlir/docs/BuildOnLinuxOSX.md) to build mlir.
+Follow the ONNX-MLIR instructions in
+`onnx-mlir/docs/BuildOnLinuxOSX.md` to build LLVM/MLIR. The local Raptor build
+expects `MLIR_DIR` to point at the MLIR CMake package, for example:
 
-Remember to set ```-DCMAKE_BUILD_TYPE=Debug``` for developing on Raptor
-
-Moreover, if compiling with build type debug, it is also suggested to use
-mold as linker (you will need to install it if you don't have it already)
-to reduce memory usage during linking. You can use it by setting the options:
-```
--DLLVM_USE_LINKER=mold
+```bash
+MLIR_DIR=$(pwd)/onnx-mlir/llvm-project/build_release/lib/cmake/mlir
 ```
 
+If your LLVM build directory is named `build` instead of `build_release`, adjust
+the path accordingly.
+
 ### Raptor
 
-Use the following commands to build Raptor.
+Configure a release build:
 
-Remember to set ```-DCMAKE_BUILD_TYPE=Debug``` for developing on Raptor.
-
-Also in this case, it is suggested to use mold as linker to reduce link time and memory usage,
-setting the options:
-```
--DCMAKE_EXE_LINKER_FLAGS="-fuse-ld=mold" \
--DCMAKE_SHARED_LINKER_FLAGS="-fuse-ld=mold" \
--DCMAKE_MODULE_LINKER_FLAGS="-fuse-ld=mold"
-```
-
-```
-git submodule update --init --recursive
-
-MLIR_DIR=$(pwd)/onnx-mlir/llvm-project/build/lib/cmake/mlir
-mkdir build && cd build
-cmake .. -G Ninja \
+```bash
+MLIR_DIR=$(pwd)/onnx-mlir/llvm-project/build_release/lib/cmake/mlir
+cmake -S . -B build_release -G Ninja \
   -DCMAKE_BUILD_TYPE=Release \
   -DONNX_MLIR_ACCELERATORS=PIM \
   -DLLVM_ENABLE_ASSERTIONS=ON \
   -DMLIR_DIR=${MLIR_DIR}
-cmake --build .
 ```
 
-If the build fails because of protobuf missing uint definitions,
-just patch the problematic files by adding ```#include <cstdint>``` to their includes.
+Configure a debug build similarly:
+
+```bash
+MLIR_DIR=$(pwd)/onnx-mlir/llvm-project/build_debug/lib/cmake/mlir
+cmake -S . -B build_debug -G Ninja \
+  -DCMAKE_BUILD_TYPE=Debug \
+  -DONNX_MLIR_ACCELERATORS=PIM \
+  -DLLVM_ENABLE_ASSERTIONS=ON \
+  -DMLIR_DIR=${MLIR_DIR}
+```
+
+For debug development, using `mold` can reduce link time and memory use:
+
+```bash
+cmake -S . -B build_debug -G Ninja \
+  -DCMAKE_BUILD_TYPE=Debug \
+  -DONNX_MLIR_ACCELERATORS=PIM \
+  -DLLVM_ENABLE_ASSERTIONS=ON \
+  -DMLIR_DIR=${MLIR_DIR} \
+  -DCMAKE_EXE_LINKER_FLAGS="-fuse-ld=mold" \
+  -DCMAKE_SHARED_LINKER_FLAGS="-fuse-ld=mold" \
+  -DCMAKE_MODULE_LINKER_FLAGS="-fuse-ld=mold"
+```
+
+Build the compiler with CMake:
+
+```bash
+cmake --build ./build_release
+cmake --build ./build_debug
+```
+
+Do not invoke `ninja` directly for this project; use `cmake --build` so CMake's
+configuration and generated shims stay consistent.
+
+If a build fails because Protobuf headers are missing fixed-width integer
+definitions, patch the affected Protobuf-generated files by adding
+`#include <cstdint>`.
+
+## Tests
+
+The Rust simulator has its own tests:
+
+```bash
+cd backend-simulators/pim/pim-simulator
+cargo test
+```
+
+## Repository Layout
+
+- `src/PIM/` - PIM accelerator implementation.
+- `test/PIM/` - PIM C++ unit tests.
+- `validation/` - functional validation scripts, ONNX operation tests, network
+  slices, and pimsim config generation.
+- `backend-simulators/pim/pim-simulator/` - in-tree Rust functional simulator.
+- `backend-simulators/pim/pimsim-nn/` - performance simulator submodule.
+- `pimcomp_utils/` - local comparison helpers for PIMCOMP-NN.
+- `.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

@@ -43,7 +43,7 @@ struct Args {
 
     /// Comma separated list of (address,size) for memory output dump
     #[arg(short, long, value_delimiter = ',', num_args = 1.., value_name = "ADDR,SIZE")]
-    dump: Vec<i32>,
+    dump: Vec<usize>,
 }
 
 fn main() -> Result<()> {
@@ -168,7 +168,7 @@ fn get_crossbars(config: &Value, args: &Args) -> anyhow::Result<HashMap<String,
 }
 
 fn dump_memory(mut executor: pimcore::Executable, args: &Args) -> Result<()> {
-    let dumps: Vec<(i32, i32)> = args
+    let dumps: Vec<(usize, usize)> = args
         .dump
         .chunks_exact(2)
         .map(|chunk| (chunk[0], chunk[1]))

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

@@ -1,3 +1,4 @@
+use crate::utility::AddressArg;
 use std::{collections::HashMap, fmt::Debug};
 use anyhow::{Context, Result, ensure};
 
@@ -9,6 +10,7 @@ use crate::{
 
 pub mod crossbar;
 
+
 #[derive(Debug, Clone)]
 pub struct CPU<'a> {
     cores: Box<[Core<'a>]>,
@@ -91,30 +93,26 @@ impl<'a> Core<'a> {
         self.memory.execute_load()
     }
 
-    pub fn execute_store<T>(&mut self, address: impl TryToUsize, element: &[T]) -> Result<()>
+    pub fn execute_store<T>(&mut self, address: impl AddressArg, element: &[T]) -> Result<()>
     where
         T: MemoryStorable,
     {
-        let address = address.try_into().context("address can not be negative")?;
+        let address = address.to_address_usize()?;
         self.memory.execute_store(address, element)
     }
 
     pub fn reserve_load(
         &mut self,
-        address: impl TryToUsize,
+        address: impl AddressArg,
         size: impl TryToUsize,
     ) -> Result<&mut CoreMemory> {
-        let address = address.try_into().context("address can not be negative")?;
+        let address = address.to_address_usize()?;
         let size = size.try_into().context("size can not be negative")?;
         self.memory.reserve_load(address, size)
     }
 
     pub fn set_register(&mut self, index: impl TryToUsize, value: i32) {
         let index = index.try_into().expect("index can not be negative");
-        assert!(
-            value >= 0,
-            "Register cannot be negative if happens remove this and go check where it's used as usize"
-        );
         self.registers[index] = value;
     }
 
@@ -123,11 +121,11 @@ impl<'a> Core<'a> {
         self.registers[index]
     }
 
-    pub fn load<T>(&mut self, address: impl TryToUsize, size: impl TryToUsize) -> Result<Vec<&[T]>>
+    pub fn load<T>(&mut self, address: impl AddressArg, size: impl TryToUsize) -> Result<Vec<&[T]>>
     where
         T: MemoryStorable,
     {
-        let address = address.try_into().context("address can not be negative")?;
+        let address = address.to_address_usize()?;
         let size = size.try_into().context("size can not be negative")?;
         self.memory.load(address, size)
     }
@@ -141,8 +139,8 @@ impl<'a> Core<'a> {
         (memory, crossbars)
     }
 
-    pub fn memset(&mut self, address: impl TryToUsize, size: impl TryToUsize, val: u8) -> Result<()> {
-        let address = address.try_into().context("address can not be negative")?;
+    pub fn memset(&mut self, address: impl AddressArg, size: impl TryToUsize, val: u8) -> Result<()> {
+        let address = address.to_address_usize()?;
         let size = size.try_into().context("size can not be negative")?;
         self.memory.memset(address, size, val)
     }

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

@@ -299,10 +299,11 @@ fn detect_deadlock(cores_instructions: &[CoreInstructions]) -> Option<DeadlockIn
             if in_path.contains(&waiting_for) {
                 let cycle_start = path.iter().position(|&c| c == waiting_for).unwrap();
                 let cycle = &path[cycle_start..];
+                let format_core = |core: &i32| (core - 1).to_string();
 
                 let cycle_str = cycle
                     .iter()
-                    .map(|c| c.to_string())
+                    .map(format_core)
                     .collect::<Vec<_>>()
                     .join(" -> ");
 
@@ -311,19 +312,19 @@ fn detect_deadlock(cores_instructions: &[CoreInstructions]) -> Option<DeadlockIn
                     .copied()
                     .chain(std::iter::once(waiting_for))
                     .collect::<Vec<_>>();
-                let cycle_msg = format!("{} -> {}", cycle_str, waiting_for);
+                let cycle_msg = format!("{} -> {}", cycle_str, waiting_for - 1);
                 let states_msg = cycle
                     .iter()
                     .filter_map(|core| {
                         states.get(core).map(|state| match state {
                             CoreState::SendingTo(target, size) => {
-                                format!("core {} send {}B -> {}", core, size, target)
+                                format!("core {} send {}B -> {}", core - 1, size, target - 1)
                             }
                             CoreState::ReceivingFrom(source, size) => {
-                                format!("core {} recv {}B <- {}", core, size, source)
+                                format!("core {} recv {}B <- {}", core - 1, size, source - 1)
                             }
-                            CoreState::Working => format!("core {} working", core),
-                            CoreState::Halted => format!("core {} halted", core),
+                            CoreState::Working => format!("core {} working", core - 1),
+                            CoreState::Halted => format!("core {} halted", core - 1),
                         })
                     })
                     .collect::<Vec<_>>()

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

@@ -1,7 +1,45 @@
+use anyhow::{Result,Context};
 use std::{fmt::Debug, mem::transmute};
 
-use crate::memory_manager::type_traits::TryToUsize;
 
+pub trait AddressArg {
+    fn to_address_usize(self) -> Result<usize>;
+}
+
+impl AddressArg for usize {
+    fn to_address_usize(self) -> Result<usize> {
+        Ok(self)
+    }
+}
+
+impl AddressArg for u32 {
+    fn to_address_usize(self) -> Result<usize> {
+        Ok(self as usize)
+    }
+}
+
+impl AddressArg for u64 {
+    fn to_address_usize(self) -> Result<usize> {
+        usize::try_from(self).context("address does not fit in usize")
+    }
+}
+
+impl AddressArg for i32 {
+    fn to_address_usize(self) -> Result<usize> {
+        Ok(self as u32 as usize)
+    }
+}
+
+impl AddressArg for i64 {
+    fn to_address_usize(self) -> Result<usize> {
+        usize::try_from(self).context("address can not be negative")
+    }
+}
+
+
+fn address_to_usize(address: i32) -> usize {
+    address as u32 as usize
+}
 
 fn add_offset_impl(address: usize, offset_select : i32, offset_value : i32, id:i32) -> usize{
      assert!(offset_select == 1 || offset_select == 2 || offset_select == 4 || offset_value == 0, "offset_select not a bit field");
@@ -14,21 +52,21 @@ fn add_offset_impl(address: usize, offset_select : i32, offset_value : i32, id:i
 }
 
 
-pub fn add_offset_rd(address: impl TryToUsize, offset_select : i32, offset_value : i32) -> usize
+pub fn add_offset_rd(address: i32, offset_select : i32, offset_value : i32) -> usize
 {
-    let address = address.try_into().expect("address can not be negative");
+    let address = address_to_usize(address);
     add_offset_impl(address, offset_select, offset_value, 4)
 }
 
-pub fn add_offset_r1(address: impl TryToUsize, offset_select : i32, offset_value : i32) -> usize
+pub fn add_offset_r1(address: i32, offset_select : i32, offset_value : i32) -> usize
 {
-    let address = address.try_into().expect("address can not be negative");
+    let address = address_to_usize(address);
     add_offset_impl(address, offset_select, offset_value, 1)
 }
 
-pub fn add_offset_r2(address: impl TryToUsize, offset_select : i32, offset_value : i32) -> usize
+pub fn add_offset_r2(address: i32, offset_select : i32, offset_value : i32) -> usize
 {
-    let address = address.try_into().expect("address can not be negative");
+    let address = address_to_usize(address);
     add_offset_impl(address, offset_select, offset_value, 2)
 }
 

src/PIM/CMakeLists.txt

@@ -10,6 +10,56 @@ set(PIM_INCLUDE_PATH ${CMAKE_INCLUDE_OUTPUT_DIRECTORY})
 set(PIM_ONNX_MLIR_SRC_ROOT ${ONNX_MLIR_SRC_ROOT})
 set(PIM_ONNX_MLIR_BIN_ROOT ${ONNX_MLIR_BIN_ROOT})
 
+set(PIM_GENERATED_PATH_SHIM_TARGET "")
+get_filename_component(PIM_BIN_ROOT_NAME "${PIM_BIN_ROOT}" NAME)
+if (PIM_BIN_ROOT_NAME STREQUAL "raptor-external")
+  get_filename_component(PIM_GENERATED_PATH_SHIM_ROOT "${PIM_BIN_ROOT}" DIRECTORY)
+  set(PIM_GENERATED_PATH_SHIM_OUTPUTS)
+
+  function(add_pim_generated_path_shim relative_path)
+    set(real_file "${PIM_BIN_ROOT}/${relative_path}")
+    set(shim_file "${PIM_GENERATED_PATH_SHIM_ROOT}/${relative_path}")
+    get_filename_component(shim_dir "${shim_file}" DIRECTORY)
+
+    add_custom_command(
+      OUTPUT "${shim_file}"
+      DEPENDS "${real_file}"
+      COMMAND "${CMAKE_COMMAND}" -E make_directory "${shim_dir}"
+      COMMAND "${CMAKE_COMMAND}" -E rm -f "${shim_file}"
+      COMMAND "${CMAKE_COMMAND}" -E create_symlink "${real_file}" "${shim_file}"
+      VERBATIM
+    )
+
+    list(APPEND PIM_GENERATED_PATH_SHIM_OUTPUTS "${shim_file}")
+    set(PIM_GENERATED_PATH_SHIM_OUTPUTS "${PIM_GENERATED_PATH_SHIM_OUTPUTS}" PARENT_SCOPE)
+  endfunction()
+
+  file(GLOB_RECURSE pim_generated_path_scan_sources
+    CONFIGURE_DEPENDS
+    "${PIM_SRC_ROOT}/*.cpp"
+    "${PIM_SRC_ROOT}/*.hpp"
+  )
+
+  set(pim_generated_path_shims)
+  foreach (source_file IN LISTS pim_generated_path_scan_sources)
+    file(READ "${source_file}" source_contents)
+    string(REGEX MATCHALL "#include \"src/Accelerators/PIM/[^\"]+\\.inc\"" source_inc_matches "${source_contents}")
+
+    foreach (inc_match IN LISTS source_inc_matches)
+      string(REGEX REPLACE "^#include \"src/Accelerators/PIM/(.+)\"$" "\\1" relative_inc_path "${inc_match}")
+      list(APPEND pim_generated_path_shims "${relative_inc_path}")
+    endforeach ()
+  endforeach ()
+
+  list(REMOVE_DUPLICATES pim_generated_path_shims)
+  foreach (relative_inc_path IN LISTS pim_generated_path_shims)
+    add_pim_generated_path_shim("${relative_inc_path}")
+  endforeach ()
+
+  add_custom_target(OMPimGeneratedPathShims DEPENDS ${PIM_GENERATED_PATH_SHIM_OUTPUTS})
+  set(PIM_GENERATED_PATH_SHIM_TARGET OMPimGeneratedPathShims)
+endif ()
+
 set(PIM_PUBLIC_INCLUDE_DIRS
   ${ONNX_MLIR_SRC_ROOT}/include
   ${ONNX_MLIR_SRC_ROOT}
@@ -37,6 +87,9 @@ set(PIM_GENERATED_INCLUDE_DIRS
 
 function(add_pim_library name)
   add_onnx_mlir_library(${name} STATIC ${ARGN})
+  if (PIM_GENERATED_PATH_SHIM_TARGET)
+    add_dependencies(${name} ${PIM_GENERATED_PATH_SHIM_TARGET})
+  endif ()
 endfunction()
 
 add_subdirectory(Dialect)

src/PIM/Common/CMakeLists.txt

@@ -1,5 +1,8 @@
 add_pim_library(OMPimCommon
+  IR/AffineUtils.cpp
   IR/AddressAnalysis.cpp
+  IR/BatchCoreUtils.cpp
+  IR/ConstantUtils.cpp
   IR/CoreBlockUtils.cpp
   IR/EntryPointUtils.cpp
   IR/ShapeUtils.cpp
@@ -16,6 +19,7 @@ add_pim_library(OMPimCommon
   ${PIM_PUBLIC_INCLUDE_DIRS}
 
   LINK_LIBS PUBLIC
+  MLIRLinalgDialect
   onnx
   SpatialOps
   PimOps

src/PIM/Common/IR/AddressAnalysis.cpp

@@ -1,7 +1,11 @@
 #include "mlir/Dialect/Arith/IR/Arith.h"
+#include "mlir/Dialect/MemRef/IR/MemRef.h"
 #include "mlir/Dialect/SCF/IR/SCF.h"
+#include "mlir/IR/BuiltinAttributes.h"
 #include "mlir/Interfaces/DestinationStyleOpInterface.h"
 
+#include <limits>
+
 #include "src/Accelerators/PIM/Common/IR/AddressAnalysis.hpp"
 #include "src/Accelerators/PIM/Common/IR/ShapeUtils.hpp"
 #include "src/Accelerators/PIM/Dialect/Pim/PimOps.hpp"
@@ -28,6 +32,14 @@ mlir::Value resolveAlias(mlir::Value value, const StaticValueKnowledge* knowledg
   return value;
 }
 
+llvm::FailureOr<CompiledIndexExpr> compileIndexValueImpl(mlir::Value value);
+llvm::FailureOr<CompiledAddressExpr> compileContiguousAddressExprImpl(mlir::Value value);
+
+template <typename... Args>
+CompiledIndexExpr makeCompiledIndexExpr(Args&&... args) {
+  return CompiledIndexExpr(std::make_shared<CompiledIndexExprNode>(std::forward<Args>(args)...));
+}
+
 mlir::Value resolveLoopCarriedAliasImpl(mlir::Value value, const StaticValueKnowledge* knowledge) {
   value = resolveAlias(value, knowledge);
 
@@ -55,6 +67,288 @@ mlir::Value resolveLoopCarriedAliasImpl(mlir::Value value, const StaticValueKnow
 }
 
 llvm::FailureOr<int64_t> resolveOpFoldResult(mlir::OpFoldResult ofr, const StaticValueKnowledge* knowledge);
+llvm::FailureOr<int64_t> resolveIndexValueImpl(mlir::Value value, const StaticValueKnowledge* knowledge);
+
+static llvm::FailureOr<llvm::SmallVector<int64_t>> getStaticMemRefStrides(mlir::MemRefType type) {
+  llvm::SmallVector<int64_t> strides;
+  int64_t offset = 0;
+  if (failed(type.getStridesAndOffset(strides, offset)))
+    return mlir::failure();
+  if (llvm::any_of(strides, mlir::ShapedType::isDynamic))
+    return mlir::failure();
+  return strides;
+}
+
+static llvm::FailureOr<int64_t> resolveConstantGlobalLoad(mlir::memref::LoadOp loadOp,
+                                                          const StaticValueKnowledge* knowledge) {
+  auto getGlobalOp = loadOp.getMemRef().getDefiningOp<mlir::memref::GetGlobalOp>();
+  if (!getGlobalOp)
+    return mlir::failure();
+
+  auto moduleOp = loadOp->getParentOfType<mlir::ModuleOp>();
+  auto globalOp = lookupGlobalForGetGlobal(moduleOp, getGlobalOp);
+  if (!globalOp || !globalOp.getConstant() || !globalOp.getInitialValue())
+    return mlir::failure();
+
+  auto denseAttr = mlir::dyn_cast<mlir::DenseElementsAttr>(*globalOp.getInitialValue());
+  auto globalType = mlir::dyn_cast<mlir::MemRefType>(getGlobalOp.getType());
+  if (!denseAttr || !globalType || !globalType.hasStaticShape())
+    return mlir::failure();
+
+  auto elementType = denseAttr.getElementType();
+  if (!elementType.isIndex() && !elementType.isInteger())
+    return mlir::failure();
+
+  llvm::SmallVector<int64_t> indices;
+  indices.reserve(loadOp.getIndices().size());
+  for (mlir::Value index : loadOp.getIndices()) {
+    auto resolvedIndex = resolveIndexValueImpl(index, knowledge);
+    if (failed(resolvedIndex))
+      return mlir::failure();
+    indices.push_back(*resolvedIndex);
+  }
+
+  if (indices.size() != static_cast<size_t>(globalType.getRank()))
+    return mlir::failure();
+
+  auto strides = computeRowMajorStrides(globalType.getShape());
+  int64_t linearIndex = linearizeIndex(indices, strides);
+  if (linearIndex < 0 || linearIndex >= globalType.getNumElements())
+    return mlir::failure();
+
+  return denseAttr.getValues<llvm::APInt>()[linearIndex].getSExtValue();
+}
+
+static bool evaluateCmpPredicate(mlir::arith::CmpIPredicate predicate, int64_t lhs, int64_t rhs) {
+  switch (predicate) {
+  case mlir::arith::CmpIPredicate::eq:  return lhs == rhs;
+  case mlir::arith::CmpIPredicate::ne:  return lhs != rhs;
+  case mlir::arith::CmpIPredicate::slt: return lhs < rhs;
+  case mlir::arith::CmpIPredicate::sle: return lhs <= rhs;
+  case mlir::arith::CmpIPredicate::sgt: return lhs > rhs;
+  case mlir::arith::CmpIPredicate::sge: return lhs >= rhs;
+  case mlir::arith::CmpIPredicate::ult: return static_cast<uint64_t>(lhs) < static_cast<uint64_t>(rhs);
+  case mlir::arith::CmpIPredicate::ule: return static_cast<uint64_t>(lhs) <= static_cast<uint64_t>(rhs);
+  case mlir::arith::CmpIPredicate::ugt: return static_cast<uint64_t>(lhs) > static_cast<uint64_t>(rhs);
+  case mlir::arith::CmpIPredicate::uge: return static_cast<uint64_t>(lhs) >= static_cast<uint64_t>(rhs);
+  }
+
+  llvm_unreachable("unknown cmpi predicate");
+}
+
+llvm::FailureOr<int64_t> evaluateCompiledIndexExpr(const CompiledIndexExpr& expr,
+                                                   const StaticValueKnowledge& knowledge) {
+  if (!expr.node)
+    return mlir::failure();
+
+  switch (expr.node->kind) {
+  case CompiledIndexExprNode::Kind::Constant: return expr.node->constant;
+  case CompiledIndexExprNode::Kind::Symbol:   {
+    auto value = resolveAlias(expr.node->symbol, &knowledge);
+    auto iter = knowledge.indexValues.find(value);
+    if (iter != knowledge.indexValues.end())
+      return iter->second;
+    return mlir::failure();
+  }
+  case CompiledIndexExprNode::Kind::Add:
+  case CompiledIndexExprNode::Kind::Sub:
+  case CompiledIndexExprNode::Kind::Mul:
+  case CompiledIndexExprNode::Kind::DivUI:
+  case CompiledIndexExprNode::Kind::DivSI:
+  case CompiledIndexExprNode::Kind::RemUI:
+  case CompiledIndexExprNode::Kind::RemSI:
+  case CompiledIndexExprNode::Kind::MinUI:
+  case CompiledIndexExprNode::Kind::CmpI:  {
+    auto lhs = evaluateCompiledIndexExpr(expr.node->operands[0], knowledge);
+    auto rhs = evaluateCompiledIndexExpr(expr.node->operands[1], knowledge);
+    if (failed(lhs) || failed(rhs))
+      return mlir::failure();
+
+    switch (expr.node->kind) {
+    case CompiledIndexExprNode::Kind::Add: return *lhs + *rhs;
+    case CompiledIndexExprNode::Kind::Sub: return *lhs - *rhs;
+    case CompiledIndexExprNode::Kind::Mul: return *lhs * *rhs;
+    case CompiledIndexExprNode::Kind::DivUI:
+      if (*rhs == 0)
+        return mlir::failure();
+      return static_cast<int64_t>(static_cast<uint64_t>(*lhs) / static_cast<uint64_t>(*rhs));
+    case CompiledIndexExprNode::Kind::DivSI:
+      if (*rhs == 0 || (*lhs == std::numeric_limits<int64_t>::min() && *rhs == -1))
+        return mlir::failure();
+      return *lhs / *rhs;
+    case CompiledIndexExprNode::Kind::RemUI:
+      if (*rhs == 0)
+        return mlir::failure();
+      return static_cast<int64_t>(static_cast<uint64_t>(*lhs) % static_cast<uint64_t>(*rhs));
+    case CompiledIndexExprNode::Kind::RemSI:
+      if (*rhs == 0)
+        return mlir::failure();
+      if (*lhs == std::numeric_limits<int64_t>::min() && *rhs == -1)
+        return 0;
+      return *lhs % *rhs;
+    case CompiledIndexExprNode::Kind::MinUI:
+      return static_cast<int64_t>(std::min(static_cast<uint64_t>(*lhs), static_cast<uint64_t>(*rhs)));
+    case CompiledIndexExprNode::Kind::CmpI: return evaluateCmpPredicate(expr.node->predicate, *lhs, *rhs) ? 1 : 0;
+    default:                                llvm_unreachable("unexpected binary compiled index kind");
+    }
+  }
+  case CompiledIndexExprNode::Kind::Select: {
+    auto condition = evaluateCompiledIndexExpr(expr.node->operands[0], knowledge);
+    if (failed(condition))
+      return mlir::failure();
+    return evaluateCompiledIndexExpr(*condition != 0 ? expr.node->operands[1] : expr.node->operands[2], knowledge);
+  }
+  case CompiledIndexExprNode::Kind::ConstantGlobalLoad: {
+    if (!expr.node->globalOp || !expr.node->globalOp.getInitialValue())
+      return mlir::failure();
+
+    auto denseAttr = mlir::dyn_cast<mlir::DenseElementsAttr>(*expr.node->globalOp.getInitialValue());
+    auto globalType = mlir::dyn_cast<mlir::MemRefType>(expr.node->globalOp.getType());
+    if (!denseAttr || !globalType)
+      return mlir::failure();
+
+    llvm::SmallVector<int64_t> indices;
+    indices.reserve(expr.node->operands.size());
+    for (const CompiledIndexExpr& operand : expr.node->operands) {
+      auto resolvedIndex = evaluateCompiledIndexExpr(operand, knowledge);
+      if (failed(resolvedIndex))
+        return mlir::failure();
+      indices.push_back(*resolvedIndex);
+    }
+
+    int64_t linearIndex = linearizeIndex(indices, expr.node->globalStrides);
+    if (linearIndex < 0 || linearIndex >= globalType.getNumElements())
+      return mlir::failure();
+    return denseAttr.getValues<llvm::APInt>()[linearIndex].getSExtValue();
+  }
+  }
+
+  llvm_unreachable("unknown compiled index kind");
+}
+
+llvm::FailureOr<CompiledIndexExpr> compileConstantGlobalLoad(mlir::memref::LoadOp loadOp) {
+  auto getGlobalOp = loadOp.getMemRef().getDefiningOp<mlir::memref::GetGlobalOp>();
+  if (!getGlobalOp)
+    return mlir::failure();
+
+  auto moduleOp = loadOp->getParentOfType<mlir::ModuleOp>();
+  auto globalOp = lookupGlobalForGetGlobal(moduleOp, getGlobalOp);
+  if (!globalOp || !globalOp.getConstant() || !globalOp.getInitialValue())
+    return mlir::failure();
+
+  auto denseAttr = mlir::dyn_cast<mlir::DenseElementsAttr>(*globalOp.getInitialValue());
+  auto globalType = mlir::dyn_cast<mlir::MemRefType>(getGlobalOp.getType());
+  if (!denseAttr || !globalType || !globalType.hasStaticShape())
+    return mlir::failure();
+
+  auto elementType = denseAttr.getElementType();
+  if (!elementType.isIndex() && !elementType.isInteger())
+    return mlir::failure();
+
+  CompiledIndexExprNode expr;
+  expr.kind = CompiledIndexExprNode::Kind::ConstantGlobalLoad;
+  expr.globalOp = globalOp;
+  expr.globalStrides = computeRowMajorStrides(globalType.getShape());
+  expr.operands.reserve(loadOp.getIndices().size());
+  for (mlir::Value index : loadOp.getIndices()) {
+    auto compiledIndex = compileIndexValueImpl(index);
+    if (failed(compiledIndex))
+      return mlir::failure();
+    expr.operands.push_back(*compiledIndex);
+  }
+  return makeCompiledIndexExpr(std::move(expr));
+}
+
+llvm::FailureOr<CompiledIndexExpr> compileIndexValueImpl(mlir::Value value) {
+  if (auto constantOp = value.getDefiningOp<mlir::arith::ConstantOp>()) {
+    if (auto integerAttr = mlir::dyn_cast<mlir::IntegerAttr>(constantOp.getValue())) {
+      CompiledIndexExprNode expr;
+      expr.kind = CompiledIndexExprNode::Kind::Constant;
+      expr.constant = integerAttr.getInt();
+      return makeCompiledIndexExpr(std::move(expr));
+    }
+  }
+
+  mlir::Operation* definingOp = value.getDefiningOp();
+  if (!definingOp) {
+    CompiledIndexExprNode expr;
+    expr.kind = CompiledIndexExprNode::Kind::Symbol;
+    expr.symbol = value;
+    return makeCompiledIndexExpr(std::move(expr));
+  }
+
+  auto buildBinaryExpr = [&](CompiledIndexExprNode::Kind kind, mlir::Value lhsValue, mlir::Value rhsValue) {
+    auto lhs = compileIndexValueImpl(lhsValue);
+    auto rhs = compileIndexValueImpl(rhsValue);
+    if (failed(lhs) || failed(rhs))
+      return llvm::FailureOr<CompiledIndexExpr>(mlir::failure());
+    CompiledIndexExprNode expr;
+    expr.kind = kind;
+    expr.operands = {*lhs, *rhs};
+    return llvm::FailureOr<CompiledIndexExpr>(makeCompiledIndexExpr(std::move(expr)));
+  };
+
+  if (auto indexCastOp = mlir::dyn_cast<mlir::arith::IndexCastOp>(definingOp))
+    return compileIndexValueImpl(indexCastOp.getIn());
+  if (auto addOp = mlir::dyn_cast<mlir::arith::AddIOp>(definingOp))
+    return buildBinaryExpr(CompiledIndexExprNode::Kind::Add, addOp.getLhs(), addOp.getRhs());
+  if (auto subOp = mlir::dyn_cast<mlir::arith::SubIOp>(definingOp))
+    return buildBinaryExpr(CompiledIndexExprNode::Kind::Sub, subOp.getLhs(), subOp.getRhs());
+  if (auto mulOp = mlir::dyn_cast<mlir::arith::MulIOp>(definingOp))
+    return buildBinaryExpr(CompiledIndexExprNode::Kind::Mul, mulOp.getLhs(), mulOp.getRhs());
+  if (auto divOp = mlir::dyn_cast<mlir::arith::DivUIOp>(definingOp))
+    return buildBinaryExpr(CompiledIndexExprNode::Kind::DivUI, divOp.getLhs(), divOp.getRhs());
+  if (auto divOp = mlir::dyn_cast<mlir::arith::DivSIOp>(definingOp))
+    return buildBinaryExpr(CompiledIndexExprNode::Kind::DivSI, divOp.getLhs(), divOp.getRhs());
+  if (auto minOp = mlir::dyn_cast<mlir::arith::MinUIOp>(definingOp))
+    return buildBinaryExpr(CompiledIndexExprNode::Kind::MinUI, minOp.getLhs(), minOp.getRhs());
+  if (auto remOp = mlir::dyn_cast<mlir::arith::RemUIOp>(definingOp))
+    return buildBinaryExpr(CompiledIndexExprNode::Kind::RemUI, remOp.getLhs(), remOp.getRhs());
+  if (auto remOp = mlir::dyn_cast<mlir::arith::RemSIOp>(definingOp))
+    return buildBinaryExpr(CompiledIndexExprNode::Kind::RemSI, remOp.getLhs(), remOp.getRhs());
+  if (auto cmpOp = mlir::dyn_cast<mlir::arith::CmpIOp>(definingOp)) {
+    auto expr = buildBinaryExpr(CompiledIndexExprNode::Kind::CmpI, cmpOp.getLhs(), cmpOp.getRhs());
+    if (failed(expr))
+      return mlir::failure();
+    auto exprNode = std::make_shared<CompiledIndexExprNode>(*expr->node);
+    exprNode->predicate = cmpOp.getPredicate();
+    return CompiledIndexExpr(exprNode);
+  }
+  if (auto maxOp = mlir::dyn_cast<mlir::arith::MaxUIOp>(definingOp)) {
+    auto lhs = compileIndexValueImpl(maxOp.getLhs());
+    auto rhs = compileIndexValueImpl(maxOp.getRhs());
+    if (failed(lhs) || failed(rhs))
+      return mlir::failure();
+
+    CompiledIndexExprNode cmpExpr;
+    cmpExpr.kind = CompiledIndexExprNode::Kind::CmpI;
+    cmpExpr.predicate = mlir::arith::CmpIPredicate::uge;
+    cmpExpr.operands = {*lhs, *rhs};
+
+    CompiledIndexExprNode selectExpr;
+    selectExpr.kind = CompiledIndexExprNode::Kind::Select;
+    selectExpr.operands = {makeCompiledIndexExpr(std::move(cmpExpr)), *lhs, *rhs};
+    return makeCompiledIndexExpr(std::move(selectExpr));
+  }
+  if (auto selectOp = mlir::dyn_cast<mlir::arith::SelectOp>(definingOp)) {
+    auto condition = compileIndexValueImpl(selectOp.getCondition());
+    auto trueValue = compileIndexValueImpl(selectOp.getTrueValue());
+    auto falseValue = compileIndexValueImpl(selectOp.getFalseValue());
+    if (failed(condition) || failed(trueValue) || failed(falseValue))
+      return mlir::failure();
+    CompiledIndexExprNode expr;
+    expr.kind = CompiledIndexExprNode::Kind::Select;
+    expr.operands = {*condition, *trueValue, *falseValue};
+    return makeCompiledIndexExpr(std::move(expr));
+  }
+  if (auto loadOp = mlir::dyn_cast<mlir::memref::LoadOp>(definingOp))
+    return compileConstantGlobalLoad(loadOp);
+
+  CompiledIndexExprNode expr;
+  expr.kind = CompiledIndexExprNode::Kind::Symbol;
+  expr.symbol = value;
+  return makeCompiledIndexExpr(std::move(expr));
+}
 
 llvm::FailureOr<int64_t> resolveIndexValueImpl(mlir::Value value, const StaticValueKnowledge* knowledge) {
   value = resolveAlias(value, knowledge);
@@ -110,6 +404,16 @@ llvm::FailureOr<int64_t> resolveIndexValueImpl(mlir::Value value, const StaticVa
     return static_cast<int64_t>(static_cast<uint64_t>(*lhs) / static_cast<uint64_t>(*rhs));
   }
 
+  if (auto divOp = mlir::dyn_cast<mlir::arith::DivSIOp>(definingOp)) {
+    auto lhs = resolveIndexValueImpl(divOp.getLhs(), knowledge);
+    auto rhs = resolveIndexValueImpl(divOp.getRhs(), knowledge);
+    if (failed(lhs) || failed(rhs) || *rhs == 0)
+      return mlir::failure();
+    if (*lhs == std::numeric_limits<int64_t>::min() && *rhs == -1)
+      return mlir::failure();
+    return *lhs / *rhs;
+  }
+
   if (auto minOp = mlir::dyn_cast<mlir::arith::MinUIOp>(definingOp)) {
     auto lhs = resolveIndexValueImpl(minOp.getLhs(), knowledge);
     auto rhs = resolveIndexValueImpl(minOp.getRhs(), knowledge);
@@ -126,6 +430,34 @@ llvm::FailureOr<int64_t> resolveIndexValueImpl(mlir::Value value, const StaticVa
     return static_cast<int64_t>(static_cast<uint64_t>(*lhs) % static_cast<uint64_t>(*rhs));
   }
 
+  if (auto remOp = mlir::dyn_cast<mlir::arith::RemSIOp>(definingOp)) {
+    auto lhs = resolveIndexValueImpl(remOp.getLhs(), knowledge);
+    auto rhs = resolveIndexValueImpl(remOp.getRhs(), knowledge);
+    if (failed(lhs) || failed(rhs) || *rhs == 0)
+      return mlir::failure();
+    if (*lhs == std::numeric_limits<int64_t>::min() && *rhs == -1)
+      return 0;
+    return *lhs % *rhs;
+  }
+
+  if (auto cmpOp = mlir::dyn_cast<mlir::arith::CmpIOp>(definingOp)) {
+    auto lhs = resolveIndexValueImpl(cmpOp.getLhs(), knowledge);
+    auto rhs = resolveIndexValueImpl(cmpOp.getRhs(), knowledge);
+    if (failed(lhs) || failed(rhs))
+      return mlir::failure();
+    return evaluateCmpPredicate(cmpOp.getPredicate(), *lhs, *rhs) ? 1 : 0;
+  }
+
+  if (auto selectOp = mlir::dyn_cast<mlir::arith::SelectOp>(definingOp)) {
+    auto condition = resolveIndexValueImpl(selectOp.getCondition(), knowledge);
+    if (failed(condition))
+      return mlir::failure();
+    return resolveIndexValueImpl(*condition != 0 ? selectOp.getTrueValue() : selectOp.getFalseValue(), knowledge);
+  }
+
+  if (auto loadOp = mlir::dyn_cast<mlir::memref::LoadOp>(definingOp))
+    return resolveConstantGlobalLoad(loadOp, knowledge);
+
   return mlir::failure();
 }
 
@@ -217,8 +549,10 @@ llvm::FailureOr<ResolvedContiguousAddress> resolveContiguousAddressImpl(mlir::Va
       if (!isMemoryContiguous(sourceType.getShape(), offsets, sizes, strides))
         return mlir::failure();
 
-      auto sourceStrides = computeRowMajorStrides(sourceType.getShape());
-      byteOffset += linearizeIndex(offsets, sourceStrides) * subviewType.getElementTypeBitWidth() / 8;
+      auto sourceStrides = getStaticMemRefStrides(sourceType);
+      if (failed(sourceStrides))
+        return mlir::failure();
+      byteOffset += linearizeIndex(offsets, *sourceStrides) * getElementTypeSizeInBytes(subviewType.getElementType());
       value = resolveAlias(subviewOp.getSource(), knowledge);
       continue;
     }
@@ -243,17 +577,206 @@ llvm::FailureOr<ResolvedContiguousAddress> resolveContiguousAddressImpl(mlir::Va
   }
 }
 
-} // namespace
+llvm::FailureOr<CompiledAddressExpr> compileContiguousAddressExprImpl(mlir::Value value) {
+  int64_t constantByteOffset = 0;
+  CompiledIndexExpr byteOffsetExpr;
+  {
+    CompiledIndexExprNode expr;
+    expr.kind = CompiledIndexExprNode::Kind::Constant;
+    expr.constant = 0;
+    byteOffsetExpr = makeCompiledIndexExpr(std::move(expr));
+  }
 
-llvm::FailureOr<int64_t> resolveIndexValue(mlir::Value value) { return resolveIndexValueImpl(value, nullptr); }
+  while (true) {
+    if (mlir::isa<mlir::BlockArgument>(value))
+      return CompiledAddressExpr {value, byteOffsetExpr};
+
+    mlir::Operation* definingOp = value.getDefiningOp();
+    if (!definingOp)
+      return mlir::failure();
+
+    if (auto dpsDefiningOp = mlir::dyn_cast<mlir::DestinationStyleOpInterface>(definingOp)) {
+      mlir::OpOperand* tiedOperand = dpsDefiningOp.getTiedOpOperand(mlir::dyn_cast<mlir::OpResult>(value));
+      if (!tiedOperand)
+        return mlir::failure();
+      value = tiedOperand->get();
+      continue;
+    }
+
+    if (auto forOp = mlir::dyn_cast<mlir::scf::ForOp>(definingOp)) {
+      auto result = mlir::dyn_cast<mlir::OpResult>(value);
+      if (!result)
+        return mlir::failure();
+
+      auto yieldOp = mlir::cast<mlir::scf::YieldOp>(forOp.getBody()->getTerminator());
+      mlir::Value yieldedValue = yieldOp.getOperand(result.getResultNumber());
+      if (auto blockArgument = mlir::dyn_cast<mlir::BlockArgument>(yieldedValue)) {
+        if (blockArgument.getOwner() == forOp.getBody() && blockArgument.getArgNumber() > 0
+            && static_cast<unsigned>(blockArgument.getArgNumber() - 1) < forOp.getInitArgs().size()) {
+          value = forOp.getInitArgs()[blockArgument.getArgNumber() - 1];
+          continue;
+        }
+      }
+
+      value = yieldedValue;
+      continue;
+    }
+
+    if (auto subviewOp = mlir::dyn_cast<mlir::memref::SubViewOp>(definingOp)) {
+      auto sourceType = mlir::dyn_cast<mlir::MemRefType>(subviewOp.getSource().getType());
+      auto subviewType = mlir::dyn_cast<mlir::MemRefType>(subviewOp.getType());
+      if (!sourceType || !subviewType || !sourceType.hasStaticShape() || !subviewType.hasStaticShape())
+        return mlir::failure();
+
+      llvm::SmallVector<int64_t> staticSizes;
+      staticSizes.reserve(subviewOp.getMixedSizes().size());
+      llvm::SmallVector<int64_t> staticStrides;
+      staticStrides.reserve(subviewOp.getMixedStrides().size());
+      llvm::SmallVector<int64_t> staticOffsets;
+      staticOffsets.reserve(subviewOp.getMixedOffsets().size());
+      bool hasOnlyStaticOffsets = true;
+
+      for (mlir::OpFoldResult offset : subviewOp.getMixedOffsets())
+        if (auto attr = mlir::dyn_cast<mlir::Attribute>(offset))
+          staticOffsets.push_back(mlir::cast<mlir::IntegerAttr>(attr).getInt());
+        else
+          hasOnlyStaticOffsets = false;
+      for (mlir::OpFoldResult size : subviewOp.getMixedSizes()) {
+        auto attr = mlir::dyn_cast<mlir::Attribute>(size);
+        if (!attr)
+          return mlir::failure();
+        staticSizes.push_back(mlir::cast<mlir::IntegerAttr>(attr).getInt());
+      }
+      for (mlir::OpFoldResult stride : subviewOp.getMixedStrides()) {
+        auto attr = mlir::dyn_cast<mlir::Attribute>(stride);
+        if (!attr)
+          return mlir::failure();
+        staticStrides.push_back(mlir::cast<mlir::IntegerAttr>(attr).getInt());
+      }
+
+      if (!isContiguousSubviewWithDynamicOffsets(
+            sourceType.getShape(), subviewOp.getMixedOffsets(), staticSizes, staticStrides)) {
+        return mlir::failure();
+      }
+
+      if (hasOnlyStaticOffsets) {
+        if (!isMemoryContiguous(sourceType.getShape(), staticOffsets, staticSizes, staticStrides))
+          return mlir::failure();
+
+        auto sourceStrides = getStaticMemRefStrides(sourceType);
+        if (failed(sourceStrides))
+          return mlir::failure();
+        constantByteOffset +=
+          linearizeIndex(staticOffsets, *sourceStrides) * getElementTypeSizeInBytes(subviewType.getElementType());
+      }
+      else {
+        auto sourceStrides = getStaticMemRefStrides(sourceType);
+        if (failed(sourceStrides))
+          return mlir::failure();
+        CompiledIndexExpr offsetExpr;
+        {
+          CompiledIndexExprNode expr;
+          expr.kind = CompiledIndexExprNode::Kind::Constant;
+          expr.constant = 0;
+          offsetExpr = makeCompiledIndexExpr(std::move(expr));
+        }
+
+        for (auto [mixedOffset, sourceStride] : llvm::zip_equal(subviewOp.getMixedOffsets(), *sourceStrides)) {
+          CompiledIndexExpr operandExpr;
+          if (auto attr = mlir::dyn_cast<mlir::Attribute>(mixedOffset)) {
+            CompiledIndexExprNode expr;
+            expr.kind = CompiledIndexExprNode::Kind::Constant;
+            expr.constant = mlir::cast<mlir::IntegerAttr>(attr).getInt() * sourceStride
+                          * getElementTypeSizeInBytes(subviewType.getElementType());
+            operandExpr = makeCompiledIndexExpr(std::move(expr));
+          }
+          else {
+            auto compiledOffset = compileIndexValueImpl(mlir::cast<mlir::Value>(mixedOffset));
+            if (failed(compiledOffset))
+              return mlir::failure();
+            CompiledIndexExpr scaleExpr;
+            {
+              CompiledIndexExprNode expr;
+              expr.kind = CompiledIndexExprNode::Kind::Constant;
+              expr.constant = sourceStride * getElementTypeSizeInBytes(subviewType.getElementType());
+              scaleExpr = makeCompiledIndexExpr(std::move(expr));
+            }
+
+            CompiledIndexExprNode expr;
+            expr.kind = CompiledIndexExprNode::Kind::Mul;
+            expr.operands = {*compiledOffset, scaleExpr};
+            operandExpr = makeCompiledIndexExpr(std::move(expr));
+          }
+
+          CompiledIndexExprNode expr;
+          expr.kind = CompiledIndexExprNode::Kind::Add;
+          expr.operands = {offsetExpr, operandExpr};
+          offsetExpr = makeCompiledIndexExpr(std::move(expr));
+        }
+
+        CompiledIndexExpr constantExpr;
+        {
+          CompiledIndexExprNode expr;
+          expr.kind = CompiledIndexExprNode::Kind::Constant;
+          expr.constant = constantByteOffset;
+          constantExpr = makeCompiledIndexExpr(std::move(expr));
+        }
+        CompiledIndexExprNode expr;
+        expr.kind = CompiledIndexExprNode::Kind::Add;
+        expr.operands = {constantExpr, offsetExpr};
+        byteOffsetExpr = makeCompiledIndexExpr(std::move(expr));
+        constantByteOffset = 0;
+      }
+
+      value = subviewOp.getSource();
+      continue;
+    }
+
+    if (auto castOp = mlir::dyn_cast<mlir::memref::CastOp>(definingOp)) {
+      value = castOp.getSource();
+      continue;
+    }
+    if (auto collapseOp = mlir::dyn_cast<mlir::memref::CollapseShapeOp>(definingOp)) {
+      value = collapseOp.getSrc();
+      continue;
+    }
+    if (auto expandOp = mlir::dyn_cast<mlir::memref::ExpandShapeOp>(definingOp)) {
+      value = expandOp.getSrc();
+      continue;
+    }
+
+    if (mlir::isa<mlir::memref::AllocOp, mlir::memref::GetGlobalOp>(definingOp)) {
+      if (constantByteOffset != 0) {
+        CompiledIndexExpr constantExpr;
+        {
+          CompiledIndexExprNode expr;
+          expr.kind = CompiledIndexExprNode::Kind::Constant;
+          expr.constant = constantByteOffset;
+          constantExpr = makeCompiledIndexExpr(std::move(expr));
+        }
+        if (byteOffsetExpr.node->kind == CompiledIndexExprNode::Kind::Constant && byteOffsetExpr.node->constant == 0)
+          byteOffsetExpr = constantExpr;
+        else {
+          CompiledIndexExprNode expr;
+          expr.kind = CompiledIndexExprNode::Kind::Add;
+          expr.operands = {constantExpr, byteOffsetExpr};
+          byteOffsetExpr = makeCompiledIndexExpr(std::move(expr));
+        }
+      }
+      return CompiledAddressExpr {value, byteOffsetExpr};
+    }
+
+    return mlir::failure();
+  }
+}
+
+} // namespace
 
 llvm::FailureOr<int64_t> resolveIndexValue(mlir::Value value, const StaticValueKnowledge& knowledge) {
   return resolveIndexValueImpl(value, &knowledge);
 }
 
-llvm::FailureOr<ResolvedContiguousAddress> resolveContiguousAddress(mlir::Value value) {
-  return resolveContiguousAddressImpl(value, nullptr);
-}
+llvm::FailureOr<CompiledIndexExpr> compileIndexExpr(mlir::Value value) { return compileIndexValueImpl(value); }
 
 llvm::FailureOr<ResolvedContiguousAddress> resolveContiguousAddress(mlir::Value value,
                                                                     const StaticValueKnowledge& knowledge) {
@@ -264,4 +787,21 @@ mlir::Value resolveLoopCarriedAlias(mlir::Value value, const StaticValueKnowledg
   return resolveLoopCarriedAliasImpl(value, &knowledge);
 }
 
+llvm::FailureOr<CompiledAddressExpr> compileContiguousAddressExpr(mlir::Value value) {
+  return compileContiguousAddressExprImpl(value);
+}
+
+llvm::FailureOr<int64_t> CompiledIndexExpr::evaluate(const StaticValueKnowledge& knowledge) const {
+  return evaluateCompiledIndexExpr(*this, knowledge);
+}
+
+llvm::FailureOr<ResolvedContiguousAddress> CompiledAddressExpr::evaluate(const StaticValueKnowledge& knowledge,
+                                                                         std::optional<unsigned> lane) const {
+  (void) lane;
+  auto resolvedOffset = byteOffset.evaluate(knowledge);
+  if (failed(resolvedOffset))
+    return mlir::failure();
+  return ResolvedContiguousAddress {base, *resolvedOffset};
+}
+
 } // namespace onnx_mlir

src/PIM/Common/IR/AddressAnalysis.hpp

@@ -1,10 +1,14 @@
 #pragma once
 
+#include "mlir/Dialect/Arith/IR/Arith.h"
 #include "mlir/Dialect/MemRef/IR/MemRef.h"
 #include "mlir/IR/Value.h"
 
 #include "llvm/ADT/DenseMap.h"
 
+#include <memory>
+#include <optional>
+
 namespace onnx_mlir {
 
 /// Describes a value as a base addressable object plus a statically known
@@ -23,21 +27,68 @@ struct StaticValueKnowledge {
   StaticValueKnowledge() {}
 };
 
+struct CompiledIndexExprNode;
+
+struct CompiledIndexExpr {
+  std::shared_ptr<CompiledIndexExprNode> node;
+
+  CompiledIndexExpr() = default;
+  explicit CompiledIndexExpr(std::shared_ptr<CompiledIndexExprNode> node)
+  : node(std::move(node)) {}
+
+  llvm::FailureOr<int64_t> evaluate(const StaticValueKnowledge& knowledge) const;
+};
+
+struct CompiledIndexExprNode {
+  enum class Kind {
+    Constant,
+    Symbol,
+    Add,
+    Sub,
+    Mul,
+    DivUI,
+    DivSI,
+    RemUI,
+    RemSI,
+    MinUI,
+    CmpI,
+    Select,
+    ConstantGlobalLoad
+  };
+
+  Kind kind = Kind::Constant;
+  int64_t constant = 0;
+  mlir::Value symbol;
+  mlir::arith::CmpIPredicate predicate = mlir::arith::CmpIPredicate::eq;
+  mlir::memref::GlobalOp globalOp;
+  llvm::SmallVector<int64_t, 4> globalStrides;
+  llvm::SmallVector<CompiledIndexExpr, 4> operands;
+};
+
+struct CompiledAddressExpr {
+  mlir::Value base;
+  CompiledIndexExpr byteOffset;
+
+  llvm::FailureOr<ResolvedContiguousAddress> evaluate(const StaticValueKnowledge& knowledge,
+                                                      std::optional<unsigned> lane) const;
+};
+
 mlir::memref::GlobalOp lookupGlobalForGetGlobal(mlir::ModuleOp moduleOp, mlir::memref::GetGlobalOp getGlobalOp);
 
 /// Resolves a value to contiguous backing storage when that storage can be
 /// proven statically from aliases, DPS ties, casts, and subviews.
-llvm::FailureOr<ResolvedContiguousAddress> resolveContiguousAddress(mlir::Value value);
 llvm::FailureOr<ResolvedContiguousAddress> resolveContiguousAddress(mlir::Value value,
-                                                                    const StaticValueKnowledge& knowledge);
+                                                                    const StaticValueKnowledge& knowledge = {});
 
 /// Statically evaluates index-like SSA values, including simple integer
 /// arithmetic and loop facts recorded in `knowledge`.
-llvm::FailureOr<int64_t> resolveIndexValue(mlir::Value value);
-llvm::FailureOr<int64_t> resolveIndexValue(mlir::Value value, const StaticValueKnowledge& knowledge);
+llvm::FailureOr<int64_t> resolveIndexValue(mlir::Value value, const StaticValueKnowledge& knowledge = {});
+llvm::FailureOr<CompiledIndexExpr> compileIndexExpr(mlir::Value value);
 
 /// Follows alias, view, and DPS chains to recover the backing value of a
 /// loop-carried memref/result.
 mlir::Value resolveLoopCarriedAlias(mlir::Value value, const StaticValueKnowledge& knowledge);
 
+llvm::FailureOr<CompiledAddressExpr> compileContiguousAddressExpr(mlir::Value value);
+
 } // namespace onnx_mlir

src/PIM/Common/IR/AffineUtils.cpp

@@ -0,0 +1,182 @@
+#include "mlir/Dialect/Arith/IR/Arith.h"
+#include "mlir/IR/Matchers.h"
+
+#include "AffineUtils.hpp"
+#include "ConstantUtils.hpp"
+
+using namespace mlir;
+
+namespace onnx_mlir {
+
+static FailureOr<int64_t> floorDivSigned(int64_t lhs, int64_t rhs) {
+  if (rhs <= 0)
+    return failure();
+
+  int64_t quotient = lhs / rhs;
+  int64_t remainder = lhs % rhs;
+  if (remainder != 0 && lhs < 0)
+    --quotient;
+  return quotient;
+}
+
+static FailureOr<int64_t> ceilDivSigned(int64_t lhs, int64_t rhs) {
+  if (rhs <= 0)
+    return failure();
+
+  int64_t quotient = lhs / rhs;
+  int64_t remainder = lhs % rhs;
+  if (remainder != 0 && lhs > 0)
+    ++quotient;
+  return quotient;
+}
+
+Value createOrFoldAffineApply(
+  RewriterBase& rewriter, Location loc, AffineMap map, ValueRange operands, Operation* constantAnchor) {
+  assert(constantAnchor && "expected a valid constant anchor");
+  assert(map.getNumResults() == 1 && "affine.apply expects a single-result affine map");
+
+  SmallVector<Attribute> operandConstants;
+  operandConstants.reserve(operands.size());
+  for (Value operand : operands) {
+    std::optional<int64_t> constantValue = matchConstantIndexValue(operand);
+    if (!constantValue)
+      return affine::AffineApplyOp::create(rewriter, loc, map, operands).getResult();
+    operandConstants.push_back(rewriter.getIndexAttr(*constantValue));
+  }
+
+  SmallVector<Attribute> foldedResults;
+  if (succeeded(map.constantFold(operandConstants, foldedResults)) && foldedResults.size() == 1)
+    if (auto constantResult = dyn_cast<IntegerAttr>(foldedResults.front()))
+      return getOrCreateIndexConstant(rewriter, constantAnchor, constantResult.getInt());
+
+  return affine::AffineApplyOp::create(rewriter, loc, map, operands).getResult();
+}
+
+Value createOrFoldAffineApply(
+  RewriterBase& rewriter, Location loc, AffineExpr expr, ValueRange dims, Operation* constantAnchor) {
+  AffineMap map = AffineMap::get(/*dimCount=*/dims.size(), /*symbolCount=*/0, expr);
+  return createOrFoldAffineApply(rewriter, loc, map, dims, constantAnchor);
+}
+
+Value affineMulConst(RewriterBase& rewriter, Location loc, Value value, int64_t multiplier, Operation* constantAnchor) {
+  assert(constantAnchor && "expected a valid constant anchor");
+  if (multiplier == 0)
+    return getOrCreateIndexConstant(rewriter, constantAnchor, 0);
+  if (multiplier == 1)
+    return value;
+
+  AffineExpr d0 = getAffineDimExpr(0, rewriter.getContext());
+  return createOrFoldAffineApply(rewriter, loc, d0 * multiplier, ValueRange {value}, constantAnchor);
+}
+
+Value affineModConst(RewriterBase& rewriter, Location loc, Value value, int64_t divisor, Operation* constantAnchor) {
+  assert(constantAnchor && "expected a valid constant anchor");
+  assert(divisor > 0 && "expected a positive affine.mod divisor");
+  if (divisor == 1)
+    return getOrCreateIndexConstant(rewriter, constantAnchor, 0);
+
+  AffineExpr d0 = getAffineDimExpr(0, rewriter.getContext());
+  return createOrFoldAffineApply(rewriter, loc, d0 % divisor, ValueRange {value}, constantAnchor);
+}
+
+Value affineFloorDivConst(
+  RewriterBase& rewriter, Location loc, Value value, int64_t divisor, Operation* constantAnchor) {
+  assert(constantAnchor && "expected a valid constant anchor");
+  assert(divisor > 0 && "expected a positive affine.floor_div divisor");
+  if (divisor == 1)
+    return value;
+
+  AffineExpr d0 = getAffineDimExpr(0, rewriter.getContext());
+  return createOrFoldAffineApply(rewriter, loc, d0.floorDiv(divisor), ValueRange {value}, constantAnchor);
+}
+
+FailureOr<int64_t> evaluateAffineExpr(AffineExpr expr, ArrayRef<int64_t> dims, ArrayRef<int64_t> symbols) {
+  if (auto constant = dyn_cast<AffineConstantExpr>(expr))
+    return constant.getValue();
+  if (auto dim = dyn_cast<AffineDimExpr>(expr)) {
+    unsigned position = dim.getPosition();
+    if (position >= dims.size())
+      return failure();
+    return dims[position];
+  }
+  if (auto symbol = dyn_cast<AffineSymbolExpr>(expr)) {
+    unsigned position = symbol.getPosition();
+    if (position >= symbols.size())
+      return failure();
+    return symbols[position];
+  }
+
+  auto binary = dyn_cast<AffineBinaryOpExpr>(expr);
+  if (!binary)
+    return failure();
+
+  FailureOr<int64_t> lhs = evaluateAffineExpr(binary.getLHS(), dims, symbols);
+  FailureOr<int64_t> rhs = evaluateAffineExpr(binary.getRHS(), dims, symbols);
+  if (failed(lhs) || failed(rhs))
+    return failure();
+
+  switch (binary.getKind()) {
+  case AffineExprKind::Add:      return *lhs + *rhs;
+  case AffineExprKind::Mul:      return *lhs * *rhs;
+  case AffineExprKind::FloorDiv: return floorDivSigned(*lhs, *rhs);
+  case AffineExprKind::CeilDiv:  return ceilDivSigned(*lhs, *rhs);
+  case AffineExprKind::Mod:      {
+    FailureOr<int64_t> div = floorDivSigned(*lhs, *rhs);
+    if (failed(div))
+      return failure();
+    return *lhs - *div * *rhs;
+  }
+  default: return failure();
+  }
+}
+
+FailureOr<int64_t> evaluateSingleResultAffineMap(AffineMap map, ArrayRef<int64_t> operands) {
+  if (map.getNumResults() != 1 || operands.size() != map.getNumInputs())
+    return failure();
+
+  ArrayRef<int64_t> dims(operands.data(), map.getNumDims());
+  ArrayRef<int64_t> symbols(operands.data() + map.getNumDims(), map.getNumSymbols());
+  return evaluateAffineExpr(map.getResult(0), dims, symbols);
+}
+
+FailureOr<int64_t> evaluateAffineApply(affine::AffineApplyOp affineApply, IndexValueResolver resolver) {
+  SmallVector<int64_t, 4> operands;
+  operands.reserve(affineApply.getMapOperands().size());
+  for (Value operand : affineApply.getMapOperands()) {
+    FailureOr<int64_t> folded = resolver(operand);
+    if (failed(folded))
+      return failure();
+    operands.push_back(*folded);
+  }
+
+  return evaluateSingleResultAffineMap(affineApply.getAffineMap(), operands);
+}
+
+bool isSingleResultSymbolFreeAffineMap(AffineMap map) { return map.getNumResults() == 1 && map.getNumSymbols() == 0; }
+
+bool isDimAndConstantAffineExpr(AffineExpr expr) {
+  switch (expr.getKind()) {
+  case AffineExprKind::Constant:
+  case AffineExprKind::DimId:    return true;
+  case AffineExprKind::SymbolId: return false;
+  case AffineExprKind::Add:      {
+    auto binaryExpr = cast<AffineBinaryOpExpr>(expr);
+    return isDimAndConstantAffineExpr(binaryExpr.getLHS()) && isDimAndConstantAffineExpr(binaryExpr.getRHS());
+  }
+  case AffineExprKind::Mul: {
+    auto binaryExpr = cast<AffineBinaryOpExpr>(expr);
+    return (isa<AffineConstantExpr>(binaryExpr.getLHS()) && isDimAndConstantAffineExpr(binaryExpr.getRHS()))
+        || (isa<AffineConstantExpr>(binaryExpr.getRHS()) && isDimAndConstantAffineExpr(binaryExpr.getLHS()));
+  }
+  case AffineExprKind::FloorDiv:
+  case AffineExprKind::CeilDiv:
+  case AffineExprKind::Mod:      {
+    auto binaryExpr = cast<AffineBinaryOpExpr>(expr);
+    return isa<AffineConstantExpr>(binaryExpr.getRHS()) && isDimAndConstantAffineExpr(binaryExpr.getLHS());
+  }
+  }
+
+  llvm_unreachable("unexpected affine expression kind");
+}
+
+} // namespace onnx_mlir

src/PIM/Common/IR/AffineUtils.hpp

@@ -0,0 +1,55 @@
+#pragma once
+
+#include "mlir/Dialect/Affine/IR/AffineOps.h"
+#include "mlir/IR/AffineExpr.h"
+#include "mlir/IR/PatternMatch.h"
+#include "mlir/Support/LogicalResult.h"
+
+#include "llvm/ADT/FunctionExtras.h"
+
+namespace onnx_mlir {
+
+using IndexValueResolver = llvm::function_ref<llvm::FailureOr<int64_t>(mlir::Value)>;
+
+mlir::Value createOrFoldAffineApply(mlir::RewriterBase& rewriter,
+                                    mlir::Location loc,
+                                    mlir::AffineMap map,
+                                    mlir::ValueRange operands,
+                                    mlir::Operation* constantAnchor);
+
+mlir::Value createOrFoldAffineApply(mlir::RewriterBase& rewriter,
+                                    mlir::Location loc,
+                                    mlir::AffineExpr expr,
+                                    mlir::ValueRange dims,
+                                    mlir::Operation* constantAnchor);
+
+mlir::Value affineMulConst(mlir::RewriterBase& rewriter,
+                           mlir::Location loc,
+                           mlir::Value value,
+                           int64_t multiplier,
+                           mlir::Operation* constantAnchor);
+
+mlir::Value affineModConst(mlir::RewriterBase& rewriter,
+                           mlir::Location loc,
+                           mlir::Value value,
+                           int64_t divisor,
+                           mlir::Operation* constantAnchor);
+
+mlir::Value affineFloorDivConst(mlir::RewriterBase& rewriter,
+                                mlir::Location loc,
+                                mlir::Value value,
+                                int64_t divisor,
+                                mlir::Operation* constantAnchor);
+
+llvm::FailureOr<int64_t>
+evaluateAffineExpr(mlir::AffineExpr expr, llvm::ArrayRef<int64_t> dims, llvm::ArrayRef<int64_t> symbols = {});
+
+llvm::FailureOr<int64_t> evaluateSingleResultAffineMap(mlir::AffineMap map, llvm::ArrayRef<int64_t> operands);
+
+llvm::FailureOr<int64_t> evaluateAffineApply(mlir::affine::AffineApplyOp affineApply, IndexValueResolver resolver);
+
+bool isSingleResultSymbolFreeAffineMap(mlir::AffineMap map);
+
+bool isDimAndConstantAffineExpr(mlir::AffineExpr expr);
+
+} // namespace onnx_mlir

src/PIM/Common/IR/BatchCoreUtils.cpp

@@ -0,0 +1,32 @@
+#include "src/Accelerators/PIM/Common/IR/BatchCoreUtils.hpp"
+#include "src/Accelerators/PIM/Common/PimCommon.hpp"
+
+namespace onnx_mlir {
+
+llvm::SmallVector<int32_t> getBatchCoreIds(pim::PimCoreBatchOp coreBatchOp) {
+  auto coreIdsAttr = coreBatchOp->getAttrOfType<mlir::DenseI32ArrayAttr>(onnx_mlir::kCoreIdsAttrName);
+  assert(coreIdsAttr && "pim.core_batch requires coreIds array attribute");
+  return llvm::SmallVector<int32_t>(coreIdsAttr.asArrayRef().begin(), coreIdsAttr.asArrayRef().end());
+}
+
+llvm::SmallVector<int32_t> getLaneChunkCoreIds(llvm::ArrayRef<int32_t> coreIds, size_t laneCount, unsigned lane) {
+  llvm::SmallVector<int32_t> laneCoreIds;
+  laneCoreIds.reserve(coreIds.size() / laneCount);
+  for (size_t chunkIndex = 0; chunkIndex < coreIds.size() / laneCount; ++chunkIndex)
+    laneCoreIds.push_back(coreIds[chunkIndex * laneCount + lane]);
+  return laneCoreIds;
+}
+
+bool isExplicitHostMemCopyOperand(mlir::Operation* op, unsigned operandIndex) {
+  if (mlir::isa<pim::PimMemCopyHostToDevOp>(op))
+    return operandIndex == 3;
+  if (mlir::isa<pim::PimMemCopyDevToHostOp>(op))
+    return operandIndex == 2;
+  return false;
+}
+
+bool isExplicitDevToHostTargetOperand(mlir::Operation* op, unsigned operandIndex) {
+  return mlir::isa<pim::PimMemCopyDevToHostOp>(op) && operandIndex == 2;
+}
+
+} // namespace onnx_mlir

src/PIM/Common/IR/BatchCoreUtils.hpp

@@ -0,0 +1,18 @@
+#pragma once
+
+#include "llvm/ADT/ArrayRef.h"
+#include "llvm/ADT/SmallVector.h"
+
+#include "src/Accelerators/PIM/Dialect/Pim/PimOps.hpp"
+
+namespace onnx_mlir {
+
+llvm::SmallVector<int32_t> getBatchCoreIds(pim::PimCoreBatchOp coreBatchOp);
+
+llvm::SmallVector<int32_t> getLaneChunkCoreIds(llvm::ArrayRef<int32_t> coreIds, size_t laneCount, unsigned lane);
+
+bool isExplicitHostMemCopyOperand(mlir::Operation* op, unsigned operandIndex);
+
+bool isExplicitDevToHostTargetOperand(mlir::Operation* op, unsigned operandIndex);
+
+} // namespace onnx_mlir

src/PIM/Common/IR/ConstantUtils.cpp

@@ -0,0 +1,93 @@
+#include "mlir/Dialect/Arith/IR/Arith.h"
+#include "mlir/Dialect/Func/IR/FuncOps.h"
+#include "mlir/IR/Builders.h"
+#include "mlir/IR/BuiltinOps.h"
+#include "mlir/IR/Dialect.h"
+
+#include "ConstantUtils.hpp"
+
+using namespace mlir;
+
+namespace onnx_mlir {
+
+Block* getConstantInsertionBlock(Operation* anchorOp) {
+  assert(anchorOp && "expected a valid anchor operation");
+
+  if (auto funcOp = dyn_cast<func::FuncOp>(anchorOp))
+    return &funcOp.getBody().front();
+  if (auto funcOp = anchorOp->getParentOfType<func::FuncOp>())
+    return &funcOp.getBody().front();
+  if (auto moduleOp = dyn_cast<ModuleOp>(anchorOp))
+    return moduleOp.getBody();
+  if (auto moduleOp = anchorOp->getParentOfType<ModuleOp>())
+    return moduleOp.getBody();
+  return anchorOp->getBlock();
+}
+
+Value getOrCreateConstant(OperationFolder& folder, Operation* anchorOp, Attribute value, Type type) {
+  assert(anchorOp && "expected a valid anchor operation");
+  Block* hostBlock = getConstantInsertionBlock(anchorOp);
+  for (Operation& op : *hostBlock) {
+    auto constantOp = dyn_cast<arith::ConstantOp>(&op);
+    if (!constantOp || constantOp.getType() != type || constantOp.getValue() != value)
+      continue;
+    return constantOp.getResult();
+  }
+
+  auto* arithDialect = anchorOp->getContext()->getOrLoadDialect<arith::ArithDialect>();
+  return folder.getOrCreateConstant(hostBlock, arithDialect, value, type);
+}
+
+Value getOrCreateConstant(RewriterBase& rewriter, Operation* anchorOp, Attribute value, Type type) {
+  assert(anchorOp && "expected a valid anchor operation");
+  Block* hostBlock = getConstantInsertionBlock(anchorOp);
+  for (Operation& op : *hostBlock) {
+    auto constantOp = dyn_cast<arith::ConstantOp>(&op);
+    if (!constantOp || constantOp.getType() != type || constantOp.getValue() != value)
+      continue;
+    return constantOp.getResult();
+  }
+
+  OpBuilder::InsertionGuard guard(rewriter);
+  rewriter.setInsertionPointToStart(hostBlock);
+  return arith::ConstantOp::create(rewriter, anchorOp->getLoc(), type, cast<TypedAttr>(value)).getResult();
+}
+
+Value getOrCreateConstantLike(OperationFolder& folder, arith::ConstantOp constantOp) {
+  return getOrCreateConstant(folder, constantOp.getOperation(), constantOp.getValue(), constantOp.getType());
+}
+
+Value getOrCreateIndexConstant(OperationFolder& folder, Operation* anchorOp, int64_t value) {
+  Builder builder(anchorOp->getContext());
+  return getOrCreateConstant(folder, anchorOp, builder.getIndexAttr(value), builder.getIndexType());
+}
+
+Value getOrCreateIndexConstant(RewriterBase& rewriter, Operation* anchorOp, int64_t value) {
+  Builder builder(anchorOp->getContext());
+  return getOrCreateConstant(rewriter, anchorOp, builder.getIndexAttr(value), builder.getIndexType());
+}
+
+std::optional<int64_t> matchConstantIndexValue(Value value) {
+  if (!value || !value.getType().isIndex())
+    return std::nullopt;
+
+  if (auto constant = value.getDefiningOp<arith::ConstantIndexOp>())
+    return constant.value();
+
+  if (auto constant = value.getDefiningOp<arith::ConstantOp>())
+    if (auto intAttr = dyn_cast<IntegerAttr>(constant.getValue()); intAttr && intAttr.getType().isIndex())
+      return intAttr.getInt();
+
+  return std::nullopt;
+}
+
+std::optional<int64_t> matchConstantIndexValue(OpFoldResult value) {
+  if (auto attr = dyn_cast<Attribute>(value))
+    if (auto intAttr = dyn_cast<IntegerAttr>(attr); intAttr && intAttr.getType().isIndex())
+      return intAttr.getInt();
+  if (auto operand = dyn_cast<Value>(value))
+    return matchConstantIndexValue(operand);
+  return std::nullopt;
+}
+
+} // namespace onnx_mlir

src/PIM/Common/IR/ConstantUtils.hpp

@@ -0,0 +1,30 @@
+#pragma once
+
+#include "mlir/Dialect/Arith/IR/Arith.h"
+#include "mlir/IR/PatternMatch.h"
+#include "mlir/IR/Value.h"
+#include "mlir/Transforms/FoldUtils.h"
+
+#include <optional>
+
+namespace onnx_mlir {
+
+mlir::Block* getConstantInsertionBlock(mlir::Operation* anchorOp);
+
+mlir::Value
+getOrCreateConstant(mlir::OperationFolder& folder, mlir::Operation* anchorOp, mlir::Attribute value, mlir::Type type);
+
+mlir::Value
+getOrCreateConstant(mlir::RewriterBase& rewriter, mlir::Operation* anchorOp, mlir::Attribute value, mlir::Type type);
+
+mlir::Value getOrCreateConstantLike(mlir::OperationFolder& folder, mlir::arith::ConstantOp constantOp);
+
+mlir::Value getOrCreateIndexConstant(mlir::OperationFolder& folder, mlir::Operation* anchorOp, int64_t value);
+
+mlir::Value getOrCreateIndexConstant(mlir::RewriterBase& rewriter, mlir::Operation* anchorOp, int64_t value);
+
+std::optional<int64_t> matchConstantIndexValue(mlir::Value value);
+
+std::optional<int64_t> matchConstantIndexValue(mlir::OpFoldResult value);
+
+} // namespace onnx_mlir

src/PIM/Common/IR/CoreBlockUtils.cpp

@@ -1,25 +1,37 @@
 #include "mlir/Dialect/Arith/IR/Arith.h"
+#include "mlir/Dialect/MemRef/IR/MemRef.h"
 #include "mlir/Dialect/SCF/IR/SCF.h"
 
+#include "llvm/ADT/SmallVector.h"
+
 #include "src/Accelerators/PIM/Common/IR/CoreBlockUtils.hpp"
 #include "src/Accelerators/PIM/Dialect/Pim/PimOps.hpp"
 
 namespace onnx_mlir {
 
 bool isCoreStaticAddressOp(mlir::Operation* op) {
-  return mlir::isa<mlir::arith::ConstantOp,
+  if (mlir::isa<mlir::arith::ConstantOp,
                 mlir::arith::AddIOp,
                 mlir::arith::SubIOp,
                 mlir::arith::MulIOp,
                 mlir::arith::DivUIOp,
+                mlir::arith::DivSIOp,
                 mlir::arith::MinUIOp,
                 mlir::arith::RemUIOp,
+                mlir::arith::RemSIOp,
                 mlir::arith::IndexCastOp,
+                mlir::arith::CmpIOp,
                 mlir::memref::AllocOp,
                 mlir::memref::SubViewOp,
                 mlir::memref::CastOp,
                 mlir::memref::CollapseShapeOp,
-                   mlir::memref::ExpandShapeOp>(op);
+                mlir::memref::ExpandShapeOp>(op))
+    return true;
+
+  if (auto selectOp = mlir::dyn_cast<mlir::arith::SelectOp>(op))
+    return selectOp.getType().isIntOrIndex();
+
+  return false;
 }
 
 mlir::LogicalResult
@@ -30,6 +42,9 @@ walkPimCoreBlock(mlir::Block& block,
   for (mlir::Operation& op : block) {
     if (mlir::isa<pim::PimHaltOp, mlir::scf::YieldOp>(op) || isCoreStaticAddressOp(&op))
       continue;
+    if (auto loadOp = mlir::dyn_cast<mlir::memref::LoadOp>(op);
+        loadOp && succeeded(resolveIndexValue(loadOp.getResult(), knowledge)))
+      continue;
 
     if (auto forOp = mlir::dyn_cast<mlir::scf::ForOp>(op)) {
       mlir::Block& loopBody = forOp.getRegion().front();
@@ -65,4 +80,58 @@ walkPimCoreBlock(mlir::Block& block,
   return mlir::success(!hasFailure);
 }
 
+mlir::LogicalResult walkPimCoreBlockStructurally(
+  mlir::Block& block,
+  const StaticValueKnowledge& knowledge,
+  llvm::function_ref<mlir::LogicalResult(mlir::Operation&, const StaticValueKnowledge&)> callback) {
+  bool hasFailure = false;
+  for (mlir::Operation& op : block) {
+    if (mlir::isa<pim::PimHaltOp, mlir::scf::YieldOp>(op) || isCoreStaticAddressOp(&op))
+      continue;
+    if (auto loadOp = mlir::dyn_cast<mlir::memref::LoadOp>(op);
+        loadOp && succeeded(resolveIndexValue(loadOp.getResult(), knowledge)))
+      continue;
+
+    if (auto forOp = mlir::dyn_cast<mlir::scf::ForOp>(op)) {
+      mlir::Block& loopBody = forOp.getRegion().front();
+      auto lowerBound = resolveIndexValue(forOp.getLowerBound(), knowledge);
+      auto upperBound = resolveIndexValue(forOp.getUpperBound(), knowledge);
+      auto step = resolveIndexValue(forOp.getStep(), knowledge);
+      if (failed(lowerBound) || failed(upperBound) || failed(step)) {
+        forOp.emitOpError("requires statically evaluable scf.for bounds for PIM verification");
+        hasFailure = true;
+        continue;
+      }
+      if (*step <= 0) {
+        forOp.emitOpError("requires positive scf.for step for PIM verification");
+        hasFailure = true;
+        continue;
+      }
+
+      llvm::SmallVector<int64_t, 2> samples;
+      if (*lowerBound < *upperBound) {
+        samples.push_back(*lowerBound);
+        int64_t last = *lowerBound + ((*upperBound - 1 - *lowerBound) / *step) * *step;
+        if (last != *lowerBound)
+          samples.push_back(last);
+      }
+
+      for (int64_t inductionValue : samples) {
+        StaticValueKnowledge loopKnowledge = knowledge;
+        loopKnowledge.indexValues[forOp.getInductionVar()] = inductionValue;
+        for (auto [iterArg, iterValue] : llvm::zip_equal(forOp.getRegionIterArgs(), forOp.getInitArgs()))
+          loopKnowledge.aliases[iterArg] = iterValue;
+
+        if (failed(walkPimCoreBlockStructurally(loopBody, loopKnowledge, callback)))
+          hasFailure = true;
+      }
+      continue;
+    }
+
+    if (failed(callback(op, knowledge)))
+      hasFailure = true;
+  }
+  return mlir::success(!hasFailure);
+}
+
 } // namespace onnx_mlir

src/PIM/Common/IR/CoreBlockUtils.hpp

@@ -21,4 +21,12 @@ walkPimCoreBlock(mlir::Block& block,
                  const StaticValueKnowledge& knowledge,
                  llvm::function_ref<mlir::LogicalResult(mlir::Operation&, const StaticValueKnowledge&)> callback);
 
+/// Walks a `pim.core`-like body structurally for verification without
+/// enumerating full loop trip counts. Loop bounds must still be statically
+/// evaluable so address resolution remains well-defined.
+mlir::LogicalResult walkPimCoreBlockStructurally(
+  mlir::Block& block,
+  const StaticValueKnowledge& knowledge,
+  llvm::function_ref<mlir::LogicalResult(mlir::Operation&, const StaticValueKnowledge&)> callback);
+
 } // namespace onnx_mlir

src/PIM/Common/IR/ShapeUtils.cpp

@@ -1,4 +1,5 @@
 #include "llvm/ADT/STLExtras.h"
+#include "llvm/Support/ErrorHandling.h"
 
 #include "src/Accelerators/PIM/Common/IR/ShapeUtils.hpp"
 
@@ -35,6 +36,30 @@ int64_t getNumElements(llvm::ArrayRef<int64_t> shape) {
   return numElements;
 }
 
+bool hasByteSizedElementType(mlir::Type elementType) {
+  if (mlir::isa<mlir::IndexType>(elementType))
+    return true;
+  if (auto intType = mlir::dyn_cast<mlir::IntegerType>(elementType))
+    return intType.getWidth() > 0 && intType.getWidth() % 8 == 0;
+  if (auto floatType = mlir::dyn_cast<mlir::FloatType>(elementType))
+    return floatType.getWidth() > 0 && floatType.getWidth() % 8 == 0;
+  return false;
+}
+
+size_t getElementTypeSizeInBytes(mlir::Type elementType) {
+  if (mlir::isa<mlir::IndexType>(elementType))
+    return mlir::IndexType::kInternalStorageBitWidth / 8;
+  if (auto intType = mlir::dyn_cast<mlir::IntegerType>(elementType))
+    return static_cast<size_t>(intType.getWidth() / 8);
+  if (auto floatType = mlir::dyn_cast<mlir::FloatType>(elementType))
+    return static_cast<size_t>(floatType.getWidth() / 8);
+  llvm_unreachable("expected byte-sized integer, float, or index element type");
+}
+
+size_t getShapedTypeSizeInBytes(mlir::ShapedType shapedType) {
+  return static_cast<size_t>(shapedType.getNumElements()) * getElementTypeSizeInBytes(shapedType.getElementType());
+}
+
 bool isMemoryContiguous(llvm::ArrayRef<int64_t> srcShape,
                         llvm::ArrayRef<int64_t> offsets,
                         llvm::ArrayRef<int64_t> sizes,
@@ -86,4 +111,56 @@ bool isMemoryContiguous(llvm::ArrayRef<int64_t> srcShape,
   return true;
 }
 
+bool isContiguousSubviewWithDynamicOffsets(llvm::ArrayRef<int64_t> sourceShape,
+                                           llvm::ArrayRef<mlir::OpFoldResult> mixedOffsets,
+                                           llvm::ArrayRef<int64_t> staticSizes,
+                                           llvm::ArrayRef<int64_t> staticStrides) {
+  if (sourceShape.size() != mixedOffsets.size() || sourceShape.size() != staticSizes.size()
+      || sourceShape.size() != staticStrides.size()) {
+    return false;
+  }
+
+  if (llvm::any_of(staticStrides, [](int64_t stride) { return stride != 1; }))
+    return false;
+
+  auto reversedTriples =
+    llvm::zip_equal(llvm::reverse(sourceShape), llvm::reverse(mixedOffsets), llvm::reverse(staticSizes));
+
+  auto firstNonZeroOrDynamicOffset = llvm::find_if(reversedTriples, [](auto triple) {
+    auto [_sourceDim, offset, _size] = triple;
+    if (auto attr = mlir::dyn_cast<mlir::Attribute>(offset))
+      return mlir::cast<mlir::IntegerAttr>(attr).getInt() != 0;
+    return true;
+  });
+
+  if (firstNonZeroOrDynamicOffset != reversedTriples.end()) {
+    auto [sourceDim, offset, size] = *firstNonZeroOrDynamicOffset;
+    if (auto attr = mlir::dyn_cast<mlir::Attribute>(offset)) {
+      int64_t staticOffset = mlir::cast<mlir::IntegerAttr>(attr).getInt();
+      if (size > sourceDim - staticOffset)
+        return false;
+    }
+
+    ++firstNonZeroOrDynamicOffset;
+    for (auto it = firstNonZeroOrDynamicOffset; it != reversedTriples.end(); ++it)
+      if (std::get<2>(*it) != 1)
+        return false;
+  }
+
+  auto reversedSizes = llvm::zip_equal(llvm::reverse(sourceShape), llvm::reverse(staticSizes));
+  auto firstDifferentSize = llvm::find_if(reversedSizes, [](auto pair) {
+    auto [sourceDim, size] = pair;
+    return size != sourceDim;
+  });
+
+  if (firstDifferentSize != reversedSizes.end()) {
+    ++firstDifferentSize;
+    for (auto it = firstDifferentSize; it != reversedSizes.end(); ++it)
+      if (std::get<1>(*it) != 1)
+        return false;
+  }
+
+  return true;
+}
+
 } // namespace onnx_mlir

src/PIM/Common/IR/ShapeUtils.hpp

@@ -1,8 +1,14 @@
 #pragma once
 
+#include "mlir/IR/BuiltinTypes.h"
+#include "mlir/IR/OpDefinition.h"
+#include "mlir/IR/Value.h"
+
 #include "llvm/ADT/ArrayRef.h"
 #include "llvm/ADT/SmallVector.h"
 
+#include <cstddef>
+
 namespace onnx_mlir {
 
 llvm::SmallVector<int64_t> computeRowMajorStrides(llvm::ArrayRef<int64_t> shape);
@@ -14,9 +20,20 @@ int64_t linearizeIndex(llvm::ArrayRef<int64_t> indices, llvm::ArrayRef<int64_t>
 
 int64_t getNumElements(llvm::ArrayRef<int64_t> shape);
 
+bool hasByteSizedElementType(mlir::Type elementType);
+
+size_t getElementTypeSizeInBytes(mlir::Type elementType);
+
+size_t getShapedTypeSizeInBytes(mlir::ShapedType shapedType);
+
 bool isMemoryContiguous(llvm::ArrayRef<int64_t> srcShape,
                         llvm::ArrayRef<int64_t> offsets,
                         llvm::ArrayRef<int64_t> sizes,
                         llvm::ArrayRef<int64_t> strides);
 
+bool isContiguousSubviewWithDynamicOffsets(llvm::ArrayRef<int64_t> sourceShape,
+                                           llvm::ArrayRef<mlir::OpFoldResult> mixedOffsets,
+                                           llvm::ArrayRef<int64_t> staticSizes,
+                                           llvm::ArrayRef<int64_t> staticStrides);
+
 } // namespace onnx_mlir

src/PIM/Common/IR/SubviewUtils.cpp

@@ -31,6 +31,19 @@ Value stripMemRefViewOps(Value value) {
   }
 }
 
+Value stripMemRefAddressingOps(Value value) {
+  while (true) {
+    if (auto subviewOp = value.getDefiningOp<memref::SubViewOp>()) {
+      value = subviewOp.getSource();
+      continue;
+    }
+    Value strippedValue = stripMemRefViewOps(value);
+    if (strippedValue == value)
+      return value;
+    value = strippedValue;
+  }
+}
+
 bool hasAllStaticSubviewParts(memref::SubViewOp subview) {
   return llvm::all_of(subview.getStaticOffsets(), [](int64_t value) { return !ShapedType::isDynamic(value); })
       && llvm::all_of(subview.getStaticSizes(), [](int64_t value) { return !ShapedType::isDynamic(value); })
@@ -81,4 +94,13 @@ FailureOr<SmallVector<int64_t>> getStaticSubviewOffsets(const StaticSubviewInfo&
   return staticOffsets;
 }
 
+bool isMemRefBaseAddressableValue(Value value) {
+  value = stripMemRefAddressingOps(value);
+  if (isa<BlockArgument>(value))
+    return true;
+
+  Operation* defOp = value.getDefiningOp();
+  return defOp && isa<memref::AllocOp, memref::GetGlobalOp>(defOp);
+}
+
 } // namespace onnx_mlir

src/PIM/Common/IR/SubviewUtils.hpp

@@ -20,6 +20,8 @@ mlir::Value stripMemRefCasts(mlir::Value value);
 
 mlir::Value stripMemRefViewOps(mlir::Value value);
 
+mlir::Value stripMemRefAddressingOps(mlir::Value value);
+
 bool hasAllStaticSubviewParts(mlir::memref::SubViewOp subview);
 
 llvm::FailureOr<StaticSubviewInfo> getStaticSubviewInfo(mlir::Value value);
@@ -27,4 +29,6 @@ llvm::FailureOr<StaticSubviewInfo> getStaticSubviewInfo(mlir::Value value);
 /// Returns the offsets in `info` as int64_t, failing if any offset is dynamic.
 llvm::FailureOr<llvm::SmallVector<int64_t>> getStaticSubviewOffsets(const StaticSubviewInfo& info);
 
+bool isMemRefBaseAddressableValue(mlir::Value value);
+
 } // namespace onnx_mlir

src/PIM/Common/IR/WeightUtils.cpp

@@ -1,8 +1,14 @@
+#include "mlir/Dialect/Linalg/IR/Linalg.h"
 #include "mlir/Dialect/Tensor/IR/Tensor.h"
+#include "mlir/IR/BuiltinAttributes.h"
+#include "mlir/IR/BuiltinTypes.h"
 
 #include "llvm/ADT/SmallPtrSet.h"
 #include "llvm/ADT/SmallSet.h"
 
+#include "src/Accelerators/PIM/Common/IR/AddressAnalysis.hpp"
+#include "src/Accelerators/PIM/Common/IR/ShapeUtils.hpp"
+#include "src/Accelerators/PIM/Common/IR/SubviewUtils.hpp"
 #include "src/Accelerators/PIM/Common/IR/WeightUtils.hpp"
 #include "src/Accelerators/PIM/Dialect/Pim/PimOps.hpp"
 #include "src/Accelerators/PIM/Dialect/Spatial/SpatialOps.hpp"
@@ -19,29 +25,57 @@ void markWeightAlways(mlir::Operation* op) {
 
 namespace {
 
-template <typename MVMOpTy, typename VMMOpTy, typename ParentOpTy>
-bool hasMvmVmmWeightUse(ParentOpTy parentOp, unsigned weightIndex) {
+CompiledIndexExpr makeConstantExpr(int64_t constant) {
+  CompiledIndexExprNode expr;
+  expr.kind = CompiledIndexExprNode::Kind::Constant;
+  expr.constant = constant;
+  return CompiledIndexExpr(std::make_shared<CompiledIndexExprNode>(std::move(expr)));
+}
+
+CompiledIndexExpr makeBinaryExpr(CompiledIndexExprNode::Kind kind, CompiledIndexExpr lhs, CompiledIndexExpr rhs) {
+  CompiledIndexExprNode expr;
+  expr.kind = kind;
+  expr.operands = {std::move(lhs), std::move(rhs)};
+  return CompiledIndexExpr(std::make_shared<CompiledIndexExprNode>(std::move(expr)));
+}
+
+CompiledIndexExpr addExpr(CompiledIndexExpr lhs, CompiledIndexExpr rhs) {
+  return makeBinaryExpr(CompiledIndexExprNode::Kind::Add, std::move(lhs), std::move(rhs));
+}
+
+CompiledIndexExpr mulExpr(CompiledIndexExpr lhs, int64_t rhs) {
+  return makeBinaryExpr(CompiledIndexExprNode::Kind::Mul, std::move(lhs), makeConstantExpr(rhs));
+}
+
+template <typename VMMOpTy, typename ParentOpTy>
+bool hasVmmWeightUse(ParentOpTy parentOp, unsigned weightIndex) {
+  auto weightArg = parentOp.getWeightArgument(weightIndex);
+  if (!weightArg)
+    return false;
   bool found = false;
   parentOp.walk([&](mlir::Operation* op) {
-    if (auto mvmOp = mlir::dyn_cast<MVMOpTy>(op))
-      found |= mvmOp.getWeightIndex() == weightIndex;
-    else if (auto vmmOp = mlir::dyn_cast<VMMOpTy>(op))
-      found |= vmmOp.getWeightIndex() == weightIndex;
+    if (auto vmmOp = mlir::dyn_cast<VMMOpTy>(op))
+      found |= vmmOp.getWeight() == *weightArg;
   });
   return found;
 }
 
-template <typename MVMOpTy, typename VMMOpTy, typename ParentOpTy>
-void walkMvmVmmWeightUses(ParentOpTy parentOp, llvm::function_ref<void(mlir::OpOperand&)> callback) {
+template <typename VMMOpTy, typename ParentOpTy>
+void walkVmmWeightUses(ParentOpTy parentOp, llvm::function_ref<void(mlir::OpOperand&)> callback) {
   auto weights = parentOp.getWeights();
   llvm::SmallSet<unsigned, 8> visited;
-  auto walkWeightIndex = [&](unsigned weightIndex) {
-    if (weightIndex < weights.size() && visited.insert(weightIndex).second)
+  auto walkWeight = [&](mlir::Value weight) {
+    for (unsigned weightIndex = 0; weightIndex < weights.size(); ++weightIndex) {
+      auto weightArg = parentOp.getWeightArgument(weightIndex);
+      if (!weightArg || *weightArg != weight)
+        continue;
+      if (visited.insert(weightIndex).second)
         callback(parentOp->getOpOperand(weightIndex));
+      break;
+    }
   };
 
-  parentOp.walk([&](MVMOpTy op) { walkWeightIndex(op.getWeightIndex()); });
-  parentOp.walk([&](VMMOpTy op) { walkWeightIndex(op.getWeightIndex()); });
+  parentOp.walk([&](VMMOpTy op) { walkWeight(op.getWeight()); });
 }
 
 } // namespace
@@ -54,7 +88,7 @@ bool isSpatialMvmVmmWeightUse(mlir::OpOperand& use) {
   if (!computeOp || operandIndex >= computeOp.getWeights().size())
     return false;
 
-  return hasMvmVmmWeightUse<spatial::SpatMVMOp, spatial::SpatVMMOp>(computeOp, operandIndex);
+  return hasVmmWeightUse<spatial::SpatVMMOp>(computeOp, operandIndex);
 }
 
 bool hasOnlySpatialMvmVmmWeightUses(mlir::Value value) {
@@ -76,8 +110,8 @@ bool hasOnlySpatialMvmVmmWeightUses(mlir::Value value) {
         return expandShapeOp.getSrc() == currentValue && self(expandShapeOp.getResult(), self);
       if (auto collapseShapeOp = mlir::dyn_cast<mlir::tensor::CollapseShapeOp>(user))
         return collapseShapeOp.getSrc() == currentValue && self(collapseShapeOp.getResult(), self);
-      if (auto transposeOp = mlir::dyn_cast<mlir::ONNXTransposeOp>(user))
-        return transposeOp.getData() == currentValue && self(transposeOp.getResult(), self);
+      if (auto transposeOp = mlir::dyn_cast<mlir::linalg::TransposeOp>(user))
+        return transposeOp.getInput() == currentValue && self(transposeOp.getResult()[0], self);
 
       return false;
     });
@@ -90,19 +124,149 @@ void walkPimMvmVmmWeightUses(mlir::Operation* root, llvm::function_ref<void(mlir
   assert(root && "expected valid root op");
   root->walk([&](pim::PimCoreOp coreOp) {
     coreOp.walk([&](pim::PimVMMOp vmmOp) {
-      auto weights = coreOp.getWeights();
-      unsigned weightIndex = vmmOp.getWeightIndex();
-      if (weightIndex < weights.size())
-        callback(coreOp->getOpOperand(weightIndex));
+      if (auto weightIndex = resolveWeightIndex(coreOp.getOperation(), vmmOp.getWeight()))
+        callback(coreOp->getOpOperand(*weightIndex));
     });
   });
   root->walk([&](pim::PimCoreBatchOp coreBatchOp) {
-    auto weights = coreBatchOp.getWeights();
-    for (auto weight : weights)
-      for (mlir::OpOperand& use : weight.getUses())
-        if (use.getOwner() == coreBatchOp.getOperation())
-          callback(use);
+    coreBatchOp.walk([&](pim::PimVMMOp vmmOp) {
+      if (auto weightIndex = resolveWeightIndex(coreBatchOp.getOperation(), vmmOp.getWeight()))
+        callback(coreBatchOp->getOpOperand(*weightIndex));
+    });
   });
 }
 
+std::optional<unsigned> resolveWeightIndex(mlir::Operation* weightOwner, mlir::Value weight) {
+  weight = stripMemRefAddressingOps(weight);
+
+  if (auto coreOp = mlir::dyn_cast_or_null<pim::PimCoreOp>(weightOwner)) {
+    for (unsigned weightIndex = 0; weightIndex < coreOp.getWeights().size(); ++weightIndex)
+      if (coreOp.getWeightArgument(weightIndex) == weight)
+        return weightIndex;
+    return std::nullopt;
+  }
+
+  if (auto coreBatchOp = mlir::dyn_cast_or_null<pim::PimCoreBatchOp>(weightOwner)) {
+    for (unsigned weightIndex = 0; weightIndex < coreBatchOp.getWeights().size(); ++weightIndex)
+      if (coreBatchOp.getWeightArgument(weightIndex) == weight)
+        return weightIndex;
+    return std::nullopt;
+  }
+
+  return std::nullopt;
+}
+
+llvm::FailureOr<ResolvedWeightView>
+resolveWeightView(mlir::Operation* weightOwner, mlir::Value weight, const StaticValueKnowledge& knowledge) {
+  llvm::SmallVector<mlir::Operation*> viewOps;
+  mlir::Value current = weight;
+
+  while (true) {
+    if (auto defOp = current.getDefiningOp()) {
+      if (auto getGlobalOp = mlir::dyn_cast<mlir::memref::GetGlobalOp>(defOp)) {
+        auto moduleOp = weightOwner ? weightOwner->getParentOfType<mlir::ModuleOp>() : mlir::ModuleOp {};
+        auto globalOp = lookupGlobalForGetGlobal(moduleOp, getGlobalOp);
+        if (!globalOp || !globalOp.getInitialValue())
+          return mlir::failure();
+
+        auto denseAttr = mlir::dyn_cast<mlir::DenseElementsAttr>(*globalOp.getInitialValue());
+        if (!denseAttr)
+          return mlir::failure();
+
+        ResolvedWeightView view;
+        view.globalOp = globalOp;
+        view.shape.assign(denseAttr.getType().getShape().begin(), denseAttr.getType().getShape().end());
+        view.strides = computeRowMajorStrides(view.shape);
+
+        CompiledIndexExpr offsetExpr = makeConstantExpr(0);
+        for (mlir::Operation* viewOp : llvm::reverse(viewOps)) {
+          if (auto subview = mlir::dyn_cast<mlir::memref::SubViewOp>(viewOp)) {
+            llvm::SmallVector<int64_t> nextStrides;
+            nextStrides.reserve(subview.getMixedOffsets().size());
+            for (auto [offset, stride, sourceStride] :
+                 llvm::zip_equal(subview.getMixedOffsets(), subview.getStaticStrides(), view.strides)) {
+              CompiledIndexExpr offsetValue = makeConstantExpr(0);
+              if (auto attr = mlir::dyn_cast<mlir::Attribute>(offset)) {
+                auto intAttr = mlir::dyn_cast<mlir::IntegerAttr>(attr);
+                if (!intAttr)
+                  return mlir::failure();
+                offsetValue = makeConstantExpr(intAttr.getInt());
+              }
+              else if (auto value = mlir::dyn_cast<mlir::Value>(offset)) {
+                auto compiledOffset = compileIndexExpr(value);
+                if (failed(compiledOffset))
+                  return mlir::failure();
+                offsetValue = *compiledOffset;
+              }
+              else {
+                return mlir::failure();
+              }
+              offsetExpr = addExpr(std::move(offsetExpr), mulExpr(std::move(offsetValue), sourceStride));
+              nextStrides.push_back(stride * sourceStride);
+            }
+            view.shape.assign(subview.getStaticSizes().begin(), subview.getStaticSizes().end());
+            view.strides = std::move(nextStrides);
+            continue;
+          }
+
+          if (auto collapse = mlir::dyn_cast<mlir::memref::CollapseShapeOp>(viewOp)) {
+            if (view.strides != computeRowMajorStrides(view.shape))
+              return mlir::failure();
+            auto resultType = mlir::cast<mlir::MemRefType>(collapse.getResult().getType());
+            view.shape.assign(resultType.getShape().begin(), resultType.getShape().end());
+            view.strides = computeRowMajorStrides(view.shape);
+            continue;
+          }
+
+          if (auto expand = mlir::dyn_cast<mlir::memref::ExpandShapeOp>(viewOp)) {
+            if (view.strides != computeRowMajorStrides(view.shape))
+              return mlir::failure();
+            auto resultType = mlir::cast<mlir::MemRefType>(expand.getResult().getType());
+            view.shape.assign(resultType.getShape().begin(), resultType.getShape().end());
+            view.strides = computeRowMajorStrides(view.shape);
+          }
+        }
+
+        auto resolvedOffset = offsetExpr.evaluate(knowledge);
+        if (failed(resolvedOffset))
+          return mlir::failure();
+        view.offset = *resolvedOffset;
+        return view;
+      }
+
+      if (mlir::isa<mlir::memref::SubViewOp, mlir::memref::CollapseShapeOp, mlir::memref::ExpandShapeOp>(defOp)) {
+        viewOps.push_back(defOp);
+        if (auto subview = mlir::dyn_cast<mlir::memref::SubViewOp>(defOp))
+          current = subview.getSource();
+        else if (auto collapse = mlir::dyn_cast<mlir::memref::CollapseShapeOp>(defOp))
+          current = collapse.getSrc();
+        else
+          current = mlir::cast<mlir::memref::ExpandShapeOp>(defOp).getSrc();
+        continue;
+      }
+
+      if (auto castOp = mlir::dyn_cast<mlir::memref::CastOp>(defOp)) {
+        current = castOp.getSource();
+        continue;
+      }
+
+      return mlir::failure();
+    }
+
+    auto weightIndex = resolveWeightIndex(weightOwner, current);
+    if (!weightIndex)
+      return mlir::failure();
+
+    if (auto coreOp = mlir::dyn_cast_or_null<pim::PimCoreOp>(weightOwner)) {
+      current = coreOp.getWeights()[*weightIndex];
+      continue;
+    }
+    if (auto coreBatchOp = mlir::dyn_cast_or_null<pim::PimCoreBatchOp>(weightOwner)) {
+      current = coreBatchOp.getWeights()[*weightIndex];
+      continue;
+    }
+    return mlir::failure();
+  }
+}
+
 } // namespace onnx_mlir

src/PIM/Common/IR/WeightUtils.hpp

@@ -1,15 +1,34 @@
 #pragma once
 
-#include "mlir/IR/Operation.h"
+#include "mlir/Dialect/MemRef/IR/MemRef.h"
+#include "mlir/IR/BuiltinOps.h"
 #include "mlir/IR/Value.h"
 
+#include "llvm/ADT/STLExtras.h"
 #include "llvm/ADT/STLFunctionalExtras.h"
+#include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/StringRef.h"
 
+#include <optional>
+
+#include "src/Accelerators/PIM/Common/IR/AddressAnalysis.hpp"
+#include "src/Accelerators/PIM/Dialect/Pim/PimOps.hpp"
+
 inline constexpr llvm::StringRef PimWeightAlwaysAttrName = "weightAlways";
 
 namespace onnx_mlir {
 
+struct ResolvedWeightView {
+  mlir::memref::GlobalOp globalOp;
+  llvm::SmallVector<int64_t> shape;
+  llvm::SmallVector<int64_t> strides;
+  int64_t offset = 0;
+
+  bool operator==(const ResolvedWeightView& other) const {
+    return globalOp == other.globalOp && shape == other.shape && strides == other.strides && offset == other.offset;
+  }
+};
+
 bool hasWeightAlways(mlir::Operation* op);
 
 /// Tags an op as producing a value that should stay materialized as a reusable
@@ -26,4 +45,20 @@ bool hasOnlySpatialMvmVmmWeightUses(mlir::Value value);
 /// passes can identify globals that must remain weight-backed.
 void walkPimMvmVmmWeightUses(mlir::Operation* root, llvm::function_ref<void(mlir::OpOperand&)> callback);
 
+std::optional<unsigned> resolveWeightIndex(mlir::Operation* weightOwner, mlir::Value weight);
+llvm::FailureOr<ResolvedWeightView>
+resolveWeightView(mlir::Operation* weightOwner, mlir::Value weight, const StaticValueKnowledge& knowledge = {});
+
+template <typename CoreLikeOpTy>
+llvm::SmallVector<unsigned, 8> getUsedWeightIndices(CoreLikeOpTy coreLikeOp) {
+  llvm::SmallVector<unsigned, 8> indices;
+  coreLikeOp.walk([&](pim::PimVMMOp vmmOp) {
+    auto weightIndex = resolveWeightIndex(coreLikeOp.getOperation(), vmmOp.getWeight());
+    if (weightIndex && !llvm::is_contained(indices, *weightIndex))
+      indices.push_back(*weightIndex);
+  });
+  llvm::sort(indices);
+  return indices;
+}
+
 } // namespace onnx_mlir

src/PIM/Common/PimCommon.hpp

@@ -12,6 +12,7 @@
 #include "llvm/ADT/StringRef.h"
 
 #include "src/Accelerators/PIM/Common/IR/AddressAnalysis.hpp"
+#include "src/Accelerators/PIM/Common/IR/ConstantUtils.hpp"
 #include "src/Accelerators/PIM/Common/IR/CoreBlockUtils.hpp"
 #include "src/Accelerators/PIM/Common/IR/EntryPointUtils.hpp"
 #include "src/Accelerators/PIM/Common/IR/ShapeUtils.hpp"

src/PIM/Common/Support/DebugDump.cpp

@@ -18,7 +18,7 @@ void dumpModule(mlir::ModuleOp moduleOp, const std::string& name) {
   std::fstream file(dialectsDir + "/" + name + ".mlir", std::ios::out);
   llvm::raw_os_ostream os(file);
   mlir::OpPrintingFlags flags;
-  flags.elideLargeElementsAttrs();
+  flags.elideLargeElementsAttrs().enableDebugInfo(true, false);
   moduleOp.print(os, flags);
   os.flush();
   file.close();

src/PIM/Common/Support/Diagnostics.hpp

@@ -13,7 +13,8 @@
 namespace onnx_mlir::pim {
 
 struct CappedDiagnosticReporter {
-  explicit CappedDiagnosticReporter(int64_t maxReportedFailures = 8) : maxReportedFailures(maxReportedFailures) {}
+  explicit CappedDiagnosticReporter(int64_t maxReportedFailures = 8)
+  : maxReportedFailures(maxReportedFailures) {}
 
   template <typename EmitFn>
   void report(mlir::Operation* op, EmitFn&& emit) {
@@ -24,8 +25,7 @@ struct CappedDiagnosticReporter {
 
   void emitSuppressedSummary(mlir::Operation* op, llvm::StringRef failureDescription) const {
     if (numFailures > maxReportedFailures)
-      op->emitError() << "suppressed " << (numFailures - maxReportedFailures) << " additional "
-                      << failureDescription;
+      op->emitError() << "suppressed " << (numFailures - maxReportedFailures) << " additional " << failureDescription;
   }
 
   bool hasFailure() const { return numFailures != 0; }

src/PIM/Compiler/CMakeLists.txt

@@ -16,7 +16,6 @@ add_pim_library(OMPimCompilerOptions
 add_pim_library(OMPimCompilerUtils
   PimCompilerUtils.cpp
   PimArtifactWriter.cpp
-  PimBatchEmission.cpp
   PimCodeGen.cpp
   PimWeightEmitter.cpp
 

src/PIM/Compiler/PimArtifactWriter.cpp

@@ -48,7 +48,7 @@ writeMemoryBinary(ModuleOp moduleOp, func::FuncOp funcOp, PimAcceleratorMemory&
     if (!denseAttr)
       return;
 
-    MemEntry memEntry = memory.hostMem.getMemEntry(getGlobalOp.getResult());
+    MemEntry memEntry = memory.hostMem.getMemEntry({getGlobalOp.getResult(), std::nullopt});
     ArrayRef<char> rawData = denseAttr.getRawData();
     char* dst = memoryBuffer.data() + memEntry.address;
 

src/PIM/Compiler/PimBatchEmission.cpp

@@ -1,136 +0,0 @@
-#include "mlir/IR/Builders.h"
-#include "mlir/IR/BuiltinOps.h"
-#include "mlir/IR/IRMapping.h"
-
-#include "src/Accelerators/PIM/Common/PimCommon.hpp"
-#include "src/Accelerators/PIM/Compiler/PimBatchEmission.hpp"
-
-using namespace mlir;
-
-namespace onnx_mlir {
-namespace {
-
-static SmallVector<int32_t> getBatchCoreIds(pim::PimCoreBatchOp coreBatchOp) {
-  auto coreIdsAttr = coreBatchOp->getAttrOfType<DenseI32ArrayAttr>(onnx_mlir::kCoreIdsAttrName);
-  assert(coreIdsAttr && "pim.core_batch requires coreIds array attribute");
-  return SmallVector<int32_t>(coreIdsAttr.asArrayRef().begin(), coreIdsAttr.asArrayRef().end());
-}
-
-static SmallVector<int32_t> getLaneChunkCoreIds(ArrayRef<int32_t> coreIds, size_t laneCount, unsigned lane) {
-  SmallVector<int32_t> laneCoreIds;
-  laneCoreIds.reserve(coreIds.size() / laneCount);
-  for (size_t chunkIndex = 0; chunkIndex < coreIds.size() / laneCount; ++chunkIndex)
-    laneCoreIds.push_back(coreIds[chunkIndex * laneCount + lane]);
-  return laneCoreIds;
-}
-
-static void scalarizeBatchOpsInCore(pim::PimCoreOp scalarCore, size_t laneCount, unsigned lane) {
-  IRRewriter rewriter(scalarCore.getContext());
-  SmallVector<Operation*> batchOps;
-  scalarCore.walk([&](Operation* op) {
-    if (isa<pim::PimSendBatchOp,
-            pim::PimSendTensorBatchOp,
-            pim::PimReceiveBatchOp,
-            pim::PimReceiveTensorBatchOp,
-            pim::PimMemCopyHostToDevBatchOp>(op)) {
-      batchOps.push_back(op);
-    }
-  });
-
-  for (Operation* op : batchOps) {
-    rewriter.setInsertionPoint(op);
-
-    if (auto sendBatchOp = dyn_cast<pim::PimSendBatchOp>(op)) {
-      pim::PimSendOp::create(rewriter,
-                             sendBatchOp.getLoc(),
-                             sendBatchOp.getInput(),
-                             sendBatchOp.getSizeAttr(),
-                             rewriter.getI32IntegerAttr(sendBatchOp.getTargetCoreIds()[lane]));
-      rewriter.eraseOp(op);
-      continue;
-    }
-
-    if (auto sendTensorBatchOp = dyn_cast<pim::PimSendTensorBatchOp>(op)) {
-      pim::PimSendTensorOp::create(
-        rewriter,
-        sendTensorBatchOp.getLoc(),
-        sendTensorBatchOp.getInput(),
-        rewriter.getDenseI32ArrayAttr(getLaneChunkCoreIds(sendTensorBatchOp.getTargetCoreIds(), laneCount, lane)));
-      rewriter.eraseOp(op);
-      continue;
-    }
-
-    if (auto receiveBatchOp = dyn_cast<pim::PimReceiveBatchOp>(op)) {
-      auto scalarReceive =
-        pim::PimReceiveOp::create(rewriter,
-                                  receiveBatchOp.getLoc(),
-                                  receiveBatchOp.getOutput().getType(),
-                                  receiveBatchOp.getOutputBuffer(),
-                                  receiveBatchOp.getSizeAttr(),
-                                  rewriter.getI32IntegerAttr(receiveBatchOp.getSourceCoreIds()[lane]));
-      rewriter.replaceOp(op, scalarReceive->getResults());
-      continue;
-    }
-
-    if (auto receiveTensorBatchOp = dyn_cast<pim::PimReceiveTensorBatchOp>(op)) {
-      auto scalarReceive = pim::PimReceiveTensorOp::create(
-        rewriter,
-        receiveTensorBatchOp.getLoc(),
-        receiveTensorBatchOp.getOutput().getType(),
-        receiveTensorBatchOp.getOutputBuffer(),
-        rewriter.getDenseI32ArrayAttr(getLaneChunkCoreIds(receiveTensorBatchOp.getSourceCoreIds(), laneCount, lane)));
-      rewriter.replaceOp(op, scalarReceive->getResults());
-      continue;
-    }
-
-    auto memcpBatchOp = cast<pim::PimMemCopyHostToDevBatchOp>(op);
-    auto scalarCopy = pim::PimMemCopyHostToDevOp::create(rewriter,
-                                                         memcpBatchOp.getLoc(),
-                                                         memcpBatchOp.getOutput().getType(),
-                                                         memcpBatchOp.getDeviceTarget(),
-                                                         memcpBatchOp.getHostSource(),
-                                                         memcpBatchOp.getDeviceTargetOffsetAttr(),
-                                                         memcpBatchOp.getHostSourceOffsetAttr(),
-                                                         memcpBatchOp.getSizeAttr());
-    rewriter.replaceOp(op, scalarCopy->getResults());
-  }
-}
-
-} // namespace
-
-LogicalResult withScalarCoreFromBatchLane(pim::PimCoreBatchOp coreBatchOp,
-                                          unsigned lane,
-                                          llvm::function_ref<LogicalResult(pim::PimCoreOp)> callback) {
-  OwningOpRef<ModuleOp> scratchModule = ModuleOp::create(coreBatchOp.getLoc());
-  OpBuilder builder(scratchModule->getContext());
-  builder.setInsertionPointToStart(scratchModule->getBody());
-
-  size_t laneCount = static_cast<size_t>(coreBatchOp.getLaneCount());
-  size_t weightsPerLane = coreBatchOp.getWeights().size() / laneCount;
-  SmallVector<Value> laneWeights;
-  laneWeights.reserve(weightsPerLane);
-  for (size_t weightIndex = 0; weightIndex < weightsPerLane; ++weightIndex)
-    laneWeights.push_back(coreBatchOp.getWeights()[lane * weightsPerLane + weightIndex]);
-
-  auto coreIds = getBatchCoreIds(coreBatchOp);
-  auto scalarCore = pim::PimCoreOp::create(
-    builder, coreBatchOp.getLoc(), ValueRange(laneWeights), builder.getI32IntegerAttr(coreIds[lane]));
-  Block* block = builder.createBlock(&scalarCore.getBody(), scalarCore.getBody().end());
-  IRMapping mapper;
-  if (coreBatchOp.getBody().front().getNumArguments() == 1)
-    mapper.map(coreBatchOp.getBody().front().getArgument(0), coreBatchOp.getInputs()[lane]);
-
-  builder.setInsertionPointToEnd(block);
-  for (Operation& op : coreBatchOp.getBody().front()) {
-    Operation* cloned = builder.clone(op, mapper);
-    for (auto [originalResult, clonedResult] : llvm::zip(op.getResults(), cloned->getResults()))
-      mapper.map(originalResult, clonedResult);
-  }
-
-  if (block->empty() || !isa<pim::PimHaltOp>(block->back()))
-    pim::PimHaltOp::create(builder, coreBatchOp.getLoc());
-  scalarizeBatchOpsInCore(scalarCore, laneCount, lane);
-  return callback(scalarCore);
-}
-
-} // namespace onnx_mlir

src/PIM/Compiler/PimBatchEmission.hpp

@@ -1,13 +0,0 @@
-#pragma once
-
-#include "llvm/ADT/STLFunctionalExtras.h"
-
-#include "src/Accelerators/PIM/Dialect/Pim/PimOps.hpp"
-
-namespace onnx_mlir {
-
-mlir::LogicalResult withScalarCoreFromBatchLane(pim::PimCoreBatchOp coreBatchOp,
-                                                unsigned lane,
-                                                llvm::function_ref<mlir::LogicalResult(pim::PimCoreOp)> callback);
-
-} // namespace onnx_mlir

src/PIM/Compiler/PimCodeGen.hpp

@@ -4,13 +4,16 @@
 
 #include "llvm-project/clang/include/clang/Basic/LLVM.h"
 #include "llvm/ADT/DenseMap.h"
+#include "llvm/ADT/Hashing.h"
 #include "llvm/Support/JSON.h"
 #include "llvm/Support/raw_os_ostream.h"
 
 #include <fstream>
+#include <limits>
 #include <optional>
 
 #include "onnx-mlir/Compiler/OMCompilerTypes.h"
+#include "src/Accelerators/PIM/Common/IR/AddressAnalysis.hpp"
 #include "src/Accelerators/PIM/Common/PimCommon.hpp"
 #include "src/Accelerators/PIM/Common/Support/ReportUtils.hpp"
 #include "src/Accelerators/PIM/Compiler/PimBinaryFormat.hpp"
@@ -23,6 +26,13 @@ struct MemEntry {
   size_t size;
 };
 
+struct MemoryValueKey {
+  mlir::Value value;
+  std::optional<unsigned> lane;
+
+  bool operator==(const MemoryValueKey& other) const { return value == other.value && lane == other.lane; }
+};
+
 struct MemoryReportRow {
   uint64_t numAlloca = 0;
   uint64_t sizeAlloca = 0;
@@ -50,33 +60,33 @@ struct MemoryReportEntry {
 };
 
 class PimMemory {
-  llvm::SmallVector<std::pair<MemEntry, mlir::Value>, 32> memEntries;
-  llvm::SmallDenseMap<mlir::Value, MemEntry, 32>& globalMemEntriesMap;
-  llvm::SmallDenseMap<mlir::Value, MemEntry, 32> ownedMemEntriesMap;
+  llvm::SmallVector<std::pair<MemEntry, MemoryValueKey>, 32> memEntries;
+  llvm::SmallDenseMap<MemoryValueKey, MemEntry, 32>& globalMemEntriesMap;
+  llvm::SmallDenseMap<MemoryValueKey, MemEntry, 32> ownedMemEntriesMap;
 
   size_t minAlignment = 4;
   size_t firstAvailableAddress = 0;
 
-  MemEntry* gatherMemEntry(mlir::Value value);
+  MemEntry* gatherMemEntry(mlir::Value value, std::optional<unsigned> lane = std::nullopt);
   void allocateGatheredMemory();
-  void allocateMemoryForValue(mlir::Value value, MemEntry& memEntry);
+  void allocateMemoryForValue(const MemoryValueKey& key, MemEntry& memEntry);
 
 public:
-  PimMemory(llvm::SmallDenseMap<mlir::Value, MemEntry, 32>& globalMemEntriesMap)
+  PimMemory(llvm::SmallDenseMap<MemoryValueKey, MemEntry, 32>& globalMemEntriesMap)
   : globalMemEntriesMap(globalMemEntriesMap) {}
 
   void allocateHost(mlir::ModuleOp moduleOp, mlir::func::FuncOp funcOp);
-  void allocateCore(mlir::Operation* op);
+  void allocateCore(mlir::Operation* op, std::optional<unsigned> lane = std::nullopt);
   MemoryReportRow getReportRow() const;
   void remove(mlir::Value val);
 
   size_t getFirstAvailableAddress() const { return firstAvailableAddress; }
-  MemEntry getMemEntry(mlir::Value value) const;
+  MemEntry getMemEntry(const MemoryValueKey& key) const;
 };
 
 class PimAcceleratorMemory {
 public:
-  llvm::SmallDenseMap<mlir::Value, MemEntry, 32> memEntriesMap;
+  llvm::SmallDenseMap<MemoryValueKey, MemEntry, 32> memEntriesMap;
   PimMemory hostMem;
 
 private:
@@ -84,14 +94,23 @@ private:
   std::fstream fileReport;
   std::optional<MemoryReportRow> hostReportRow;
   llvm::SmallVector<MemoryReportEntry, 32> reportEntries;
+  mutable llvm::DenseMap<mlir::Value, CompiledIndexExpr> compiledIndexExprs;
+  mutable llvm::DenseMap<mlir::Value, CompiledAddressExpr> compiledAddressExprs;
 
 public:
   PimAcceleratorMemory()
   : hostMem(memEntriesMap), fileReport(openReportFile("memory_report")) {}
+  PimAcceleratorMemory(const llvm::SmallDenseMap<MemoryValueKey, MemEntry, 32>& initialMemEntries, bool enableReport)
+  : memEntriesMap(initialMemEntries),
+    hostMem(memEntriesMap),
+    fileReport(enableReport ? openReportFile("memory_report") : std::fstream()) {}
 
   PimMemory& getOrCreateDeviceMem(size_t id);
 
-  size_t getValueAddress(mlir::Value value, const StaticValueKnowledge& knowledge = {}) const;
+  size_t getValueAddress(mlir::Value value,
+                         const StaticValueKnowledge& knowledge = {},
+                         std::optional<unsigned> lane = std::nullopt) const;
+  llvm::FailureOr<int64_t> getIndexValue(mlir::Value value, const StaticValueKnowledge& knowledge = {}) const;
   void reportHost();
   void recordCoreReport(size_t coreId, const MemoryReportRow& row);
   void recordBatchReport(uint64_t batchId,
@@ -103,15 +122,24 @@ public:
   void clean(mlir::Operation* op);
 };
 
+struct CoreEmissionJob {
+  mlir::Operation* coreLikeOp = nullptr;
+  size_t originalCoreId = 0;
+  size_t emittedCoreId = 0;
+  llvm::SmallVector<unsigned, 4> lanes;
+  std::optional<uint64_t> batchReportId;
+};
+
 class PimCodeGen {
   PimAcceleratorMemory& memory;
   llvm::raw_fd_ostream& coreBinaryStream;
   llvm::raw_fd_ostream* coreJsonStream;
   const llvm::DenseMap<size_t, size_t>& emittedCoreIds;
+  std::optional<unsigned> batchLane;
   mutable uint32_t emittedInstructionCount = 0;
 
   size_t addressOf(mlir::Value value, const StaticValueKnowledge& knowledge) const {
-    return memory.getValueAddress(value, knowledge);
+    return memory.getValueAddress(value, knowledge, batchLane);
   }
   size_t remapCoreId(size_t coreId) const;
 
@@ -141,15 +169,17 @@ public:
   : memory(memory), coreBinaryStream(coreBinary), coreJsonStream(coreJson), emittedCoreIds(emittedCoreIds) {}
 
   uint32_t getEmittedInstructionCount() const { return emittedInstructionCount; }
+  void setBatchLane(std::optional<unsigned> lane) { batchLane = lane; }
+  llvm::FailureOr<int64_t> indexOf(mlir::Value value, const StaticValueKnowledge& knowledge) const {
+    return memory.getIndexValue(value, knowledge);
+  }
 
   void codeGenLoadOp(pim::PimMemCopyHostToDevOp loadOp, const StaticValueKnowledge& knowledge) const;
   void codeGenStoreOp(pim::PimMemCopyDevToHostOp storeOp, const StaticValueKnowledge& knowledge) const;
   void codeGenLmvOp(pim::PimMemCopyOp lmvOp, const StaticValueKnowledge& knowledge) const;
 
   void codeGenReceiveOp(pim::PimReceiveOp receiveOp, const StaticValueKnowledge& knowledge) const;
-  void codeGenReceiveTensorOp(pim::PimReceiveTensorOp receiveTensorOp, const StaticValueKnowledge& knowledge) const;
   void codeGenSendOp(pim::PimSendOp sendOp, const StaticValueKnowledge& knowledge) const;
-  void codeGenSendTensorOp(pim::PimSendTensorOp sendTensorOp, const StaticValueKnowledge& knowledge) const;
   void codeGenConcatOp(pim::PimConcatOp concatOp, const StaticValueKnowledge& knowledge) const;
 
   template <typename MVMTy>
@@ -172,3 +202,20 @@ public:
 OnnxMlirCompilerErrorCodes compileToPimCode(mlir::ModuleOp& moduleOpRef, std::string& outputDirName);
 
 } // namespace onnx_mlir
+
+namespace llvm {
+
+template <>
+struct DenseMapInfo<onnx_mlir::MemoryValueKey> {
+  static onnx_mlir::MemoryValueKey getEmptyKey() { return {DenseMapInfo<mlir::Value>::getEmptyKey(), 0}; }
+
+  static onnx_mlir::MemoryValueKey getTombstoneKey() { return {DenseMapInfo<mlir::Value>::getTombstoneKey(), 0}; }
+
+  static unsigned getHashValue(const onnx_mlir::MemoryValueKey& key) {
+    return hash_combine(key.value, key.lane.value_or(std::numeric_limits<unsigned>::max()));
+  }
+
+  static bool isEqual(const onnx_mlir::MemoryValueKey& lhs, const onnx_mlir::MemoryValueKey& rhs) { return lhs == rhs; }
+};
+
+} // namespace llvm

src/PIM/Compiler/PimCompilerOptions.cpp

@@ -1,7 +1,7 @@
-#include "src/Accelerators/PIM/Compiler/PimCompilerOptions.hpp"
-
 #include "llvm/Support/ErrorHandling.h"
 
+#include "src/Accelerators/PIM/Compiler/PimCompilerOptions.hpp"
+
 #define DEBUG_TYPE "PimCompilerOptions"
 
 namespace onnx_mlir {
@@ -15,11 +15,10 @@ llvm::cl::opt<PimEmissionTargetType> pimEmissionTarget(
   llvm::cl::init(EmitPimCodegen),
   llvm::cl::cat(OnnxMlirOptions));
 
-llvm::cl::opt<PimMergeSchedulerType> pimMergeScheduler(
-  "pim-merge-scheduler",
+llvm::cl::opt<PimMergeSchedulerType>
+  pimMergeScheduler("pim-merge-scheduler",
                     llvm::cl::desc("Scheduler used by the Spatial merge-compute-nodes pass"),
                     llvm::cl::values(clEnumValN(MergeSchedulerPeft, "peft", "Use PEFT scheduling")),
-  llvm::cl::values(clEnumValN(MergeSchedulerDcp, "dcp", "Use the legacy DCP-inspired scheduler")),
                     llvm::cl::init(MergeSchedulerPeft),
                     llvm::cl::cat(OnnxMlirOptions));
 
@@ -49,12 +48,6 @@ llvm::cl::opt<long> coresCount("core-count",
                                llvm::cl::desc("Number of cores in the chip. Required for PIM compilation."),
                                llvm::cl::init(-1));
 
-llvm::cl::opt<size_t> dcpCriticalWindowSize(
-  "dcp-critical-window-size",
-  llvm::cl::desc("Number of lowest-slack virtual nodes considered by each DCP coarsening iteration. "
-                 "Use 0 to run the legacy full-graph DCP analysis. Only used by the DCP scheduler."),
-  llvm::cl::init(4000));
-
 llvm::cl::opt<bool>
   ignoreConcatError("ignore-concat-error",
                     llvm::cl::desc("Ignore ConcatOp corner case: do not assert and do a simplification"),

src/PIM/Compiler/PimCompilerOptions.hpp

@@ -22,7 +22,6 @@ typedef enum {
 
 typedef enum {
   MergeSchedulerPeft = 0,
-  MergeSchedulerDcp = 1,
 } PimMergeSchedulerType;
 
 extern llvm::cl::OptionCategory OnnxMlirOptions;
@@ -36,7 +35,6 @@ extern llvm::cl::opt<bool> pimEmitJson;
 extern llvm::cl::opt<size_t> crossbarSize;
 extern llvm::cl::opt<size_t> crossbarCountInCore;
 extern llvm::cl::opt<long> coresCount;
-extern llvm::cl::opt<size_t> dcpCriticalWindowSize;
 
 bool hasExplicitPimCoreCount();
 void verifyExplicitPimCoreCount();

src/PIM/Compiler/PimCompilerUtils.cpp

@@ -30,20 +30,17 @@ void addPassesPim(OwningOpRef<ModuleOp>& module,
   if (pimEmissionTarget >= EmitSpatial) {
     pm.addPass(createONNXToSpatialPass());
     pm.addPass(createMergeComputeNodesPass());
-    // pm.addPass(createCountInstructionPass());
     pm.addPass(createMessagePass("Onnx lowered to Spatial"));
   }
 
   if (pimEmissionTarget >= EmitPim) {
     pm.addPass(createSpatialToPimPass());
-    // pm.addPass(createCountInstructionPass());
     pm.addPass(createMessagePass("Spatial lowered to Pim"));
   }
 
   if (pimEmissionTarget >= EmitPimBufferized) {
     pm.addPass(createPimBufferizationPass());
     pm.addPass(createPimStaticMemoryCoalescingPass());
-    // pm.addPass(createCountInstructionPass());
     pm.addPass(createMessagePass("Pim bufferized"));
   }
 
@@ -54,7 +51,6 @@ void addPassesPim(OwningOpRef<ModuleOp>& module,
     pm.addPass(createPimVerificationPass());
     pm.addPass(createMessagePass("Pim verified"));
     pm.addPass(createEmitPimCodePass());
-    // pm.addPass(createCountInstructionPass());
     pm.addPass(createMessagePass("Pim code emitted"));
   }
 }

src/PIM/Compiler/PimWeightEmitter.cpp

@@ -1,9 +1,7 @@
 #include "mlir/Dialect/MemRef/IR/MemRef.h"
 #include "mlir/IR/BuiltinAttributes.h"
-#include "mlir/IR/BuiltinOps.h"
 #include "mlir/IR/BuiltinTypes.h"
 
-#include "llvm/ADT/STLExtras.h"
 #include "llvm/Support/FileSystem.h"
 #include "llvm/Support/raw_ostream.h"
 
@@ -11,196 +9,42 @@
 
 #include "Conversion/ONNXToSpatial/Common/Common.hpp"
 #include "src/Accelerators/PIM/Common/IR/ShapeUtils.hpp"
-#include "src/Accelerators/PIM/Common/IR/SubviewUtils.hpp"
-#include "src/Accelerators/PIM/Common/IR/WeightUtils.hpp"
-#include "src/Accelerators/PIM/Compiler/PimBatchEmission.hpp"
-#include "src/Accelerators/PIM/Compiler/PimCodeGen.hpp"
 #include "src/Accelerators/PIM/Compiler/PimCompilerOptions.hpp"
 #include "src/Accelerators/PIM/Compiler/PimWeightEmitter.hpp"
-#include "src/Accelerators/PIM/Dialect/Pim/PimOps.hpp"
 
 using namespace llvm;
 using namespace mlir;
 
 namespace onnx_mlir {
-namespace {
+namespace {} // namespace
 
-struct DenseWeightView {
-  DenseElementsAttr denseAttr;
-  SmallVector<int64_t> shape;
-  SmallVector<int64_t> strides;
-  int64_t offset = 0;
-};
-
-FailureOr<DenseWeightView> resolveDenseWeightView(ModuleOp moduleOp, mlir::Value weight) {
-  SmallVector<Operation*> viewOps;
-  mlir::Value current = weight;
-  memref::GetGlobalOp getGlobalOp;
-
-  while (true) {
-    Operation* defOp = current.getDefiningOp();
-    if (!defOp)
-      return failure();
-    if ((getGlobalOp = dyn_cast<memref::GetGlobalOp>(defOp)))
-      break;
-    if (auto subview = dyn_cast<memref::SubViewOp>(defOp)) {
-      if (!hasAllStaticSubviewParts(subview))
-        return failure();
-      viewOps.push_back(subview);
-      current = subview.getSource();
-      continue;
-    }
-    if (auto cast = dyn_cast<memref::CastOp>(defOp)) {
-      current = cast.getSource();
-      continue;
-    }
-    if (auto collapse = dyn_cast<memref::CollapseShapeOp>(defOp)) {
-      auto srcType = dyn_cast<MemRefType>(collapse.getSrc().getType());
-      auto resultType = dyn_cast<MemRefType>(collapse.getResult().getType());
-      if (!srcType || !resultType || !srcType.hasStaticShape() || !resultType.hasStaticShape())
-        return failure();
-      viewOps.push_back(collapse);
-      current = collapse.getSrc();
-      continue;
-    }
-    if (auto expand = dyn_cast<memref::ExpandShapeOp>(defOp)) {
-      auto srcType = dyn_cast<MemRefType>(expand.getSrc().getType());
-      auto resultType = dyn_cast<MemRefType>(expand.getResult().getType());
-      if (!srcType || !resultType || !srcType.hasStaticShape() || !resultType.hasStaticShape())
-        return failure();
-      viewOps.push_back(expand);
-      current = expand.getSrc();
-      continue;
-    }
-    return failure();
-  }
-
-  auto globalOp = lookupGlobalForGetGlobal(moduleOp, getGlobalOp);
-  if (!globalOp || !globalOp.getInitialValue())
-    return failure();
-
-  auto denseAttr = dyn_cast<DenseElementsAttr>(*globalOp.getInitialValue());
-  if (!denseAttr)
-    return failure();
-
-  DenseWeightView view;
-  view.denseAttr = denseAttr;
-  view.shape.assign(denseAttr.getType().getShape().begin(), denseAttr.getType().getShape().end());
-  view.strides = computeRowMajorStrides(view.shape);
-
-  for (Operation* viewOp : llvm::reverse(viewOps)) {
-    if (auto subview = dyn_cast<memref::SubViewOp>(viewOp)) {
-      SmallVector<int64_t> nextStrides;
-      nextStrides.reserve(subview.getStaticStrides().size());
-      for (auto [offset, stride, sourceStride] :
-           llvm::zip_equal(subview.getStaticOffsets(), subview.getStaticStrides(), view.strides)) {
-        view.offset += offset * sourceStride;
-        nextStrides.push_back(stride * sourceStride);
-      }
-      view.shape.assign(subview.getStaticSizes().begin(), subview.getStaticSizes().end());
-      view.strides = std::move(nextStrides);
-      continue;
-    }
-
-    // Collapse/expand are accepted only as contiguous static reshapes of a
-    // dense global view, so a row-major stride recomputation preserves layout.
-    if (auto collapse = dyn_cast<memref::CollapseShapeOp>(viewOp)) {
-      if (view.strides != computeRowMajorStrides(view.shape))
-        return failure();
-      auto resultType = cast<MemRefType>(collapse.getResult().getType());
-      view.shape.assign(resultType.getShape().begin(), resultType.getShape().end());
-      view.strides = computeRowMajorStrides(view.shape);
-      continue;
-    }
-
-    if (auto expand = dyn_cast<memref::ExpandShapeOp>(viewOp)) {
-      if (view.strides != computeRowMajorStrides(view.shape))
-        return failure();
-      auto resultType = cast<MemRefType>(expand.getResult().getType());
-      view.shape.assign(resultType.getShape().begin(), resultType.getShape().end());
-      view.strides = computeRowMajorStrides(view.shape);
-      continue;
-    }
-  }
-
-  return view;
-}
-
-SmallVector<unsigned, 8> getUsedWeightIndices(Block& block) {
-  SmallVector<unsigned, 8> indices;
-  auto addIndex = [&](unsigned weightIndex) {
-    if (!llvm::is_contained(indices, weightIndex))
-      indices.push_back(weightIndex);
-  };
-
-  block.walk([&](pim::PimVMMOp vmmOp) { addIndex(vmmOp.getWeightIndex()); });
-  llvm::sort(indices);
-  return indices;
-}
-
-SmallVector<unsigned, 8> getUsedWeightIndices(pim::PimCoreOp coreOp) {
-  return getUsedWeightIndices(coreOp.getBody().front());
-}
-
-SmallVector<Operation*> collectTopLevelCoreLikeOps(func::FuncOp funcOp) {
-  SmallVector<Operation*> coreLikeOps;
-  for (Operation& op : funcOp.getBody().front())
-    if (dyn_cast<pim::PimCoreOp>(&op) || dyn_cast<pim::PimCoreBatchOp>(&op))
-      coreLikeOps.push_back(&op);
-  return coreLikeOps;
-}
-
-} // namespace
-
-llvm::DenseMap<size_t, llvm::DenseMap<mlir::Value, std::string>>
-createAndPopulateWeightFolder(func::FuncOp funcOp, StringRef outputDirPath) {
-  ModuleOp moduleOp = funcOp->getParentOfType<ModuleOp>();
+llvm::DenseMap<size_t, llvm::SmallVector<std::string, 8>>
+createAndPopulateWeightFolder(ArrayRef<WeightFileRequest> requests, StringRef outputDirPath) {
   auto coreWeightsDirPath = outputDirPath + "/weights";
   auto error = sys::fs::create_directory(coreWeightsDirPath);
   assert(!error && "Error creating weights directory");
   size_t indexFileName = 0;
 
   int64_t xbarSize = crossbarSize.getValue();
-  llvm::DenseMap<size_t, llvm::DenseMap<mlir::Value, std::string>> mapCoreWeightToFileName;
-  llvm::DenseMap<memref::GlobalOp, std::string> mapGlobalOpToFileName;
+  llvm::DenseMap<size_t, llvm::SmallVector<std::string, 8>> mapCoreWeightToFileName;
+  llvm::SmallVector<std::pair<ResolvedWeightView, std::string>, 16> materializedWeights;
 
-  SmallVector<Operation*> coreLikeOps = collectTopLevelCoreLikeOps(funcOp);
+  auto materializeWeight = [&](const ResolvedWeightView& weightView) -> std::string {
+    if (auto it = llvm::find_if(materializedWeights, [&](const auto& entry) { return entry.first == weightView; });
+        it != materializedWeights.end())
+      return it->second;
 
-  for (Operation* op : coreLikeOps) {
-    auto processCore = [&](pim::PimCoreOp coreOp) {
-      size_t coreId = static_cast<size_t>(coreOp.getCoreId());
-      for (unsigned index : getUsedWeightIndices(coreOp)) {
-        if (index >= coreOp.getWeights().size()) {
-          coreOp.emitWarning("Weight index " + std::to_string(index) + " is out of range");
-          assert(index < coreOp.getWeights().size() && "Weight index is out of range");
-        }
-        mlir::Value weight = coreOp.getWeights()[index];
+    auto globalOp = weightView.globalOp;
+    auto denseAttr = mlir::dyn_cast<DenseElementsAttr>(*globalOp.getInitialValue());
+    assert(denseAttr && "Weight global must have dense initial value");
 
-        auto weightView = resolveDenseWeightView(moduleOp, weight);
-        if (failed(weightView)) {
-          coreOp.emitWarning("Weight is not from a memref.get_global at index " + std::to_string(index));
-          assert(succeeded(weightView) && "Weight is not from a dense memref.global view");
-        }
-
-        if (mapCoreWeightToFileName[coreId].contains(weight))
-          continue;
-
-        auto getGlobalOp = weight.getDefiningOp<memref::GetGlobalOp>();
-        auto globalOp = getGlobalOp ? lookupGlobalForGetGlobal(moduleOp, getGlobalOp) : memref::GlobalOp {};
-        if (globalOp && mapGlobalOpToFileName.contains(globalOp)) {
-          auto& fileName = mapGlobalOpToFileName[globalOp];
-          mapCoreWeightToFileName[coreId].insert({weight, fileName});
-          continue;
-        }
-
-        DenseElementsAttr denseAttr = weightView->denseAttr;
-        ArrayRef<int64_t> shape = weightView->shape;
+    ArrayRef<int64_t> shape = weightView.shape;
     assert(isMatrixShape(shape) && "Weight matrix must be 2-dimensional");
     int64_t numRows = shape[0];
     int64_t numCols = shape[1];
     assert(numRows <= xbarSize && numCols <= xbarSize && "Weight dimensions must not exceed crossbar size");
 
-        size_t elementByteWidth = denseAttr.getElementType().getIntOrFloatBitWidth() / 8;
+    size_t elementByteWidth = getElementTypeSizeInBytes(denseAttr.getElementType());
 
     std::string newFileName = "crossbar_" + std::to_string(indexFileName++) + ".bin";
     auto weightFilePath = (coreWeightsDirPath + "/" + newFileName).str();
@@ -215,7 +59,7 @@ createAndPopulateWeightFolder(func::FuncOp funcOp, StringRef outputDirPath) {
     for (int64_t row = 0; row < xbarSize; row++) {
       for (int64_t col = 0; col < xbarSize; col++) {
         if (row < numRows && col < numCols) {
-              int64_t elementIndex = weightView->offset + row * weightView->strides[0] + col * weightView->strides[1];
+          int64_t elementIndex = weightView.offset + row * weightView.strides[0] + col * weightView.strides[1];
           APInt bits = denseAttr.getValues<APFloat>()[elementIndex].bitcastToAPInt();
           uint64_t word = bits.getZExtValue();
           weightFileStream.write(reinterpret_cast<const char*>(&word), elementByteWidth);
@@ -227,23 +71,17 @@ createAndPopulateWeightFolder(func::FuncOp funcOp, StringRef outputDirPath) {
     }
 
     weightFileStream.close();
-        if (globalOp)
-          mapGlobalOpToFileName.insert({globalOp, newFileName});
-        mapCoreWeightToFileName[coreId].insert({weight, newFileName});
-      }
-      return success();
+    materializedWeights.push_back({weightView, newFileName});
+    return newFileName;
   };
 
-    if (auto coreOp = dyn_cast<pim::PimCoreOp>(op)) {
-      (void) processCore(coreOp);
-      continue;
+  for (const WeightFileRequest& request : requests) {
+    auto& coreFiles = mapCoreWeightToFileName[request.coreId];
+    coreFiles.reserve(request.weights.size());
+    for (const ResolvedWeightView& weight : request.weights)
+      coreFiles.push_back(materializeWeight(weight));
   }
 
-    auto coreBatchOp = cast<pim::PimCoreBatchOp>(op);
-    for (unsigned lane = 0; lane < static_cast<unsigned>(coreBatchOp.getLaneCount()); ++lane)
-      if (failed(withScalarCoreFromBatchLane(coreBatchOp, lane, processCore)))
-        return mapCoreWeightToFileName;
-  }
   return mapCoreWeightToFileName;
 }
 

src/PIM/Compiler/PimWeightEmitter.hpp

@@ -1,16 +1,23 @@
 #pragma once
 
 #include "mlir/Dialect/Func/IR/FuncOps.h"
-#include "mlir/IR/Value.h"
 
 #include "llvm/ADT/DenseMap.h"
+#include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/StringRef.h"
 
 #include <string>
 
+#include "src/Accelerators/PIM/Common/IR/WeightUtils.hpp"
+
 namespace onnx_mlir {
 
-llvm::DenseMap<size_t, llvm::DenseMap<mlir::Value, std::string>>
-createAndPopulateWeightFolder(mlir::func::FuncOp funcOp, llvm::StringRef outputDirPath);
+struct WeightFileRequest {
+  size_t coreId = 0;
+  llvm::SmallVector<ResolvedWeightView, 8> weights;
+};
+
+llvm::DenseMap<size_t, llvm::SmallVector<std::string, 8>>
+createAndPopulateWeightFolder(llvm::ArrayRef<WeightFileRequest> requests, llvm::StringRef outputDirPath);
 
 } // namespace onnx_mlir

src/PIM/Conversion/ONNXToSpatial/CMakeLists.txt

@@ -3,11 +3,12 @@ mlir_tablegen(ONNXToSpatial.hpp.inc -gen-rewriters "-I${ONNX_MLIR_SRC_ROOT}")
 add_public_tablegen_target(ONNXToSpatialIncGen)
 
 add_pim_library(OMONNXToSpatial
-  ConversionPatterns.cpp
-  HostFoldability.cpp
-  HostLegality.cpp
-  PrePatterns.cpp
-  PostPatterns.cpp
+  Patterns.cpp
+  CompileTime.cpp
+  ONNXToSpatialVerifier.cpp
+  Patterns/Pre.cpp
+  Patterns/Post.cpp
+  Patterns/GeneratedConversion.cpp
   Patterns/Math/Conv.cpp
   Patterns/Math/Elementwise.cpp
   Patterns/Math/Gemm.cpp
@@ -22,8 +23,11 @@ add_pim_library(OMONNXToSpatial
   Patterns/Tensor/Resize.cpp
   Patterns/Tensor/Reshape.cpp
   Patterns/Tensor/Split.cpp
+  Patterns/Tensor/Transpose.cpp
   ONNXToSpatialPass.cpp
+  Common/AttributeUtils.cpp
   Common/ComputeRegionBuilder.cpp
+  Common/IndexingUtils.cpp
   Common/ShapeTilingUtils.cpp
   Common/WeightMaterialization.cpp
 
@@ -33,6 +37,7 @@ add_pim_library(OMONNXToSpatial
   ONNXToSpatialIncGen
 
   LINK_LIBS PUBLIC
+  MLIRLinalgDialect
   MLIRSCFDialect
   MLIRTosaDialect
   OMCompilerOptions

src/PIM/Conversion/ONNXToSpatial/Common/AttributeUtils.cpp

@@ -0,0 +1,23 @@
+#include "mlir/IR/BuiltinAttributes.h"
+
+#include "AttributeUtils.hpp"
+
+using namespace mlir;
+
+namespace onnx_mlir {
+
+int64_t getI64Attr(ArrayAttr attr, size_t index) { return cast<IntegerAttr>(attr[index]).getInt(); }
+
+int64_t getOptionalI64Attr(std::optional<ArrayAttr> attr, size_t index, int64_t defaultValue) {
+  return attr ? getI64Attr(*attr, index) : defaultValue;
+}
+
+llvm::SmallVector<int64_t> getI64ArrayAttrValues(ArrayAttr attr) {
+  llvm::SmallVector<int64_t> values;
+  values.reserve(attr.size());
+  for (Attribute value : attr)
+    values.push_back(cast<IntegerAttr>(value).getInt());
+  return values;
+}
+
+} // namespace onnx_mlir

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

@@ -0,0 +1,18 @@
+#pragma once
+
+#include "mlir/IR/BuiltinAttributes.h"
+
+#include "llvm/ADT/SmallVector.h"
+
+#include <cstddef>
+#include <optional>
+
+namespace onnx_mlir {
+
+int64_t getI64Attr(mlir::ArrayAttr attr, size_t index);
+
+int64_t getOptionalI64Attr(std::optional<mlir::ArrayAttr> attr, size_t index, int64_t defaultValue);
+
+llvm::SmallVector<int64_t> getI64ArrayAttrValues(mlir::ArrayAttr attr);
+
+} // namespace onnx_mlir

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

@@ -1,6 +1,8 @@
 #pragma once
 
+#include "AttributeUtils.hpp"
 #include "ComputeRegionBuilder.hpp"
+#include "IndexingUtils.hpp"
 #include "ShapeTilingUtils.hpp"
 #include "WeightMaterialization.hpp"
 #include "src/Accelerators/PIM/Common/PimCommon.hpp"

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

@@ -1,5 +1,6 @@
 #pragma once
 
+#include "mlir/Dialect/Tensor/IR/Tensor.h"
 #include "mlir/IR/Block.h"
 #include "mlir/IR/BuiltinTypes.h"
 #include "mlir/IR/ValueRange.h"
@@ -7,9 +8,11 @@
 
 #include <cassert>
 #include <cstddef>
+#include <limits>
 #include <type_traits>
 #include <utility>
 
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/CompileTime.hpp"
 #include "src/Accelerators/PIM/Dialect/Spatial/SpatialOps.hpp"
 
 namespace onnx_mlir {
@@ -18,13 +21,17 @@ namespace detail {
 
 inline mlir::ValueRange getBlockArgs(mlir::Block* block) { return mlir::ValueRange(block->getArguments()); }
 
+inline mlir::ValueRange getInputBlockArgs(mlir::Block* block, size_t weightCount) {
+  return mlir::ValueRange(block->getArguments()).drop_front(weightCount);
+}
+
 template <typename Fn, size_t... Is>
 decltype(auto) invokeWithBlockArgs(Fn&& fn, mlir::Block* block, std::index_sequence<Is...>) {
   return std::forward<Fn>(fn)(block->getArgument(Is)...);
 }
 
 template <typename Fn, size_t... Is>
-decltype(auto) invokeWithValues(Fn&& fn, mlir::ArrayRef<mlir::Value> values, std::index_sequence<Is...>) {
+decltype(auto) invokeWithValues(Fn&& fn, mlir::ValueRange values, std::index_sequence<Is...>) {
   return std::forward<Fn>(fn)(values[Is]...);
 }
 
@@ -45,6 +52,13 @@ using InvokeWithBlockArgsResultT = typename InvokeWithBlockArgsResult<Fn, Seq>::
 template <typename Fn>
 using InvokeWithValueRangeResultT = std::invoke_result_t<Fn, mlir::ValueRange>;
 
+struct SpatComputeBatchBodyArgs {
+  mlir::Value lane;
+  mlir::ValueRange weights;
+  mlir::ValueRange inputs;
+  mlir::ValueRange outputs;
+};
+
 } // namespace detail
 
 template <typename RewriterT>
@@ -85,6 +99,8 @@ auto createSpatCompute(RewriterT& rewriter,
   auto computeOp = spatial::SpatCompute::create(rewriter, loc, resultTypes, weights, inputs);
 
   auto* block = new mlir::Block();
+  for (mlir::Value weight : weights)
+    block->addArgument(weight.getType(), loc);
   for (mlir::Value input : inputs)
     block->addArgument(input.getType(), loc);
 
@@ -93,14 +109,17 @@ auto createSpatCompute(RewriterT& rewriter,
 
   using BodyResult = detail::InvokeWithBlockArgsResultT<std::decay_t<BodyFn>, std::make_index_sequence<NumInputs>>;
   if constexpr (std::is_same_v<BodyResult, void>) {
-    detail::invokeWithBlockArgs(std::forward<BodyFn>(body), block, std::make_index_sequence<NumInputs> {});
+    detail::invokeWithValues(std::forward<BodyFn>(body),
+                             detail::getInputBlockArgs(block, weights.size()),
+                             std::make_index_sequence<NumInputs> {});
 
     rewriter.setInsertionPointAfter(computeOp);
     return computeOp;
   }
   else {
-    auto bodyResult =
-      detail::invokeWithBlockArgs(std::forward<BodyFn>(body), block, std::make_index_sequence<NumInputs> {});
+    auto bodyResult = detail::invokeWithValues(std::forward<BodyFn>(body),
+                                               detail::getInputBlockArgs(block, weights.size()),
+                                               std::make_index_sequence<NumInputs> {});
     if (mlir::failed(bodyResult)) {
       rewriter.setInsertionPointAfter(computeOp);
       rewriter.eraseOp(computeOp);
@@ -123,6 +142,8 @@ auto createSpatCompute(RewriterT& rewriter,
   auto computeOp = spatial::SpatCompute::create(rewriter, loc, resultTypes, weights, inputs);
 
   auto* block = new mlir::Block();
+  for (mlir::Value weight : weights)
+    block->addArgument(weight.getType(), loc);
   for (mlir::Value input : inputs)
     block->addArgument(input.getType(), loc);
 
@@ -131,13 +152,13 @@ auto createSpatCompute(RewriterT& rewriter,
 
   using BodyResult = detail::InvokeWithValueRangeResultT<std::decay_t<BodyFn>>;
   if constexpr (std::is_same_v<BodyResult, void>) {
-    std::forward<BodyFn>(body)(detail::getBlockArgs(block));
+    std::forward<BodyFn>(body)(detail::getInputBlockArgs(block, weights.size()));
 
     rewriter.setInsertionPointAfter(computeOp);
     return computeOp;
   }
   else {
-    auto bodyResult = std::forward<BodyFn>(body)(detail::getBlockArgs(block));
+    auto bodyResult = std::forward<BodyFn>(body)(detail::getInputBlockArgs(block, weights.size()));
     if (mlir::failed(bodyResult)) {
       rewriter.setInsertionPointAfter(computeOp);
       rewriter.eraseOp(computeOp);
@@ -148,6 +169,95 @@ auto createSpatCompute(RewriterT& rewriter,
   }
 }
 
+template <typename RewriterT, typename BodyFn>
+auto createSpatComputeBatch(RewriterT& rewriter,
+                            mlir::Location loc,
+                            mlir::TypeRange resultTypes,
+                            int64_t laneCount,
+                            mlir::ValueRange weights,
+                            mlir::ValueRange inputs,
+                            BodyFn&& body) {
+  if (laneCount <= 0 || laneCount > std::numeric_limits<int32_t>::max())
+    return mlir::FailureOr<spatial::SpatComputeBatch>(mlir::failure());
+
+  auto batchOp = spatial::SpatComputeBatch::create(
+    rewriter, loc, resultTypes, rewriter.getI32IntegerAttr(static_cast<int32_t>(laneCount)), weights, inputs);
+
+  mlir::SmallVector<mlir::Type> blockArgTypes {rewriter.getIndexType()};
+  mlir::SmallVector<mlir::Location> blockArgLocs {loc};
+  blockArgTypes.reserve(1 + weights.size() + inputs.size() + resultTypes.size());
+  blockArgLocs.reserve(1 + weights.size() + inputs.size() + resultTypes.size());
+  for (mlir::Value weight : weights) {
+    blockArgTypes.push_back(weight.getType());
+    blockArgLocs.push_back(weight.getLoc());
+  }
+  for (mlir::Value input : inputs) {
+    blockArgTypes.push_back(input.getType());
+    blockArgLocs.push_back(input.getLoc());
+  }
+  for (mlir::Type resultType : resultTypes) {
+    blockArgTypes.push_back(resultType);
+    blockArgLocs.push_back(loc);
+  }
+
+  auto* block =
+    rewriter.createBlock(&batchOp.getBody(), batchOp.getBody().end(), mlir::TypeRange(blockArgTypes), blockArgLocs);
+  rewriter.setInsertionPointToStart(block);
+
+  detail::SpatComputeBatchBodyArgs args {
+    block->getArgument(0),
+    mlir::ValueRange(block->getArguments()).slice(1, weights.size()),
+    mlir::ValueRange(block->getArguments()).slice(1 + weights.size(), inputs.size()),
+    mlir::ValueRange(block->getArguments()).drop_front(1 + weights.size() + inputs.size())};
+
+  using BodyResult = std::invoke_result_t<BodyFn, detail::SpatComputeBatchBodyArgs>;
+  if constexpr (std::is_same_v<BodyResult, void>) {
+    std::forward<BodyFn>(body)(args);
+    rewriter.setInsertionPointAfter(batchOp);
+    return mlir::FailureOr<spatial::SpatComputeBatch>(batchOp);
+  }
+  else {
+    auto bodyResult = std::forward<BodyFn>(body)(args);
+    if (mlir::failed(bodyResult)) {
+      rewriter.setInsertionPointAfter(batchOp);
+      rewriter.eraseOp(batchOp);
+      return mlir::FailureOr<spatial::SpatComputeBatch>(mlir::failure());
+    }
+    rewriter.setInsertionPointAfter(batchOp);
+    return mlir::FailureOr<spatial::SpatComputeBatch>(batchOp);
+  }
+}
+
+inline void createParallelInsertSliceIntoBatchOutput(mlir::PatternRewriter& rewriter,
+                                                     mlir::Location loc,
+                                                     mlir::Value source,
+                                                     mlir::Value dest,
+                                                     mlir::ArrayRef<mlir::OpFoldResult> offsets,
+                                                     mlir::ArrayRef<mlir::OpFoldResult> sizes,
+                                                     mlir::ArrayRef<mlir::OpFoldResult> strides) {
+  auto inParallelOp = spatial::SpatInParallelOp::create(rewriter, loc);
+  rewriter.setInsertionPointToStart(&inParallelOp.getRegion().front());
+  mlir::tensor::ParallelInsertSliceOp::create(rewriter, loc, source, dest, offsets, sizes, strides);
+}
+
+template <typename BodyFn>
+mlir::Value materializeOrComputeUnary(mlir::Value input,
+                                      mlir::RankedTensorType resultType,
+                                      mlir::PatternRewriter& rewriter,
+                                      mlir::Location loc,
+                                      BodyFn&& build) {
+  auto&& buildFn = build;
+  if (isCompileTimeComputable(input))
+    return buildFn(input);
+
+  auto computeOp = createSpatCompute<1>(
+    rewriter, loc, mlir::TypeRange {resultType}, {}, mlir::ValueRange {input}, [&](mlir::Value computeInput) {
+      mlir::Value result = buildFn(computeInput);
+      spatial::SpatYieldOp::create(rewriter, loc, result);
+    });
+  return computeOp.getResult(0);
+}
+
 mlir::Value sumTensors(mlir::ArrayRef<mlir::Value> tensors, mlir::ConversionPatternRewriter& rewriter);
 
 } // namespace onnx_mlir

src/PIM/Conversion/ONNXToSpatial/Common/IndexingUtils.cpp

@@ -0,0 +1,45 @@
+#include <algorithm>
+
+#include "IndexingUtils.hpp"
+
+using namespace mlir;
+
+namespace onnx_mlir {
+
+int64_t normalizeAxis(int64_t axis, int64_t rank) { return axis >= 0 ? axis : rank + axis; }
+
+FailureOr<int64_t> normalizeAxisChecked(int64_t axis, int64_t rank) {
+  int64_t normalizedAxis = normalizeAxis(axis, rank);
+  if (normalizedAxis < 0 || normalizedAxis >= rank)
+    return failure();
+  return normalizedAxis;
+}
+
+int64_t normalizeIndex(int64_t index, int64_t dimSize) { return index >= 0 ? index : dimSize + index; }
+
+static SmallVector<int64_t> normalizeAxesImpl(std::optional<ArrayAttr> axesAttr, int64_t rank) {
+  SmallVector<int64_t> normalizedAxes;
+  if (!axesAttr) {
+    normalizedAxes.reserve(rank);
+    for (int64_t axis = 0; axis < rank; ++axis)
+      normalizedAxes.push_back(axis);
+  }
+  else {
+    normalizedAxes.reserve(axesAttr->size());
+    for (Attribute attr : *axesAttr)
+      normalizedAxes.push_back(normalizeAxis(cast<IntegerAttr>(attr).getInt(), rank));
+    llvm::sort(normalizedAxes);
+    normalizedAxes.erase(std::unique(normalizedAxes.begin(), normalizedAxes.end()), normalizedAxes.end());
+  }
+  return normalizedAxes;
+}
+
+FailureOr<SmallVector<int64_t>> normalizeAxesChecked(std::optional<ArrayAttr> axesAttr, int64_t rank) {
+  SmallVector<int64_t> normalizedAxes = normalizeAxesImpl(axesAttr, rank);
+  for (int64_t axis : normalizedAxes)
+    if (axis < 0 || axis >= rank)
+      return failure();
+  return normalizedAxes;
+}
+
+} // namespace onnx_mlir

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

@@ -0,0 +1,20 @@
+#pragma once
+
+#include "mlir/IR/BuiltinAttributes.h"
+#include "mlir/Support/LogicalResult.h"
+
+#include "llvm/ADT/SmallVector.h"
+
+#include <optional>
+
+namespace onnx_mlir {
+
+int64_t normalizeAxis(int64_t axis, int64_t rank);
+
+mlir::FailureOr<int64_t> normalizeAxisChecked(int64_t axis, int64_t rank);
+
+int64_t normalizeIndex(int64_t index, int64_t dimSize);
+
+mlir::FailureOr<llvm::SmallVector<int64_t>> normalizeAxesChecked(std::optional<mlir::ArrayAttr> axesAttr, int64_t rank);
+
+} // namespace onnx_mlir

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

@@ -3,26 +3,103 @@
 
 #include "llvm/ADT/SmallVector.h"
 
+#include <functional>
+
+#include "IndexingUtils.hpp"
 #include "ShapeTilingUtils.hpp"
+#include "src/Accelerators/PIM/Common/IR/ConstantUtils.hpp"
 #include "src/Accelerators/PIM/Compiler/PimCompilerOptions.hpp"
 #include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Common/Common.hpp"
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/HostFoldability.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/CompileTime.hpp"
+#include "src/Dialect/ONNX/ONNXOps.hpp"
 
 using namespace mlir;
 
 namespace onnx_mlir {
 
+bool hasStaticPositiveShape(ArrayRef<int64_t> shape) {
+  return llvm::all_of(shape, [](int64_t dim) { return dim > 0; });
+}
+
+bool hasStaticPositiveShape(RankedTensorType type) {
+  return type.hasStaticShape() && hasStaticPositiveShape(type.getShape());
+}
+
+int64_t getStaticShapeElementCount(ArrayRef<int64_t> shape) {
+  return std::accumulate(shape.begin(), shape.end(), int64_t {1}, std::multiplies<int64_t> {});
+}
+
+SmallVector<int64_t> permuteShape(ArrayRef<int64_t> shape, ArrayRef<int64_t> permutation) {
+  SmallVector<int64_t> permutedShape;
+  permutedShape.reserve(permutation.size());
+  for (int64_t axis : permutation)
+    permutedShape.push_back(shape[axis]);
+  return permutedShape;
+}
+
+SmallVector<int64_t> invertPermutation(ArrayRef<int64_t> permutation) {
+  SmallVector<int64_t> inversePermutation(permutation.size());
+  for (auto [newIndex, oldIndex] : llvm::enumerate(permutation))
+    inversePermutation[oldIndex] = static_cast<int64_t>(newIndex);
+  return inversePermutation;
+}
+
+FailureOr<SmallVector<int64_t>> getTransposePermutationChecked(std::optional<ArrayAttr> permAttr, int64_t rank) {
+  SmallVector<int64_t> permutation;
+  if (!permAttr) {
+    permutation.reserve(rank);
+    for (int64_t dim = rank - 1; dim >= 0; --dim)
+      permutation.push_back(dim);
+    return permutation;
+  }
+
+  if (static_cast<int64_t>(permAttr->size()) != rank)
+    return failure();
+
+  permutation.reserve(permAttr->size());
+  SmallVector<bool> seen(rank, false);
+  for (IntegerAttr attr : permAttr->getAsRange<IntegerAttr>()) {
+    int64_t axis = attr.getInt();
+    if (axis < 0 || axis >= rank || seen[axis])
+      return failure();
+    seen[axis] = true;
+    permutation.push_back(axis);
+  }
+  return permutation;
+}
+
+Value transposeMaybeInCompute(
+  Value value, RankedTensorType resultType, ArrayRef<int64_t> permutation, PatternRewriter& rewriter, Location loc) {
+  auto buildTranspose = [&](Value input) -> Value {
+    return ONNXTransposeOp::create(rewriter, loc, resultType, input, rewriter.getI64ArrayAttr(permutation)).getResult();
+  };
+  return materializeOrComputeUnary(value, resultType, rewriter, loc, buildTranspose);
+}
+
+SmallVector<OpFoldResult> getUnitStrides(PatternRewriter& rewriter, int64_t rank) {
+  return SmallVector<OpFoldResult>(rank, rewriter.getIndexAttr(1));
+}
+
+SmallVector<OpFoldResult> getZeroOffsets(PatternRewriter& rewriter, int64_t rank) {
+  return SmallVector<OpFoldResult>(rank, rewriter.getIndexAttr(0));
+}
+
+SmallVector<OpFoldResult> getStaticSizes(PatternRewriter& rewriter, ArrayRef<int64_t> shape) {
+  SmallVector<OpFoldResult> sizes;
+  sizes.reserve(shape.size());
+  for (int64_t dim : shape)
+    sizes.push_back(rewriter.getIndexAttr(dim));
+  return sizes;
+}
+
 SmallVector<Value> sliceTensor(
   const Value& tensorToSlice, size_t axis, int64_t sliceSize, ConversionPatternRewriter& rewriter, Location loc) {
   ArrayRef<long> shape = getTensorShape(tensorToSlice);
   assert("Invalid axis" && axis < shape.size());
 
   SmallVector<OpFoldResult> strides(shape.size(), rewriter.getIndexAttr(1));
-  SmallVector<OpFoldResult> offsets(shape.size(), rewriter.getIndexAttr(0));
-  SmallVector<OpFoldResult> sizes;
-  sizes.reserve(shape.size());
-  for (const auto size : shape)
-    sizes.push_back(rewriter.getIndexAttr(size));
+  SmallVector<OpFoldResult> offsets = getZeroOffsets(rewriter, shape.size());
+  SmallVector<OpFoldResult> sizes = getStaticSizes(rewriter, shape);
   sizes[axis] = rewriter.getIndexAttr(sliceSize);
 
   long length = shape[axis];
@@ -44,7 +121,7 @@ SmallVector<Value> sliceTensor(
       RankedTensorType::get(sliceShape, cast<RankedTensorType>(tensorToSlice.getType()).getElementType());
 
     Value slice;
-    if (isHostFoldableValue(tensorToSlice)) {
+    if (isCompileTimeComputable(tensorToSlice)) {
       slice = tensor::ExtractSliceOp::create(rewriter, loc, tensorToSlice, offsets, sizes, strides);
     }
     else {
@@ -80,47 +157,33 @@ sliceVectorPerCrossbarPerCore(const Value& vectorToSlice, ConversionPatternRewri
   return slicesPerCore;
 }
 
-DenseMap<HSliceId, DenseMap<CoreId, SmallVector<Value>>> tileMatrix(
-  Value& matrixToTile, int64_t hSliceSize, int64_t vSliceSize, ConversionPatternRewriter& rewriter, Location& loc) {
-  assert("Not a matrix" && isMatrixShape(getTensorShape(matrixToTile)));
+Value extractAxisSlice(
+  PatternRewriter& rewriter, Location loc, Value source, int64_t axis, int64_t offset, int64_t size) {
+  auto sourceType = cast<RankedTensorType>(source.getType());
+  SmallVector<int64_t> resultShape(sourceType.getShape());
+  resultShape[axis] = size;
+  auto resultType = RankedTensorType::get(resultShape, sourceType.getElementType(), sourceType.getEncoding());
 
-  DenseMap<HSliceId, DenseMap<CoreId, SmallVector<Value>>> tiles;
-
-  SmallVector<Value> hSlices = sliceTensor(matrixToTile, 1, hSliceSize, rewriter, loc);
-  size_t numHSlices = hSlices.size();
-  for (size_t hSliceId = 0; hSliceId < numHSlices; hSliceId++) {
-    Value hSlice = hSlices[hSliceId];
-    SmallVector<Value> vSlices = sliceTensor(hSlice, 0, vSliceSize, rewriter, loc);
-    for (size_t vSliceId = 0; vSliceId < vSlices.size(); vSliceId++) {
-      size_t coreId = vSliceId / crossbarCountInCore;
-      Value vSlice = vSlices[vSliceId];
-      tiles[hSliceId][coreId].push_back(vSlice);
-    }
-  }
-  return tiles;
+  SmallVector<OpFoldResult> offsets = getZeroOffsets(rewriter, sourceType.getRank());
+  SmallVector<OpFoldResult> sizes = getStaticSizes(rewriter, sourceType.getShape());
+  offsets[axis] = rewriter.getIndexAttr(offset);
+  sizes[axis] = rewriter.getIndexAttr(size);
+  return tensor::ExtractSliceOp::create(
+           rewriter, loc, resultType, source, offsets, sizes, getUnitStrides(rewriter, sourceType.getRank()))
+    .getResult();
 }
 
-Value broadcastToVector(Value scalarToBroadcast, int64_t length, ConversionPatternRewriter& rewriter, Location loc) {
-  auto oldType = cast<RankedTensorType>(scalarToBroadcast.getType());
-  Type elementType = oldType.getElementType();
-  int64_t shape[2] = {1, length};
-  Type type = oldType.cloneWith(ArrayRef(shape), elementType);
-
-  auto buildBroadcast = [&](Value input) -> Value {
-    auto zero = arith::ConstantIndexOp::create(rewriter, loc, 0).getResult();
-    SmallVector<Value> index(oldType.getRank(), zero);
-    auto elementValue = tensor::ExtractOp::create(rewriter, loc, input, index).getResult();
-    return tensor::SplatOp::create(rewriter, loc, type, elementValue);
-  };
-
-  if (isHostFoldableValue(scalarToBroadcast))
-    return buildBroadcast(scalarToBroadcast);
-
-  auto broadcastCompute =
-    createSpatCompute<1>(rewriter, loc, TypeRange {type}, {}, ValueRange {scalarToBroadcast}, [&](Value input) {
-      spatial::SpatYieldOp::create(rewriter, loc, buildBroadcast(input));
-    });
-  return broadcastCompute.getResult(0);
+Value insertStaticSlice(
+  PatternRewriter& rewriter, Location loc, Value source, Value dest, ArrayRef<OpFoldResult> offsets) {
+  auto sourceType = cast<RankedTensorType>(source.getType());
+  return tensor::InsertSliceOp::create(rewriter,
+                                       loc,
+                                       source,
+                                       dest,
+                                       offsets,
+                                       getStaticSizes(rewriter, sourceType.getShape()),
+                                       getUnitStrides(rewriter, sourceType.getRank()))
+    .getResult();
 }
 
 } // namespace onnx_mlir

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

@@ -3,6 +3,7 @@
 #include "mlir/Dialect/Tensor/IR/Tensor.h"
 #include "mlir/IR/BuiltinTypes.h"
 #include "mlir/IR/Value.h"
+#include "mlir/IR/ValueRange.h"
 #include "mlir/Transforms/DialectConversion.h"
 
 #include "llvm/ADT/ArrayRef.h"
@@ -11,46 +12,12 @@
 
 #include <cassert>
 #include <cstddef>
+#include <optional>
 #include <type_traits>
 #include <utility>
 
 namespace onnx_mlir {
 
-template <class ShapedType>
-inline auto getImageWidth(const ShapedType& shapedType) {
-  return shapedType.getDimSize(2);
-}
-
-template <class ShapedType>
-inline auto getImageHeight(const ShapedType& shapedType) {
-  return shapedType.getDimSize(3);
-}
-
-template <class ShapedType>
-inline auto getImageChannel(const ShapedType& shapedType) {
-  return shapedType.getDimSize(1);
-}
-
-template <class ShapedType>
-inline auto getImageN(const ShapedType& shapedType) {
-  return shapedType.getDimSize(0);
-}
-
-template <class ShapedType>
-inline auto getKernelWidth(const ShapedType& shapedType) {
-  return shapedType.getDimSize(2);
-}
-
-template <class ShapedType>
-inline auto getKernelHeight(const ShapedType& shapedType) {
-  return shapedType.getDimSize(3);
-}
-
-template <class ShapedType>
-inline auto getFilterCount(const ShapedType& shapedType) {
-  return shapedType.getDimSize(0);
-}
-
 using HSliceId = size_t;
 using CoreId = size_t;
 
@@ -87,17 +54,6 @@ bool isHVectorShape(mlir::ArrayRef<T> shape) {
   return shape.size() == 2 && shape[0] == 1;
 }
 
-template <class T>
-bool isVVectorShape(mlir::ArrayRef<T> shape) {
-  return shape.size() == 2 && shape[1] == 1;
-}
-
-template <class T>
-T getVectorLength(mlir::ArrayRef<T> shape) {
-  assert(isVectorShape(shape));
-  return shape[0] != 1 ? shape[0] : shape[1];
-}
-
 inline auto getTensorShape(mlir::Value tensor) {
   return mlir::cast<mlir::RankedTensorType>(tensor.getType()).getShape();
 }
@@ -109,6 +65,31 @@ inline bool haveSameStaticShape(mlir::Value lhs, mlir::Value rhs) {
       && lhsType.getShape() == rhsType.getShape();
 }
 
+bool hasStaticPositiveShape(mlir::ArrayRef<int64_t> shape);
+
+bool hasStaticPositiveShape(mlir::RankedTensorType type);
+
+int64_t getStaticShapeElementCount(mlir::ArrayRef<int64_t> shape);
+
+llvm::SmallVector<int64_t> permuteShape(mlir::ArrayRef<int64_t> shape, mlir::ArrayRef<int64_t> permutation);
+
+llvm::SmallVector<int64_t> invertPermutation(mlir::ArrayRef<int64_t> permutation);
+
+mlir::FailureOr<llvm::SmallVector<int64_t>> getTransposePermutationChecked(std::optional<mlir::ArrayAttr> permAttr,
+                                                                           int64_t rank);
+
+mlir::Value transposeMaybeInCompute(mlir::Value value,
+                                    mlir::RankedTensorType resultType,
+                                    mlir::ArrayRef<int64_t> permutation,
+                                    mlir::PatternRewriter& rewriter,
+                                    mlir::Location loc);
+
+llvm::SmallVector<mlir::OpFoldResult> getUnitStrides(mlir::PatternRewriter& rewriter, int64_t rank);
+
+llvm::SmallVector<mlir::OpFoldResult> getZeroOffsets(mlir::PatternRewriter& rewriter, int64_t rank);
+
+llvm::SmallVector<mlir::OpFoldResult> getStaticSizes(mlir::PatternRewriter& rewriter, mlir::ArrayRef<int64_t> shape);
+
 /// Slices a statically shaped tensor along one axis into contiguous pieces of
 /// at most `sliceSize` elements.
 llvm::SmallVector<mlir::Value> sliceTensor(const mlir::Value& tensorToSlice,
@@ -127,18 +108,13 @@ llvm::SmallVector<mlir::Value> sliceVector(const mlir::Value& vectorToSlice,
 llvm::DenseMap<CoreId, llvm::SmallVector<mlir::Value>> sliceVectorPerCrossbarPerCore(
   const mlir::Value& vectorToSlice, mlir::ConversionPatternRewriter& rewriter, mlir::Location loc);
 
-/// Tiles a matrix first across output columns and then across input rows so it
-/// can be assigned to crossbars grouped by core.
-llvm::DenseMap<HSliceId, llvm::DenseMap<CoreId, llvm::SmallVector<mlir::Value>>>
-tileMatrix(mlir::Value& matrixToTile,
-           int64_t hSliceSize,
-           int64_t vSliceSize,
-           mlir::ConversionPatternRewriter& rewriter,
-           mlir::Location& loc);
+mlir::Value extractAxisSlice(
+  mlir::PatternRewriter& rewriter, mlir::Location loc, mlir::Value source, int64_t axis, int64_t offset, int64_t size);
 
-mlir::Value broadcastToVector(mlir::Value scalarToBroadcast,
-                              int64_t length,
-                              mlir::ConversionPatternRewriter& rewriter,
-                              mlir::Location loc);
+mlir::Value insertStaticSlice(mlir::PatternRewriter& rewriter,
+                              mlir::Location loc,
+                              mlir::Value source,
+                              mlir::Value dest,
+                              llvm::ArrayRef<mlir::OpFoldResult> offsets);
 
 } // namespace onnx_mlir

src/PIM/Conversion/ONNXToSpatial/Common/WeightMaterialization.cpp

@@ -1,4 +1,5 @@
 #include "mlir/Dialect/Arith/IR/Arith.h"
+#include "mlir/Dialect/Linalg/IR/Linalg.h"
 #include "mlir/Dialect/Tensor/IR/Tensor.h"
 #include "mlir/IR/BuiltinTypes.h"
 #include "mlir/IR/IRMapping.h"
@@ -43,8 +44,8 @@ bool isWeightLikeComputeOperand(Value value) {
       value = collapseShapeOp.getSrc();
       continue;
     }
-    if (auto transposeOp = dyn_cast<ONNXTransposeOp>(definingOp)) {
-      value = transposeOp.getData();
+    if (auto transposeOp = dyn_cast<linalg::TransposeOp>(definingOp)) {
+      value = transposeOp.getInput();
       continue;
     }
 
@@ -80,7 +81,7 @@ FailureOr<Value> materializeWeightLikeValueInBlock(Value value, IRRewriter& rewr
     return referencedValue.getResult();
   }
 
-  if (!isa<tensor::ExtractSliceOp, tensor::ExpandShapeOp, tensor::CollapseShapeOp, ONNXTransposeOp>(definingOp))
+  if (!isa<tensor::ExtractSliceOp, tensor::ExpandShapeOp, tensor::CollapseShapeOp, linalg::TransposeOp>(definingOp))
     return failure();
 
   IRMapping localMapper;

src/PIM/Conversion/ONNXToSpatial/CompileTime.cpp

@@ -1,4 +1,5 @@
 #include "mlir/Dialect/Arith/IR/Arith.h"
+#include "mlir/Dialect/Linalg/IR/Linalg.h"
 #include "mlir/Dialect/Tensor/IR/Tensor.h"
 #include "mlir/IR/BuiltinAttributes.h"
 #include "mlir/IR/BuiltinTypes.h"
@@ -8,7 +9,11 @@
 #include "llvm/ADT/SmallPtrSet.h"
 #include "llvm/ADT/SmallVector.h"
 
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/HostFoldability.hpp"
+#include <utility>
+
+#include "src/Accelerators/PIM/Common/IR/ConstantUtils.hpp"
+#include "src/Accelerators/PIM/Common/IR/ShapeUtils.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/CompileTime.hpp"
 #include "src/Accelerators/PIM/Dialect/Spatial/SpatialOps.hpp"
 #include "src/Dialect/ONNX/ONNXOps.hpp"
 
@@ -23,8 +28,7 @@ static bool hasStaticUnitStrides(tensor::ExtractSliceOp extractSliceOp) {
 }
 
 static bool hasConstantIndices(tensor::ExtractOp extractOp) {
-  return llvm::all_of(extractOp.getIndices(),
-                      [](Value index) { return isa_and_nonnull<arith::ConstantIndexOp>(index.getDefiningOp()); });
+  return llvm::all_of(extractOp.getIndices(), [](Value index) { return matchConstantIndexValue(index).has_value(); });
 }
 
 static bool isStaticTensorResult(Operation* op) {
@@ -34,13 +38,6 @@ static bool isStaticTensorResult(Operation* op) {
   });
 }
 
-static SmallVector<int64_t> computeRowMajorStrides(ArrayRef<int64_t> shape) {
-  SmallVector<int64_t> strides(shape.size(), 1);
-  for (int64_t dim = static_cast<int64_t>(shape.size()) - 2; dim >= 0; --dim)
-    strides[dim] = strides[dim + 1] * shape[dim + 1];
-  return strides;
-}
-
 static FailureOr<DenseElementsAttr> transposeDenseElements(DenseElementsAttr denseAttr, ArrayRef<int64_t> perms) {
   auto tensorType = dyn_cast<RankedTensorType>(denseAttr.getType());
   if (!tensorType)
@@ -145,7 +142,7 @@ static DenseElementsAttr getDirectDenseConstantAttr(Value value) {
   return nullptr;
 }
 
-static DenseElementsAttr getHostFoldableDenseElementsAttrImpl(Value value, llvm::SmallPtrSetImpl<Operation*>& visited) {
+static DenseElementsAttr getHostConstantDenseElementsAttrImpl(Value value, llvm::SmallPtrSetImpl<Operation*>& visited) {
   auto* definingOp = value.getDefiningOp();
   if (!definingOp || !visited.insert(definingOp).second)
     return nullptr;
@@ -156,7 +153,7 @@ static DenseElementsAttr getHostFoldableDenseElementsAttrImpl(Value value, llvm:
     return denseAttr;
 
   if (auto transposeOp = dyn_cast<ONNXTransposeOp>(definingOp)) {
-    auto inputAttr = getHostFoldableDenseElementsAttrImpl(transposeOp.getData(), visited);
+    auto inputAttr = getHostConstantDenseElementsAttrImpl(transposeOp.getData(), visited);
     if (!inputAttr)
       return nullptr;
 
@@ -168,8 +165,18 @@ static DenseElementsAttr getHostFoldableDenseElementsAttrImpl(Value value, llvm:
     return succeeded(transposedAttr) ? *transposedAttr : nullptr;
   }
 
+  if (auto transposeOp = dyn_cast<linalg::TransposeOp>(definingOp)) {
+    auto inputAttr = getHostConstantDenseElementsAttrImpl(transposeOp.getInput(), visited);
+    if (!inputAttr)
+      return nullptr;
+
+    SmallVector<int64_t> perm(transposeOp.getPermutation().begin(), transposeOp.getPermutation().end());
+    auto transposedAttr = transposeDenseElements(inputAttr, perm);
+    return succeeded(transposedAttr) ? *transposedAttr : nullptr;
+  }
+
   if (auto collapseShapeOp = dyn_cast<tensor::CollapseShapeOp>(definingOp)) {
-    auto inputAttr = getHostFoldableDenseElementsAttrImpl(collapseShapeOp.getSrc(), visited);
+    auto inputAttr = getHostConstantDenseElementsAttrImpl(collapseShapeOp.getSrc(), visited);
     if (!inputAttr)
       return nullptr;
     auto reshapedAttr = reshapeDenseElements(inputAttr, cast<RankedTensorType>(collapseShapeOp.getType()));
@@ -177,7 +184,7 @@ static DenseElementsAttr getHostFoldableDenseElementsAttrImpl(Value value, llvm:
   }
 
   if (auto expandShapeOp = dyn_cast<tensor::ExpandShapeOp>(definingOp)) {
-    auto inputAttr = getHostFoldableDenseElementsAttrImpl(expandShapeOp.getSrc(), visited);
+    auto inputAttr = getHostConstantDenseElementsAttrImpl(expandShapeOp.getSrc(), visited);
     if (!inputAttr)
       return nullptr;
     auto reshapedAttr = reshapeDenseElements(inputAttr, cast<RankedTensorType>(expandShapeOp.getType()));
@@ -185,7 +192,7 @@ static DenseElementsAttr getHostFoldableDenseElementsAttrImpl(Value value, llvm:
   }
 
   if (auto extractSliceOp = dyn_cast<tensor::ExtractSliceOp>(definingOp)) {
-    auto inputAttr = getHostFoldableDenseElementsAttrImpl(extractSliceOp.getSource(), visited);
+    auto inputAttr = getHostConstantDenseElementsAttrImpl(extractSliceOp.getSource(), visited);
     if (!inputAttr)
       return nullptr;
     auto slicedAttr = extractSliceDenseElements(inputAttr, extractSliceOp);
@@ -195,62 +202,98 @@ static DenseElementsAttr getHostFoldableDenseElementsAttrImpl(Value value, llvm:
   return nullptr;
 }
 
-static bool isHostFoldableOpImpl(Operation* op, llvm::SmallPtrSetImpl<Operation*>& visited) {
-  if (!op || !visited.insert(op).second)
-    return false;
+static std::optional<CompileTimeSource>
+getCompileTimeSourceImpl(Operation* op, llvm::SmallPtrSetImpl<Operation*>& visited, size_t chainLength = 0) {
+  if (!op)
+    return std::nullopt;
+
+  if (!visited.insert(op).second)
+    return {
+      {op, chainLength}
+    };
 
   if (isa<arith::ConstantOp, ONNXConstantOp, ONNXNoneOp>(op))
-    return true;
+    return {
+      {op, chainLength}
+    };
+
+  chainLength += 1;
 
   if (auto extractOp = dyn_cast<tensor::ExtractOp>(op))
-    return hasConstantIndices(extractOp) && isHostFoldableValue(extractOp.getTensor());
+    return hasConstantIndices(extractOp)
+           ? getCompileTimeSourceImpl(extractOp.getTensor().getDefiningOp(), visited, chainLength)
+           : std::nullopt;
 
   if (!isStaticTensorResult(op))
-    return false;
+    return std::nullopt;
 
   if (auto transposeOp = dyn_cast<ONNXTransposeOp>(op))
-    return isHostFoldableValue(transposeOp.getData());
+    return getCompileTimeSourceImpl(transposeOp.getData().getDefiningOp(), visited, chainLength);
+
+  if (auto transposeOp = dyn_cast<linalg::TransposeOp>(op))
+    return getCompileTimeSourceImpl(transposeOp.getInput().getDefiningOp(), visited, chainLength);
 
   if (auto collapseShapeOp = dyn_cast<tensor::CollapseShapeOp>(op))
-    return isHostFoldableValue(collapseShapeOp.getSrc());
+    return getCompileTimeSourceImpl(collapseShapeOp.getSrc().getDefiningOp(), visited, chainLength);
 
   if (auto expandShapeOp = dyn_cast<tensor::ExpandShapeOp>(op))
-    return isHostFoldableValue(expandShapeOp.getSrc());
+    return getCompileTimeSourceImpl(expandShapeOp.getSrc().getDefiningOp(), visited, chainLength);
 
   if (auto extractSliceOp = dyn_cast<tensor::ExtractSliceOp>(op))
-    return hasStaticUnitStrides(extractSliceOp) && isHostFoldableValue(extractSliceOp.getSource());
+    return hasStaticUnitStrides(extractSliceOp)
+           ? getCompileTimeSourceImpl(extractSliceOp.getSource().getDefiningOp(), visited, chainLength)
+           : std::nullopt;
 
   if (auto splatOp = dyn_cast<tensor::SplatOp>(op))
-    return isHostFoldableValue(splatOp.getInput());
+    return getCompileTimeSourceImpl(splatOp.getInput().getDefiningOp(), visited, chainLength);
 
   if (auto extractRowsOp = dyn_cast<spatial::SpatExtractRowsOp>(op))
-    return isHostFoldableValue(extractRowsOp.getInput());
+    return getCompileTimeSourceImpl(extractRowsOp.getInput().getDefiningOp(), visited, chainLength);
 
-  if (auto concatOp = dyn_cast<spatial::SpatConcatOp>(op))
-    return llvm::all_of(concatOp.getInputs(), isHostFoldableValue);
+  if (auto concatOp = dyn_cast<spatial::SpatConcatOp>(op)) {
+    std::optional<CompileTimeSource> res = {};
+    for (auto operandValue : concatOp.getOperands()) {
+      auto partialRes = getCompileTimeSourceImpl(operandValue.getDefiningOp(), visited, chainLength);
+      if (!partialRes)
+        return std::nullopt;
 
-  return false;
+      if (!res) {
+        res = partialRes;
+        continue;
+      }
+      if (res->chainLength < partialRes->chainLength)
+        res = partialRes;
+    }
+    return res;
+  }
+
+  return std::nullopt;
 }
 
 } // namespace
 
-bool isHostFoldableValue(Value value) {
+std::optional<CompileTimeSource> getCompileTimeSource(Operation* op) {
+  llvm::SmallPtrSet<Operation*, 8> visited;
+  return getCompileTimeSourceImpl(op, visited);
+}
+
+bool isCompileTimeComputable(Value value) {
   auto* definingOp = value.getDefiningOp();
   if (!definingOp)
     return false;
 
   llvm::SmallPtrSet<Operation*, 8> visited;
-  return isHostFoldableOpImpl(definingOp, visited);
+  return getCompileTimeSourceImpl(definingOp, visited).has_value();
 }
 
-bool isHostFoldableOp(Operation* op) {
+bool isCompileTimeOp(Operation* op) {
   llvm::SmallPtrSet<Operation*, 8> visited;
-  return isHostFoldableOpImpl(op, visited);
+  return getCompileTimeSourceImpl(op, visited).has_value();
 }
 
-DenseElementsAttr getHostFoldableDenseElementsAttr(Value value) {
+DenseElementsAttr getHostConstDenseElementsAttr(Value value) {
   llvm::SmallPtrSet<Operation*, 8> visited;
-  return getHostFoldableDenseElementsAttrImpl(value, visited);
+  return getHostConstantDenseElementsAttrImpl(value, visited);
 }
 
 } // namespace onnx_mlir

src/PIM/Conversion/ONNXToSpatial/CompileTime.hpp

@@ -0,0 +1,22 @@
+#pragma once
+
+#include "mlir/IR/BuiltinAttributes.h"
+#include "mlir/IR/Operation.h"
+#include "mlir/IR/Value.h"
+
+namespace onnx_mlir {
+
+struct CompileTimeSource {
+  mlir::Operation* source;
+  size_t chainLength;
+};
+
+std::optional<CompileTimeSource> getCompileTimeSource(mlir::Operation* op);
+
+bool isCompileTimeComputable(mlir::Value value);
+
+bool isCompileTimeOp(mlir::Operation* op);
+
+mlir::DenseElementsAttr getHostConstDenseElementsAttr(mlir::Value value);
+
+} // namespace onnx_mlir

src/PIM/Conversion/ONNXToSpatial/HostFoldability.hpp

@@ -1,15 +0,0 @@
-#pragma once
-
-#include "mlir/IR/BuiltinAttributes.h"
-#include "mlir/IR/Operation.h"
-#include "mlir/IR/Value.h"
-
-namespace onnx_mlir {
-
-bool isHostFoldableValue(mlir::Value value);
-
-bool isHostFoldableOp(mlir::Operation* op);
-
-mlir::DenseElementsAttr getHostFoldableDenseElementsAttr(mlir::Value value);
-
-} // namespace onnx_mlir

src/PIM/Conversion/ONNXToSpatial/HostLegality.cpp

@@ -1,34 +0,0 @@
-#include "mlir/Dialect/Arith/IR/Arith.h"
-#include "mlir/Dialect/Func/IR/FuncOps.h"
-#include "mlir/Dialect/Tensor/IR/Tensor.h"
-
-#include "src/Accelerators/PIM/Common/Support/Diagnostics.hpp"
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/HostFoldability.hpp"
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/HostLegality.hpp"
-#include "src/Accelerators/PIM/Dialect/Spatial/SpatialOps.hpp"
-
-using namespace mlir;
-
-namespace onnx_mlir {
-
-LogicalResult verifyONNXToSpatialHostLegality(func::FuncOp funcOp) {
-  pim::CappedDiagnosticReporter diagnostics;
-
-  for (Operation& op : funcOp.getFunctionBody().front()) {
-    if (isa<func::ReturnOp, spatial::SpatCompute, spatial::SpatComputeBatch>(&op))
-      continue;
-    if (isHostFoldableOp(&op))
-      continue;
-
-    diagnostics.report(&op, [](Operation* illegalOp) {
-      illegalOp->emitOpError("non-foldable top-level runtime op remains after ONNX-to-Spatial; lower it inside "
-                             "spat.compute");
-    });
-  }
-
-  diagnostics.emitSuppressedSummary(funcOp, "ONNX-to-Spatial host legality failures");
-
-  return success(!diagnostics.hasFailure());
-}
-
-} // namespace onnx_mlir

src/PIM/Conversion/ONNXToSpatial/HostLegality.hpp

@@ -1,10 +0,0 @@
-#pragma once
-
-#include "mlir/Dialect/Func/IR/FuncOps.h"
-#include "mlir/Support/LogicalResult.h"
-
-namespace onnx_mlir {
-
-mlir::LogicalResult verifyONNXToSpatialHostLegality(mlir::func::FuncOp funcOp);
-
-} // namespace onnx_mlir

src/PIM/Conversion/ONNXToSpatial/ONNXToSpatialPass.cpp

@@ -1,5 +1,7 @@
+#include "mlir/Dialect/Affine/IR/AffineOps.h"
 #include "mlir/Dialect/Arith/IR/Arith.h"
 #include "mlir/Dialect/Func/IR/FuncOps.h"
+#include "mlir/Dialect/Linalg/IR/Linalg.h"
 #include "mlir/Dialect/SCF/IR/SCF.h"
 #include "mlir/Dialect/Tensor/IR/Tensor.h"
 #include "mlir/IR/IRMapping.h"
@@ -12,13 +14,10 @@
 
 #include "Common/Common.hpp"
 #include "Common/PimCommon.hpp"
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/ConversionPatterns.hpp"
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/HostFoldability.hpp"
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/HostLegality.hpp"
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/PostPatterns.hpp"
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/PrePatterns.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/CompileTime.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/ONNXToSpatialVerifier.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Patterns.hpp"
 #include "src/Accelerators/PIM/Dialect/Spatial/SpatialOps.hpp"
-#include "src/Compiler/CompilerOptions.hpp"
 #include "src/Dialect/ONNX/ONNXOps.hpp"
 
 using namespace mlir;
@@ -44,7 +43,8 @@ static void populateEmptyFunction(func::FuncOp funcOp) {
   IRRewriter rewriter(funcOp.getContext());
   IRMapping mapper;
   SmallVector<spatial::SpatCompute> computes(funcOp.getOps<spatial::SpatCompute>());
-  if (!computes.empty())
+  SmallVector<spatial::SpatComputeBatch> computeBatches(funcOp.getOps<spatial::SpatComputeBatch>());
+  if (!computes.empty() || !computeBatches.empty())
     return;
 
   auto returnOp = cast<func::ReturnOp>(funcOp.getFunctionBody().front().getTerminator());
@@ -85,30 +85,6 @@ static void populateEmptyFunction(func::FuncOp funcOp) {
     returnOp.setOperand(index, computeResult);
 }
 
-static void wrapTopLevelRuntimeTransposes(func::FuncOp funcOp) {
-  IRRewriter rewriter(funcOp.getContext());
-  Block& entryBlock = funcOp.getFunctionBody().front();
-
-  for (Operation& op : llvm::make_early_inc_range(entryBlock)) {
-    auto transposeOp = dyn_cast<ONNXTransposeOp>(&op);
-    if (!transposeOp || isHostFoldableOp(transposeOp))
-      continue;
-
-    // Transpose stays globally legal because constant/view-only cases are
-    // allowed on the host. Any residual runtime transpose must be sunk into
-    // spat.compute before the host legality check.
-    auto resultType = transposeOp.getResult().getType();
-    rewriter.setInsertionPoint(transposeOp);
-    auto computeOp = createSpatCompute<1>(
-      rewriter, transposeOp.getLoc(), TypeRange {resultType}, {}, ValueRange {transposeOp.getData()}, [&](Value input) {
-        Value transposed =
-          ONNXTransposeOp::create(rewriter, transposeOp.getLoc(), resultType, input, transposeOp.getPermAttr());
-        spatial::SpatYieldOp::create(rewriter, transposeOp.getLoc(), transposed);
-      });
-    rewriter.replaceOp(transposeOp, computeOp.getResult(0));
-  }
-}
-
 void ONNXToSpatialPass::runOnOperation() {
   ModuleOp moduleOp = getOperation();
   MLIRContext* ctx = &getContext();
@@ -116,7 +92,9 @@ void ONNXToSpatialPass::runOnOperation() {
   ConversionTarget preTarget(*ctx);
   preTarget.addLegalDialect<spatial::SpatialDialect,
                             ONNXDialect,
+                            linalg::LinalgDialect,
                             tensor::TensorDialect,
+                            affine::AffineDialect,
                             arith::ArithDialect,
                             scf::SCFDialect>();
   preTarget.addIllegalOp<ONNXConstantOp, ONNXFlattenOp>();
@@ -154,10 +132,13 @@ void ONNXToSpatialPass::runOnOperation() {
   ConversionTarget target(*ctx);
   target.addLegalDialect<spatial::SpatialDialect,
                          ONNXDialect,
+                         linalg::LinalgDialect,
                          tensor::TensorDialect,
+                         affine::AffineDialect,
                          arith::ArithDialect,
                          scf::SCFDialect>();
   target.addIllegalOp<ONNXMatMulOp>();
+  target.addIllegalOp<ONNXTransposeOp>();
   target.addIllegalOp<ONNXAddOp>();
   target.addIllegalOp<ONNXDivOp>();
   target.addIllegalOp<ONNXMulOp>();
@@ -184,22 +165,18 @@ void ONNXToSpatialPass::runOnOperation() {
     return;
   }
 
+  RewritePatternSet transposePatterns(ctx);
+  populateTransposePatterns(transposePatterns, ctx);
+  walkAndApplyPatterns(moduleOp, std::move(transposePatterns));
+
   ConversionTarget earlyPostTarget(*ctx);
   earlyPostTarget.addLegalDialect<spatial::SpatialDialect,
                                   ONNXDialect,
+                                  linalg::LinalgDialect,
                                   tensor::TensorDialect,
+                                  affine::AffineDialect,
                                   arith::ArithDialect,
                                   scf::SCFDialect>();
-  earlyPostTarget.addDynamicallyLegalOp<spatial::SpatComputeBatch>(
-    [](spatial::SpatComputeBatch batchOp) { return !requiresEarlyPostRewrite(batchOp); });
-
-  RewritePatternSet earlyPostPatterns(ctx);
-  populateEarlyPostPatterns(earlyPostPatterns, ctx);
-  if (failed(applyPartialConversion(*entryFunc, earlyPostTarget, std::move(earlyPostPatterns)))) {
-    moduleOp.emitError("failed to normalize single-lane spat.compute_batch ops before core assignment checks");
-    signalPassFailure();
-    return;
-  }
 
   PassManager cleanupPM(ctx);
   cleanupPM.addPass(createCanonicalizerPass());
@@ -211,7 +188,9 @@ void ONNXToSpatialPass::runOnOperation() {
   ConversionTarget postTarget(*ctx);
   postTarget.addLegalDialect<spatial::SpatialDialect,
                              ONNXDialect,
+                             linalg::LinalgDialect,
                              tensor::TensorDialect,
+                             affine::AffineDialect,
                              arith::ArithDialect,
                              scf::SCFDialect>();
   postTarget.addDynamicallyLegalOp<spatial::SpatCompute>(
@@ -227,9 +206,7 @@ void ONNXToSpatialPass::runOnOperation() {
     return;
   }
 
-  wrapTopLevelRuntimeTransposes(*entryFunc);
-
-  if (failed(verifyONNXToSpatialHostLegality(*entryFunc))) {
+  if (failed(verifyONNXToSpatial(*entryFunc))) {
     moduleOp.emitError("ONNX-to-Spatial host legality verification failed");
     signalPassFailure();
     return;

src/PIM/Conversion/ONNXToSpatial/ONNXToSpatialVerifier.cpp

@@ -0,0 +1,157 @@
+#include "mlir/Dialect/Func/IR/FuncOps.h"
+#include "mlir/Support/LLVM.h"
+
+#include "Common/IR/WeightUtils.hpp"
+#include "src/Accelerators/PIM/Common/Support/Diagnostics.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/CompileTime.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/ONNXToSpatialVerifier.hpp"
+#include "src/Accelerators/PIM/Dialect/Spatial/SpatialOps.hpp"
+
+using namespace mlir;
+
+namespace onnx_mlir {
+
+namespace {
+
+void checkWeightUseChains(func::FuncOp func, pim::CappedDiagnosticReporter& diagnostics) {
+  func.walk([&](Operation* op) {
+    if (!hasWeightAlways(op))
+      return;
+
+    for (Value result : op->getResults()) {
+      if (hasOnlySpatialMvmVmmWeightUses(result))
+        continue;
+
+      diagnostics.report(op, [&](Operation* illegalOp) {
+        illegalOp->emitOpError(
+          "weight-marked values may only flow through static view/slice helper chains into Spatial VMM weights");
+      });
+      return;
+    }
+  });
+}
+
+Region* getParentRegion(Value value) {
+  if (auto blockArg = dyn_cast<BlockArgument>(value))
+    return blockArg.getOwner()->getParent();
+  if (Operation* definingOp = value.getDefiningOp())
+    return definingOp->getParentRegion();
+  return nullptr;
+}
+
+bool isDefinedInsideRegion(Value value, Region& region) {
+  Region* parentRegion = getParentRegion(value);
+  return parentRegion && (&region == parentRegion || region.isAncestor(parentRegion));
+}
+
+bool isLegalHostBackedValue(Value value) {
+  Operation* definingOp = value.getDefiningOp();
+  if (!definingOp)
+    return isa<BlockArgument>(value);
+
+  if (isa<spatial::SpatChannelReceiveOp>(definingOp))
+    return false;
+
+  return definingOp->getDialect()->getNamespace() != "spat";
+}
+
+LogicalResult verifyComputeLikeInputs(Operation* computeLikeOp,
+                                      ValueRange inputs,
+                                      bool allowChannelReceiveInputs,
+                                      StringRef kind,
+                                      pim::CappedDiagnosticReporter& diagnostics) {
+  for (auto [inputIndex, input] : llvm::enumerate(inputs)) {
+    unsigned currentInputIndex = inputIndex;
+    Operation* definingOp = input.getDefiningOp();
+    if (allowChannelReceiveInputs && isa_and_nonnull<spatial::SpatChannelReceiveOp>(definingOp))
+      continue;
+    if (isLegalHostBackedValue(input))
+      continue;
+
+    diagnostics.report(computeLikeOp, [&](Operation* illegalOp) {
+      InFlightDiagnostic diagnostic = illegalOp->emitOpError()
+                                   << kind << " input #" << currentInputIndex
+                                   << (allowChannelReceiveInputs ? " must come from the host or an explicit "
+                                                                   "spat.channel_receive"
+                                                                 : " must come from the host");
+      if (definingOp)
+        diagnostic.attachNote(definingOp->getLoc()) << "illegal Spatial producer is " << definingOp->getName();
+    });
+    return failure();
+  }
+  return success();
+}
+
+void verifyNoExternalTensorCaptures(Operation* ownerOp,
+                                    Region& region,
+                                    StringRef kind,
+                                    pim::CappedDiagnosticReporter& diagnostics) {
+  region.walk([&](Operation* op) {
+    for (OpOperand& operand : op->getOpOperands()) {
+      Value value = operand.get();
+      if (!isa<TensorType>(value.getType()))
+        continue;
+      if (isDefinedInsideRegion(value, region) || isa<BlockArgument>(value))
+        continue;
+
+      Operation* definingOp = value.getDefiningOp();
+      if (definingOp && definingOp->hasTrait<OpTrait::ConstantLike>())
+        continue;
+
+      diagnostics.report(ownerOp, [&](Operation* illegalOp) {
+        InFlightDiagnostic diagnostic = illegalOp->emitOpError() << kind << " body may not capture external tensor "
+                                                                 << "values";
+        diagnostic.attachNote(op->getLoc())
+          << "tensor operand #" << operand.getOperandNumber() << " is defined outside the compute body by "
+          << (definingOp ? definingOp->getName().getStringRef() : StringRef("<block argument>"));
+      });
+    }
+  });
+}
+
+} // namespace
+
+LogicalResult verifyONNXToSpatial(func::FuncOp funcOp) {
+  pim::CappedDiagnosticReporter diagnostics;
+
+  for (Operation& op : funcOp.getOps()) {
+    if (isa<func::ReturnOp, spatial::SpatCompute, spatial::SpatComputeBatch>(&op))
+      continue;
+    if (isCompileTimeOp(&op))
+      continue;
+
+    diagnostics.report(&op, [](Operation* illegalOp) {
+      illegalOp->emitOpError(
+        "non-foldable top-level runtime op remains after ONNX-to-Spatial; lower it inside spat.compute");
+    });
+  }
+  checkWeightUseChains(funcOp, diagnostics);
+  diagnostics.emitSuppressedSummary(funcOp, "ONNX-to-Spatial verification failed");
+
+  return success(!diagnostics.hasFailure());
+}
+
+LogicalResult verifySpatialCommunicationInvariants(func::FuncOp funcOp) {
+  pim::CappedDiagnosticReporter diagnostics;
+
+  for (auto computeOp : funcOp.getOps<spatial::SpatCompute>()) {
+    (void) verifyComputeLikeInputs(
+      computeOp.getOperation(), computeOp.getInputs(), /*allowChannelReceiveInputs=*/true, "spat.compute", diagnostics);
+    verifyNoExternalTensorCaptures(computeOp.getOperation(), computeOp.getBody(), "spat.compute", diagnostics);
+  }
+
+  for (auto computeBatchOp : funcOp.getOps<spatial::SpatComputeBatch>()) {
+    (void) verifyComputeLikeInputs(computeBatchOp.getOperation(),
+                                   computeBatchOp.getInputs(),
+                                   /*allowChannelReceiveInputs=*/false,
+                                   "spat.compute_batch",
+                                   diagnostics);
+    verifyNoExternalTensorCaptures(
+      computeBatchOp.getOperation(), computeBatchOp.getBody(), "spat.compute_batch", diagnostics);
+  }
+
+  diagnostics.emitSuppressedSummary(funcOp, "Spatial communication invariant verification failed");
+  return success(!diagnostics.hasFailure());
+}
+
+} // namespace onnx_mlir

src/PIM/Conversion/ONNXToSpatial/ONNXToSpatialVerifier.hpp

@@ -0,0 +1,11 @@
+#pragma once
+
+#include "mlir/Dialect/Func/IR/FuncOps.h"
+#include "mlir/Support/LogicalResult.h"
+
+namespace onnx_mlir {
+
+mlir::LogicalResult verifyONNXToSpatial(mlir::func::FuncOp funcOp);
+mlir::LogicalResult verifySpatialCommunicationInvariants(mlir::func::FuncOp funcOp);
+
+} // namespace onnx_mlir

src/PIM/Conversion/ONNXToSpatial/Patterns.cpp

@@ -1,19 +1,14 @@
 #include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Common/Common.hpp"
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/ConversionPatterns.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Patterns.hpp"
 
 using namespace mlir;
 
 namespace onnx_mlir {
 
-namespace {
-
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/ONNXToSpatial.hpp.inc"
-
-} // namespace
-
-void populateConversionPatterns(mlir::RewritePatternSet& patterns, mlir::MLIRContext* ctx) {
-  patterns.add<removeLRN>(ctx);
+void populatePrePatterns(RewritePatternSet& patterns, MLIRContext* ctx) { populateGeneratedPrePatterns(patterns, ctx); }
 
+void populateConversionPatterns(RewritePatternSet& patterns, MLIRContext* ctx) {
+  populateGeneratedConversionPatterns(patterns, ctx);
   populateElementwisePatterns(patterns, ctx);
   populateGemmPatterns(patterns, ctx);
   populateConvPatterns(patterns, ctx);
@@ -27,6 +22,11 @@ void populateConversionPatterns(mlir::RewritePatternSet& patterns, mlir::MLIRCon
   populateResizePatterns(patterns, ctx);
   populateReshapePatterns(patterns, ctx);
   populateSplitPatterns(patterns, ctx);
+  populateTransposePatterns(patterns, ctx);
+}
+
+void populatePostPatterns(RewritePatternSet& patterns, MLIRContext* ctx) {
+  populateWeightPromotionPatterns(patterns, ctx);
 }
 
 } // namespace onnx_mlir

src/PIM/Conversion/ONNXToSpatial/Patterns.hpp

@@ -1,38 +1,39 @@
 #pragma once
 
+#include "mlir/Dialect/Func/IR/FuncOps.h"
 #include "mlir/IR/MLIRContext.h"
 #include "mlir/Transforms/DialectConversion.h"
 
+#include "src/Accelerators/PIM/Dialect/Spatial/SpatialOps.hpp"
+
 namespace onnx_mlir {
 
+void populatePrePatterns(mlir::RewritePatternSet& patterns, mlir::MLIRContext* ctx);
 void populateConversionPatterns(mlir::RewritePatternSet& patterns, mlir::MLIRContext* ctx);
+void populatePostPatterns(mlir::RewritePatternSet& patterns, mlir::MLIRContext* ctx);
+
+void populateGeneratedPrePatterns(mlir::RewritePatternSet& patterns, mlir::MLIRContext* ctx);
+void populateGeneratedConversionPatterns(mlir::RewritePatternSet& patterns, mlir::MLIRContext* ctx);
+void populateWeightPromotionPatterns(mlir::RewritePatternSet& patterns, mlir::MLIRContext* ctx);
 
 void populateConvPatterns(mlir::RewritePatternSet& patterns, mlir::MLIRContext* ctx);
-
 void populateElementwisePatterns(mlir::RewritePatternSet& patterns, mlir::MLIRContext* ctx);
-
 void populateGemmPatterns(mlir::RewritePatternSet& patterns, mlir::MLIRContext* ctx);
-
 void populateMatMulRewritePatterns(mlir::RewritePatternSet& patterns, mlir::MLIRContext* ctx);
-
 void populatePoolPatterns(mlir::RewritePatternSet& patterns, mlir::MLIRContext* ctx);
-
 void populateReduceMeanPatterns(mlir::RewritePatternSet& patterns, mlir::MLIRContext* ctx);
-
 void populateReluPatterns(mlir::RewritePatternSet& patterns, mlir::MLIRContext* ctx);
-
 void populateSigmoidPatterns(mlir::RewritePatternSet& patterns, mlir::MLIRContext* ctx);
-
 void populateSoftmaxPatterns(mlir::RewritePatternSet& patterns, mlir::MLIRContext* ctx);
-
 void populateConcatPatterns(mlir::RewritePatternSet& patterns, mlir::MLIRContext* ctx);
-
 void populateGatherPatterns(mlir::RewritePatternSet& patterns, mlir::MLIRContext* ctx);
-
 void populateResizePatterns(mlir::RewritePatternSet& patterns, mlir::MLIRContext* ctx);
-
 void populateReshapePatterns(mlir::RewritePatternSet& patterns, mlir::MLIRContext* ctx);
-
 void populateSplitPatterns(mlir::RewritePatternSet& patterns, mlir::MLIRContext* ctx);
+void populateTransposePatterns(mlir::RewritePatternSet& patterns, mlir::MLIRContext* ctx);
+
+bool requiresPostRewrite(spatial::SpatCompute computeOp);
+bool requiresPostRewrite(spatial::SpatComputeBatch computeOp);
+void annotateWeightsConstants(mlir::func::FuncOp funcOp);
 
 } // namespace onnx_mlir

src/PIM/Conversion/ONNXToSpatial/Patterns/GeneratedConversion.cpp

@@ -0,0 +1,18 @@
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Common/Common.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Patterns.hpp"
+
+using namespace mlir;
+
+namespace onnx_mlir {
+
+namespace {
+
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/ONNXToSpatial.hpp.inc"
+
+} // namespace
+
+void populateGeneratedConversionPatterns(RewritePatternSet& patterns, MLIRContext* ctx) {
+  patterns.add<removeLRN>(ctx);
+}
+
+} // namespace onnx_mlir

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

@@ -11,7 +11,7 @@
 #include "src/Accelerators/PIM/Common/Support/Diagnostics.hpp"
 #include "src/Accelerators/PIM/Compiler/PimCompilerOptions.hpp"
 #include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Common/Common.hpp"
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/HostFoldability.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/CompileTime.hpp"
 #include "src/Accelerators/PIM/Dialect/Spatial/SpatialOps.hpp"
 #include "src/Dialect/ONNX/ONNXOps.hpp"
 
@@ -28,8 +28,6 @@ struct ConvToGemm : OpConversionPattern<ONNXConvOp> {
                                 ConversionPatternRewriter& rewriter) const override;
 };
 
-static int64_t getI64FromArrayAttr(ArrayAttr arr, size_t idx) { return cast<IntegerAttr>(arr[idx]).getInt(); }
-
 static Value expandBiasIfNeeded(Value bias, ConversionPatternRewriter& rewriter, Location loc) {
   auto biasType = cast<RankedTensorType>(bias.getType());
   if (biasType.getRank() != 1)
@@ -53,23 +51,100 @@ static Value createPaddedRows(Value tensorValue,
   if (tensorType.getDimSize(0) == paddedRows)
     return tensorValue;
 
-  auto paddedType = RankedTensorType::get({paddedRows, tensorType.getDimSize(1)}, tensorType.getElementType());
+  auto paddedType = RankedTensorType::get(
+    {paddedRows, tensorType.getDimSize(1)}, tensorType.getElementType(), tensorType.getEncoding());
   SmallVector<OpFoldResult> lowPads = {rewriter.getIndexAttr(0), rewriter.getIndexAttr(0)};
   SmallVector<OpFoldResult> highPads = {rewriter.getIndexAttr(paddedRows - tensorType.getDimSize(0)),
                                         rewriter.getIndexAttr(0)};
   auto padOp = tensor::PadOp::create(rewriter, loc, paddedType, tensorValue, lowPads, highPads);
   auto* padBlock = new Block();
-  for (int i = 0; i < 2; i++)
+  for (int i = 0; i < 2; ++i)
     padBlock->addArgument(rewriter.getIndexType(), loc);
   padOp.getRegion().push_back(padBlock);
   rewriter.setInsertionPointToStart(padBlock);
-  auto zero = arith::ConstantOp::create(
-    rewriter, loc, tensorType.getElementType(), rewriter.getZeroAttr(tensorType.getElementType()));
-  tensor::YieldOp::create(rewriter, loc, zero.getResult());
+  auto zero = getOrCreateConstant(
+    rewriter, padOp.getOperation(), rewriter.getZeroAttr(tensorType.getElementType()), tensorType.getElementType());
+  tensor::YieldOp::create(rewriter, loc, zero);
   rewriter.setInsertionPointAfter(padOp);
   return padOp.getResult();
 }
 
+static Value packRowsForParallelGemm(
+  Value rows, RankedTensorType rowsType, int64_t packFactor, ConversionPatternRewriter& rewriter, Location loc) {
+  if (packFactor == 1)
+    return rows;
+
+  const int64_t packedNumRows = ceilIntegerDivide(rowsType.getDimSize(0), packFactor);
+  const int64_t paddedNumRows = packedNumRows * packFactor;
+  const int64_t rowWidth = rowsType.getDimSize(1);
+  auto groupedType =
+    RankedTensorType::get({packedNumRows, packFactor, rowWidth}, rowsType.getElementType(), rowsType.getEncoding());
+  auto packedType =
+    RankedTensorType::get({packedNumRows, packFactor * rowWidth}, rowsType.getElementType(), rowsType.getEncoding());
+
+  Value paddedRows = createPaddedRows(rows, rowsType, paddedNumRows, rewriter, loc);
+  Value groupedRows = tensor::ExpandShapeOp::create(rewriter,
+                                                    loc,
+                                                    groupedType,
+                                                    paddedRows,
+                                                    SmallVector<ReassociationIndices> {
+                                                      {0, 1},
+                                                      {2}
+  });
+  return tensor::CollapseShapeOp::create(rewriter,
+                                         loc,
+                                         packedType,
+                                         groupedRows,
+                                         SmallVector<ReassociationIndices> {
+                                           {0},
+                                           {1, 2}
+  });
+}
+
+static Value unpackRowsFromParallelGemm(Value packedRows,
+                                        RankedTensorType packedRowsType,
+                                        int64_t unpackedRows,
+                                        int64_t rowWidth,
+                                        int64_t packFactor,
+                                        ConversionPatternRewriter& rewriter,
+                                        Location loc) {
+  if (packFactor == 1)
+    return packedRows;
+
+  const int64_t packedNumRows = packedRowsType.getDimSize(0);
+  const int64_t paddedNumRows = packedNumRows * packFactor;
+  auto expandedType = RankedTensorType::get(
+    {packedNumRows, packFactor, rowWidth}, packedRowsType.getElementType(), packedRowsType.getEncoding());
+  auto paddedType =
+    RankedTensorType::get({paddedNumRows, rowWidth}, packedRowsType.getElementType(), packedRowsType.getEncoding());
+  auto unpackedType =
+    RankedTensorType::get({unpackedRows, rowWidth}, packedRowsType.getElementType(), packedRowsType.getEncoding());
+
+  Value expandedRows = tensor::ExpandShapeOp::create(rewriter,
+                                                     loc,
+                                                     expandedType,
+                                                     packedRows,
+                                                     SmallVector<ReassociationIndices> {
+                                                       {0},
+                                                       {1, 2}
+  });
+  Value paddedRows = tensor::CollapseShapeOp::create(rewriter,
+                                                     loc,
+                                                     paddedType,
+                                                     expandedRows,
+                                                     SmallVector<ReassociationIndices> {
+                                                       {0, 1},
+                                                       {2}
+  });
+  if (paddedNumRows == unpackedRows)
+    return paddedRows;
+
+  SmallVector<OpFoldResult> offsets {rewriter.getIndexAttr(0), rewriter.getIndexAttr(0)};
+  SmallVector<OpFoldResult> sizes {rewriter.getIndexAttr(unpackedRows), rewriter.getIndexAttr(rowWidth)};
+  SmallVector<OpFoldResult> strides {rewriter.getIndexAttr(1), rewriter.getIndexAttr(1)};
+  return tensor::ExtractSliceOp::create(rewriter, loc, unpackedType, paddedRows, offsets, sizes, strides);
+}
+
 static Value buildPackedWeight(DenseElementsAttr wDenseAttr,
                                Value wTrans,
                                RankedTensorType wType,
@@ -108,7 +183,30 @@ static Value buildPackedWeight(DenseElementsAttr wDenseAttr,
   }
 
   auto packedAttr = DenseElementsAttr::get(packedWeightType, packedValues);
-  return arith::ConstantOp::create(rewriter, loc, packedWeightType, packedAttr);
+  return getOrCreateConstant(rewriter, rewriter.getInsertionBlock()->getParentOp(), packedAttr, packedWeightType);
+}
+
+static Value createConvWeightMatrix(
+  Value w, RankedTensorType wFlatType, RankedTensorType wTransType, ConversionPatternRewriter& rewriter, Location loc) {
+  auto buildWeightMatrix = [&](Value weight) -> Value {
+    Value wFlat = tensor::CollapseShapeOp::create(rewriter,
+                                                  loc,
+                                                  wFlatType,
+                                                  weight,
+                                                  SmallVector<ReassociationIndices> {
+                                                    {0},
+                                                    {1, 2, 3}
+    });
+    return ONNXTransposeOp::create(rewriter, loc, wTransType, wFlat, rewriter.getI64ArrayAttr({1, 0})).getResult();
+  };
+
+  if (isCompileTimeComputable(w))
+    return buildWeightMatrix(w);
+
+  auto computeOp = createSpatCompute<1>(rewriter, loc, TypeRange {wTransType}, {}, ValueRange {w}, [&](Value weight) {
+    spatial::SpatYieldOp::create(rewriter, loc, buildWeightMatrix(weight));
+  });
+  return computeOp.getResult(0);
 }
 
 static Value buildPackedBias(bool hasBias,
@@ -134,7 +232,7 @@ static Value buildPackedBias(bool hasBias,
 
   auto packedBiasType = RankedTensorType::get({1, packFactor * numChannelsOut}, outType.getElementType());
   auto packedBiasAttr = DenseElementsAttr::get(packedBiasType, packedValues);
-  return arith::ConstantOp::create(rewriter, loc, packedBiasType, packedBiasAttr).getResult();
+  return getOrCreateConstant(rewriter, rewriter.getInsertionBlock()->getParentOp(), packedBiasAttr, packedBiasType);
 }
 
 static Value createIm2colRowComputes(Value x,
@@ -165,7 +263,6 @@ static Value createIm2colRowComputes(Value x,
                                      Location loc) {
   auto elemType = xType.getElementType();
   constexpr size_t numInputs = 1;
-  const int64_t packedNumRows = ceilIntegerDivide(numPatches, packFactor);
   auto im2colComputeOp =
     createSpatCompute<numInputs>(rewriter, loc, TypeRange {gemmInputRowsType}, {}, x, [&](Value xArg) {
       Value paddedInput = xArg;
@@ -190,8 +287,8 @@ static Value createIm2colRowComputes(Value x,
           padBlock->addArgument(rewriter.getIndexType(), loc);
         padOp.getRegion().push_back(padBlock);
         rewriter.setInsertionPointToStart(padBlock);
-        auto zero = arith::ConstantOp::create(rewriter, loc, elemType, rewriter.getFloatAttr(elemType, 0.0));
-        tensor::YieldOp::create(rewriter, loc, zero.getResult());
+        auto zero = getOrCreateConstant(rewriter, padOp.getOperation(), rewriter.getFloatAttr(elemType, 0.0), elemType);
+        tensor::YieldOp::create(rewriter, loc, zero);
         rewriter.setInsertionPointAfter(padOp);
         paddedInput = padOp.getResult();
       }
@@ -199,13 +296,14 @@ static Value createIm2colRowComputes(Value x,
       // Build im2col [numPatches, patchSize] incrementally to keep the IR small
       // until the late PIM unrolling step.
       Value im2colInit = tensor::EmptyOp::create(rewriter, loc, im2colType.getShape(), elemType);
-      auto c0 = arith::ConstantIndexOp::create(rewriter, loc, 0);
-      auto c1 = arith::ConstantIndexOp::create(rewriter, loc, 1);
-      auto cNumPatches = arith::ConstantIndexOp::create(rewriter, loc, numPatches);
-      auto cNumPatchesPerBatch = arith::ConstantIndexOp::create(rewriter, loc, numPatchesPerBatch);
-      auto cOutWidth = arith::ConstantIndexOp::create(rewriter, loc, outWidth);
-      auto cStrideHeight = arith::ConstantIndexOp::create(rewriter, loc, strideHeight);
-      auto cStrideWidth = arith::ConstantIndexOp::create(rewriter, loc, strideWidth);
+      Operation* anchorOp = rewriter.getInsertionBlock()->getParentOp();
+      auto c0 = getOrCreateIndexConstant(rewriter, anchorOp, 0);
+      auto c1 = getOrCreateIndexConstant(rewriter, anchorOp, 1);
+      auto cNumPatches = getOrCreateIndexConstant(rewriter, anchorOp, numPatches);
+      auto cNumPatchesPerBatch = getOrCreateIndexConstant(rewriter, anchorOp, numPatchesPerBatch);
+      auto cOutWidth = getOrCreateIndexConstant(rewriter, anchorOp, outWidth);
+      auto cStrideHeight = getOrCreateIndexConstant(rewriter, anchorOp, strideHeight);
+      auto cStrideWidth = getOrCreateIndexConstant(rewriter, anchorOp, strideWidth);
 
       auto im2colLoop = scf::ForOp::create(rewriter, loc, c0, cNumPatches, c1, ValueRange {im2colInit});
       rewriter.setInsertionPointToStart(im2colLoop.getBody());
@@ -252,28 +350,8 @@ static Value createIm2colRowComputes(Value x,
       Value im2col = im2colLoop.getResult(0);
 
       Value gemmInputRows = im2col;
-      if (packFactor != 1) {
-        const int64_t paddedNumPatches = packedNumRows * packFactor;
-        auto groupedType = RankedTensorType::get({packedNumRows, packFactor, patchSize}, elemType);
-        auto packedType = RankedTensorType::get({packedNumRows, packFactor * patchSize}, elemType);
-        Value paddedIm2col = createPaddedRows(im2col, im2colType, paddedNumPatches, rewriter, loc);
-        Value groupedIm2col = tensor::ExpandShapeOp::create(rewriter,
-                                                            loc,
-                                                            groupedType,
-                                                            paddedIm2col,
-                                                            SmallVector<ReassociationIndices> {
-                                                              {0, 1},
-                                                              {2}
-        });
-        gemmInputRows = tensor::CollapseShapeOp::create(rewriter,
-                                                        loc,
-                                                        packedType,
-                                                        groupedIm2col,
-                                                        SmallVector<ReassociationIndices> {
-                                                          {0},
-                                                          {1, 2}
-        });
-      }
+      if (packFactor != 1)
+        gemmInputRows = packRowsForParallelGemm(im2col, im2colType, packFactor, rewriter, loc);
 
       spatial::SpatYieldOp::create(rewriter, loc, gemmInputRows);
     });
@@ -291,41 +369,20 @@ static Value createCollectedConvOutput(ValueRange gemmRows,
                                        int64_t packFactor,
                                        ConversionPatternRewriter& rewriter,
                                        Location loc) {
-  const int64_t packedNumRows = ceilIntegerDivide(numPatches, packFactor);
-  const int64_t paddedNumPatches = packedNumRows * packFactor;
   auto collectComputeOp = createSpatCompute(rewriter, loc, convType, {}, gemmRows, [&](ValueRange gemmRowArgs) {
     Value gemmOut;
     if (packFactor == 1) {
       gemmOut = createSpatConcat(rewriter, loc, /*axis=*/0, gemmRowArgs);
     }
     else {
-      auto expandedType = RankedTensorType::get({packedNumRows, packFactor, numChannelsOut}, outType.getElementType());
-      auto paddedType = RankedTensorType::get({paddedNumPatches, numChannelsOut}, outType.getElementType());
       Value packedOutput = createSpatConcat(rewriter, loc, /*axis=*/0, gemmRowArgs);
-      Value expandedOutput = tensor::ExpandShapeOp::create(rewriter,
-                                                           loc,
-                                                           expandedType,
-                                                           packedOutput,
-                                                           SmallVector<ReassociationIndices> {
-                                                             {0},
-                                                             {1, 2}
-      });
-      Value paddedOutput = tensor::CollapseShapeOp::create(rewriter,
-                                                           loc,
-                                                           paddedType,
-                                                           expandedOutput,
-                                                           SmallVector<ReassociationIndices> {
-                                                             {0, 1},
-                                                             {2}
-      });
-
-      gemmOut = paddedOutput;
-      if (paddedNumPatches != numPatches) {
-        SmallVector<OpFoldResult> offsets = {rewriter.getIndexAttr(0), rewriter.getIndexAttr(0)};
-        SmallVector<OpFoldResult> sizes = {rewriter.getIndexAttr(numPatches), rewriter.getIndexAttr(numChannelsOut)};
-        SmallVector<OpFoldResult> strides = {rewriter.getIndexAttr(1), rewriter.getIndexAttr(1)};
-        gemmOut = tensor::ExtractSliceOp::create(rewriter, loc, gemmOutType, paddedOutput, offsets, sizes, strides);
-      }
+      gemmOut = unpackRowsFromParallelGemm(packedOutput,
+                                           cast<RankedTensorType>(packedOutput.getType()),
+                                           numPatches,
+                                           numChannelsOut,
+                                           packFactor,
+                                           rewriter,
+                                           loc);
     }
 
     // Restore to NCHW layout:
@@ -391,19 +448,11 @@ static Value lowerSingleConvGroup(Value x,
   const int64_t xbarSize = static_cast<int64_t>(crossbarSize.getValue());
   const int64_t wMaxDim = std::max(patchSize, numChannelsOut);
   const int64_t maxParallelPixels = std::max<int64_t>(1, xbarSize / wMaxDim);
-  auto wDenseAttr = getHostFoldableDenseElementsAttr(w);
+  auto wDenseAttr = getHostConstDenseElementsAttr(w);
 
   // Prepare weight matrix W for crossbar storage:
   // W: [numChannelsOut, numChannelsIn, wHeight, wWidth] -> [numChannelsOut, patchSize] -> [patchSize, numChannelsOut]
-  Value wFlat = tensor::CollapseShapeOp::create(rewriter,
-                                                loc,
-                                                wFlatType,
-                                                w,
-                                                SmallVector<ReassociationIndices> {
-                                                  {0},
-                                                  {1, 2, 3}
-  });
-  Value wTrans = ONNXTransposeOp::create(rewriter, loc, wTransType, wFlat, rewriter.getI64ArrayAttr({1, 0}));
+  Value wTrans = createConvWeightMatrix(w, wFlatType, wTransType, rewriter, loc);
 
   // Pass bias through directly; Gemm handles rank-1 C canonicalization.
   bool hasB = !isa<ONNXNoneOp>(b.getDefiningOp());
@@ -412,7 +461,7 @@ static Value lowerSingleConvGroup(Value x,
   DenseElementsAttr biasDenseAttr;
   if (hasB) {
     gemmBias = b;
-    biasDenseAttr = getHostFoldableDenseElementsAttr(b);
+    biasDenseAttr = getHostConstDenseElementsAttr(b);
     biasMatrix = expandBiasIfNeeded(b, rewriter, loc);
   }
   const bool canPackWeightsAsConstants = static_cast<bool>(wDenseAttr);
@@ -597,10 +646,10 @@ LogicalResult ConvToGemm::matchAndRewrite(ONNXConvOp convOp,
     return failure();
   }
 
-  const int64_t strideHeight = stridesAttr ? getI64FromArrayAttr(*stridesAttr, 0) : 1;
-  const int64_t strideWidth = stridesAttr ? getI64FromArrayAttr(*stridesAttr, 1) : 1;
-  const int64_t dilationHeight = dilationsAttr ? getI64FromArrayAttr(*dilationsAttr, 0) : 1;
-  const int64_t dilationWidth = dilationsAttr ? getI64FromArrayAttr(*dilationsAttr, 1) : 1;
+  const int64_t strideHeight = getOptionalI64Attr(stridesAttr, 0, 1);
+  const int64_t strideWidth = getOptionalI64Attr(stridesAttr, 1, 1);
+  const int64_t dilationHeight = getOptionalI64Attr(dilationsAttr, 0, 1);
+  const int64_t dilationWidth = getOptionalI64Attr(dilationsAttr, 1, 1);
 
   int64_t padHeightBegin = 0;
   int64_t padHeightEnd = 0;
@@ -608,10 +657,10 @@ LogicalResult ConvToGemm::matchAndRewrite(ONNXConvOp convOp,
   int64_t padWidthEnd = 0;
 
   if (padsAttr) {
-    padHeightBegin = getI64FromArrayAttr(*padsAttr, 0);
-    padWidthBegin = getI64FromArrayAttr(*padsAttr, 1);
-    padHeightEnd = getI64FromArrayAttr(*padsAttr, 2);
-    padWidthEnd = getI64FromArrayAttr(*padsAttr, 3);
+    padHeightBegin = getI64Attr(*padsAttr, 0);
+    padWidthBegin = getI64Attr(*padsAttr, 1);
+    padHeightEnd = getI64Attr(*padsAttr, 2);
+    padWidthEnd = getI64Attr(*padsAttr, 3);
   }
   else {
     // Compute padding from auto_pad attribute
@@ -717,7 +766,7 @@ LogicalResult ConvToGemm::matchAndRewrite(ONNXConvOp convOp,
   }
 
   Value result;
-  if (llvm::all_of(groupResults, isHostFoldableValue)) {
+  if (llvm::all_of(groupResults, isCompileTimeComputable)) {
     result = createSpatConcat(rewriter, loc, /*axis=*/1, groupResults);
   }
   else {

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

@@ -7,7 +7,7 @@
 
 #include "src/Accelerators/PIM/Common/IR/ShapeUtils.hpp"
 #include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Common/Common.hpp"
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/ConversionPatterns.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Patterns.hpp"
 #include "src/Accelerators/PIM/Dialect/Spatial/SpatialOps.hpp"
 #include "src/Dialect/ONNX/ONNXOps.hpp"
 
@@ -83,7 +83,7 @@ static FailureOr<Value> materializeBroadcastedConstantTensor(Value value,
   }
 
   auto broadcastedAttr = DenseElementsAttr::get(resultType, resultValues);
-  return arith::ConstantOp::create(rewriter, loc, resultType, broadcastedAttr).getResult();
+  return getOrCreateConstant(rewriter, rewriter.getInsertionBlock()->getParentOp(), broadcastedAttr, resultType);
 }
 
 static FailureOr<Value>
@@ -121,7 +121,7 @@ static FailureOr<Value> materializeReciprocalTensor(Value value,
   }
 
   auto reciprocalAttr = DenseFPElementsAttr::get(resultType, reciprocalValues);
-  return arith::ConstantOp::create(rewriter, loc, resultType, reciprocalAttr).getResult();
+  return getOrCreateConstant(rewriter, rewriter.getInsertionBlock()->getParentOp(), reciprocalAttr, resultType);
 }
 
 template <typename OnnxOp, typename SpatialOp>

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

@@ -1,3 +1,5 @@
+#include "mlir/Dialect/Arith/IR/Arith.h"
+#include "mlir/Dialect/SCF/IR/SCF.h"
 #include "mlir/Dialect/Tensor/IR/Tensor.h"
 #include "mlir/IR/BuiltinTypes.h"
 #include "mlir/IR/PatternMatch.h"
@@ -5,12 +7,10 @@
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/ADT/SmallVector.h"
 
-#include <functional>
-#include <numeric>
-
+#include "src/Accelerators/PIM/Common/IR/AffineUtils.hpp"
 #include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Common/Common.hpp"
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/ConversionPatterns.hpp"
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/HostFoldability.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/CompileTime.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Patterns.hpp"
 #include "src/Accelerators/PIM/Dialect/Spatial/SpatialOps.hpp"
 #include "src/Dialect/ONNX/ONNXOps.hpp"
 
@@ -19,14 +19,6 @@ using namespace mlir;
 namespace onnx_mlir {
 namespace {
 
-static bool haveStaticPositiveShape(ArrayRef<int64_t> shape) {
-  return llvm::all_of(shape, [](int64_t dim) { return dim > 0; });
-}
-
-static int64_t getStaticShapeElementCount(ArrayRef<int64_t> shape) {
-  return std::accumulate(shape.begin(), shape.end(), int64_t {1}, std::multiplies<int64_t> {});
-}
-
 static FailureOr<SmallVector<int64_t>> inferSupportedBatchShape(ArrayRef<int64_t> lhsBatchShape,
                                                                 ArrayRef<int64_t> rhsBatchShape) {
   if (lhsBatchShape.empty())
@@ -38,62 +30,60 @@ static FailureOr<SmallVector<int64_t>> inferSupportedBatchShape(ArrayRef<int64_t
   return SmallVector<int64_t>(lhsBatchShape.begin(), lhsBatchShape.end());
 }
 
-static Value collapseBatchDims(Value value,
-                               int64_t batchSize,
-                               int64_t rows,
-                               int64_t cols,
-                               PatternRewriter& rewriter,
-                               Location loc) {
+static Value
+collapseBatchDims(Value value, int64_t batchSize, int64_t rows, int64_t cols, PatternRewriter& rewriter, Location loc) {
   auto type = cast<RankedTensorType>(value.getType());
   if (type.getRank() == 2 || type.getRank() == 3)
     return value;
 
-  auto collapsedType =
-    RankedTensorType::get({batchSize, rows, cols}, type.getElementType(), type.getEncoding());
-  SmallVector<ReassociationIndices> reassociation = {
-    ReassociationIndices {},
+  auto collapsedType = RankedTensorType::get({batchSize, rows, cols}, type.getElementType(), type.getEncoding());
+  SmallVector<ReassociationIndices> reassociation = {ReassociationIndices {},
                                                      ReassociationIndices {static_cast<int64_t>(type.getRank() - 2)},
-    ReassociationIndices {static_cast<int64_t>(type.getRank() - 1)}
-  };
+                                                     ReassociationIndices {static_cast<int64_t>(type.getRank() - 1)}};
   for (int64_t dim = 0; dim < type.getRank() - 2; ++dim)
     reassociation.front().push_back(dim);
 
   auto buildCollapsed = [&](Value input) -> Value {
     return tensor::CollapseShapeOp::create(rewriter, loc, collapsedType, input, reassociation);
   };
-
-  if (isHostFoldableValue(value))
-    return buildCollapsed(value);
-
-  auto collapseCompute =
-    createSpatCompute<1>(rewriter, loc, TypeRange {collapsedType}, {}, ValueRange {value}, [&](Value input) {
-      spatial::SpatYieldOp::create(rewriter, loc, buildCollapsed(input));
-    });
-  return collapseCompute.getResult(0);
+  return materializeOrComputeUnary(value, collapsedType, rewriter, loc, buildCollapsed);
 }
 
-static Value expandBatchDims(Value value,
-                             RankedTensorType outputType,
-                             size_t batchRank,
-                             PatternRewriter& rewriter,
-                             Location loc) {
+static Value
+expandBatchDims(Value value, RankedTensorType outputType, size_t batchRank, PatternRewriter& rewriter, Location loc) {
   if (cast<RankedTensorType>(value.getType()) == outputType)
     return value;
 
-  SmallVector<ReassociationIndices> reassociation = {
-    ReassociationIndices {},
+  SmallVector<ReassociationIndices> reassociation = {ReassociationIndices {},
                                                      ReassociationIndices {static_cast<int64_t>(batchRank)},
-    ReassociationIndices {static_cast<int64_t>(batchRank + 1)}
-  };
+                                                     ReassociationIndices {static_cast<int64_t>(batchRank + 1)}};
   for (size_t dim = 0; dim < batchRank; ++dim)
     reassociation.front().push_back(static_cast<int64_t>(dim));
 
-  auto expandCompute =
-    createSpatCompute<1>(rewriter, loc, TypeRange {outputType}, {}, ValueRange {value}, [&](Value input) {
-      Value expanded = tensor::ExpandShapeOp::create(rewriter, loc, outputType, input, reassociation);
-      spatial::SpatYieldOp::create(rewriter, loc, expanded);
+  auto buildExpanded = [&](Value input) -> Value {
+    return tensor::ExpandShapeOp::create(rewriter, loc, outputType, input, reassociation).getResult();
+  };
+  return materializeOrComputeUnary(value, outputType, rewriter, loc, buildExpanded);
+}
+
+static Value ensureBatchedTensor(
+  Value value, int64_t batchSize, int64_t rows, int64_t cols, PatternRewriter& rewriter, Location loc) {
+  auto type = cast<RankedTensorType>(value.getType());
+  if (type.getRank() == 3)
+    return value;
+
+  auto batchedType = RankedTensorType::get({batchSize, rows, cols}, type.getElementType(), type.getEncoding());
+  auto buildExpanded = [&](Value input) -> Value {
+    return tensor::ExpandShapeOp::create(rewriter,
+                                         loc,
+                                         batchedType,
+                                         input,
+                                         SmallVector<ReassociationIndices> {
+                                           {0, 1},
+                                           {2}
     });
-  return expandCompute.getResult(0);
+  };
+  return materializeOrComputeUnary(value, batchedType, rewriter, loc, buildExpanded);
 }
 
 static Value extractBatchMatrix(Value value,
@@ -112,7 +102,7 @@ static Value extractBatchMatrix(Value value,
     rewriter.getIndexAttr(batchSize == 1 ? 0 : batchIndex), rewriter.getIndexAttr(0), rewriter.getIndexAttr(0)};
   SmallVector<OpFoldResult> sizes = {
     rewriter.getIndexAttr(1), rewriter.getIndexAttr(rows), rewriter.getIndexAttr(cols)};
-  SmallVector<OpFoldResult> strides = {rewriter.getIndexAttr(1), rewriter.getIndexAttr(1), rewriter.getIndexAttr(1)};
+  SmallVector<OpFoldResult> strides = getUnitStrides(rewriter, 3);
   auto matrixType = RankedTensorType::get({rows, cols}, type.getElementType());
   auto buildMatrix = [&](Value input) -> Value {
     Value slice = tensor::ExtractSliceOp::create(rewriter, loc, sliceType, input, offsets, sizes, strides);
@@ -126,14 +116,7 @@ static Value extractBatchMatrix(Value value,
     });
   };
 
-  if (isHostFoldableValue(value))
-    return buildMatrix(value);
-
-  auto batchMatrixCompute =
-    createSpatCompute<1>(rewriter, loc, TypeRange {matrixType}, {}, ValueRange {value}, [&](Value input) {
-      spatial::SpatYieldOp::create(rewriter, loc, buildMatrix(input));
-    });
-  return batchMatrixCompute.getResult(0);
+  return materializeOrComputeUnary(value, matrixType, rewriter, loc, buildMatrix);
 }
 
 static Value transposeLastTwoDims(Value value, PatternRewriter& rewriter, Location loc) {
@@ -150,18 +133,7 @@ static Value transposeLastTwoDims(Value value, PatternRewriter& rewriter, Locati
     perm = {0, 2, 1};
   }
 
-  auto buildTranspose = [&](Value input) -> Value {
-    return ONNXTransposeOp::create(rewriter, loc, transposedType, input, rewriter.getI64ArrayAttr(perm));
-  };
-
-  if (isHostFoldableValue(value))
-    return buildTranspose(value);
-
-  auto transposeCompute =
-    createSpatCompute<1>(rewriter, loc, TypeRange {transposedType}, {}, ValueRange {value}, [&](Value input) {
-      spatial::SpatYieldOp::create(rewriter, loc, buildTranspose(input));
-    });
-  return transposeCompute.getResult(0);
+  return transposeMaybeInCompute(value, transposedType, perm, rewriter, loc);
 }
 
 static Value transposeLastTwoDimsInCompute(Value value, PatternRewriter& rewriter, Location loc) {
@@ -185,28 +157,527 @@ static Value transposeLastTwoDimsInCompute(Value value, PatternRewriter& rewrite
   return transposeCompute.getResult(0);
 }
 
-static Value concatValues(ValueRange inputs, int64_t axis, PatternRewriter& rewriter, Location loc) {
-  auto firstType = cast<RankedTensorType>(inputs.front().getType());
-  SmallVector<int64_t> outputShape(firstType.getShape().begin(), firstType.getShape().end());
-  int64_t concatDimSize = 0;
-  for (Value input : inputs)
-    concatDimSize += cast<RankedTensorType>(input.getType()).getDimSize(axis);
-  outputShape[axis] = concatDimSize;
-  auto resultType = RankedTensorType::get(outputShape, firstType.getElementType(), firstType.getEncoding());
+static Value createZeroPaddedTensor(Value value, RankedTensorType resultType, PatternRewriter& rewriter, Location loc) {
+  auto sourceType = cast<RankedTensorType>(value.getType());
+  SmallVector<OpFoldResult> lowPads(sourceType.getRank(), rewriter.getIndexAttr(0));
+  SmallVector<OpFoldResult> highPads;
+  highPads.reserve(sourceType.getRank());
+  for (auto [sourceDim, resultDim] : llvm::zip(sourceType.getShape(), resultType.getShape()))
+    highPads.push_back(rewriter.getIndexAttr(resultDim - sourceDim));
 
-  if (llvm::all_of(inputs, isHostFoldableValue))
-    return createSpatConcat(rewriter, loc, axis, inputs);
-
-  auto concatCompute = createSpatCompute(rewriter, loc, TypeRange {resultType}, {}, inputs, [&](ValueRange args) {
-    spatial::SpatYieldOp::create(rewriter, loc, createSpatConcat(rewriter, loc, axis, args));
-  });
-  return concatCompute.getResult(0);
+  auto padOp = tensor::PadOp::create(rewriter, loc, resultType, value, lowPads, highPads);
+  auto* padBlock = new Block();
+  for (int64_t i = 0; i < sourceType.getRank(); ++i)
+    padBlock->addArgument(rewriter.getIndexType(), loc);
+  padOp.getRegion().push_back(padBlock);
+  rewriter.setInsertionPointToStart(padBlock);
+  auto zero = getOrCreateConstant(
+    rewriter, padOp.getOperation(), rewriter.getZeroAttr(sourceType.getElementType()), sourceType.getElementType());
+  tensor::YieldOp::create(rewriter, loc, zero);
+  rewriter.setInsertionPointAfter(padOp);
+  return padOp.getResult();
 }
 
-struct MatMulToGemm : OpRewritePattern<ONNXMatMulOp> {
-  using OpRewritePattern::OpRewritePattern;
+static Value createPaddedBatchedInputCompute(Value input,
+                                             RankedTensorType paddedInputType,
+                                             PatternRewriter& rewriter,
+                                             Location loc) {
+  auto inputType = cast<RankedTensorType>(input.getType());
+  if (inputType == paddedInputType)
+    return input;
 
-  LogicalResult matchAndRewrite(ONNXMatMulOp matmulOp, PatternRewriter& rewriter) const override {
+  auto computeOp = createSpatCompute<1>(rewriter, loc, TypeRange {paddedInputType}, {}, input, [&](Value computeInput) {
+    Value paddedInput = createZeroPaddedTensor(computeInput, paddedInputType, rewriter, loc);
+    spatial::SpatYieldOp::create(rewriter, loc, paddedInput);
+  });
+  return computeOp.getResult(0);
+}
+
+static FailureOr<Value> materializePaddedBatchedWeight(
+  Value value, int64_t sourceBatch, int64_t targetBatch, RankedTensorType resultType, PatternRewriter& rewriter) {
+  auto sourceType = cast<RankedTensorType>(value.getType());
+  if (sourceType == resultType)
+    return value;
+
+  auto denseAttr = getHostConstDenseElementsAttr(value);
+  if (!denseAttr)
+    return failure();
+
+  const int64_t sourceRows = sourceType.getRank() == 2 ? sourceType.getDimSize(0) : sourceType.getDimSize(1);
+  const int64_t sourceCols = sourceType.getRank() == 2 ? sourceType.getDimSize(1) : sourceType.getDimSize(2);
+  const int64_t targetRows = resultType.getDimSize(1);
+  const int64_t targetCols = resultType.getDimSize(2);
+  SmallVector<Attribute> sourceValues(denseAttr.getValues<Attribute>());
+  SmallVector<Attribute> resultValues(resultType.getNumElements(), rewriter.getZeroAttr(resultType.getElementType()));
+
+  for (int64_t batchIdx = 0; batchIdx < targetBatch; ++batchIdx) {
+    const int64_t sourceBatchIdx = sourceType.getRank() == 2 ? 0 : (sourceBatch == 1 ? 0 : batchIdx);
+    const int64_t sourceBatchBase = sourceType.getRank() == 2 ? 0 : sourceBatchIdx * sourceRows * sourceCols;
+    const int64_t targetBatchBase = batchIdx * targetRows * targetCols;
+    for (int64_t row = 0; row < sourceRows; ++row)
+      for (int64_t col = 0; col < sourceCols; ++col)
+        resultValues[targetBatchBase + row * targetCols + col] = sourceValues[sourceBatchBase + row * sourceCols + col];
+  }
+
+  auto resultAttr = DenseElementsAttr::get(resultType, resultValues);
+  return getOrCreateConstant(rewriter, rewriter.getInsertionBlock()->getParentOp(), resultAttr, resultType);
+}
+
+static Value extractBatchedATile(Value a,
+                                 int64_t sourceBatchCount,
+                                 Value batch,
+                                 Value row,
+                                 Value kOffset,
+                                 RankedTensorType aTileType,
+                                 PatternRewriter& rewriter,
+                                 Location loc) {
+  auto aSliceType = RankedTensorType::get({1, 1, aTileType.getDimSize(1)}, aTileType.getElementType());
+  SmallVector<OpFoldResult> offsets {
+    sourceBatchCount == 1 ? OpFoldResult(rewriter.getIndexAttr(0)) : OpFoldResult(batch), row, kOffset};
+  SmallVector<OpFoldResult> sizes {
+    rewriter.getIndexAttr(1), rewriter.getIndexAttr(1), rewriter.getIndexAttr(aTileType.getDimSize(1))};
+  auto slice =
+    tensor::ExtractSliceOp::create(rewriter, loc, aSliceType, a, offsets, sizes, getUnitStrides(rewriter, 3));
+  return tensor::CollapseShapeOp::create(rewriter,
+                                         loc,
+                                         aTileType,
+                                         slice,
+                                         SmallVector<ReassociationIndices> {
+                                           {0, 1},
+                                           {2}
+  });
+}
+
+static Value extractBatchedBTile(Value b,
+                                 int64_t sourceBatchCount,
+                                 Value batch,
+                                 Value kOffset,
+                                 Value hOffset,
+                                 RankedTensorType bTileType,
+                                 PatternRewriter& rewriter,
+                                 Location loc) {
+  auto bSliceType =
+    RankedTensorType::get({1, bTileType.getDimSize(0), bTileType.getDimSize(1)}, bTileType.getElementType());
+  SmallVector<OpFoldResult> offsets {
+    sourceBatchCount == 1 ? OpFoldResult(rewriter.getIndexAttr(0)) : OpFoldResult(batch), kOffset, hOffset};
+  SmallVector<OpFoldResult> sizes {rewriter.getIndexAttr(1),
+                                   rewriter.getIndexAttr(bTileType.getDimSize(0)),
+                                   rewriter.getIndexAttr(bTileType.getDimSize(1))};
+  auto slice =
+    tensor::ExtractSliceOp::create(rewriter, loc, bSliceType, b, offsets, sizes, getUnitStrides(rewriter, 3));
+  return tensor::CollapseShapeOp::create(rewriter,
+                                         loc,
+                                         bTileType,
+                                         slice,
+                                         SmallVector<ReassociationIndices> {
+                                           {0, 1},
+                                           {2}
+  });
+}
+
+static Value getBatchLaneIndex(
+  Value lane, int64_t numOutRows, int64_t numKSlices, int64_t numOutHSlices, PatternRewriter& rewriter, Location loc) {
+  return affineFloorDivConst(
+    rewriter, loc, lane, numOutRows * numKSlices * numOutHSlices, rewriter.getInsertionBlock()->getParentOp());
+}
+
+static spatial::SpatComputeBatch createBatchedVmmBatch(Value a,
+                                                       Value b,
+                                                       RankedTensorType aType,
+                                                       int64_t aBatchCount,
+                                                       RankedTensorType bType,
+                                                       int64_t bBatchCount,
+                                                       RankedTensorType partialPiecesType,
+                                                       int64_t numOutRows,
+                                                       int64_t numKSlices,
+                                                       int64_t numOutHSlices,
+                                                       PatternRewriter& rewriter,
+                                                       Location loc) {
+  const int64_t laneCount = partialPiecesType.getDimSize(0);
+  auto batchOp = createSpatComputeBatch(
+    rewriter,
+    loc,
+    TypeRange {partialPiecesType},
+    laneCount,
+    ValueRange {b},
+    ValueRange {a},
+    [&](detail::SpatComputeBatchBodyArgs args) {
+      Operation* anchorOp = rewriter.getInsertionBlock()->getParentOp();
+      Value row = affineModConst(rewriter, loc, args.lane, numOutRows, anchorOp);
+      Value outerLane = affineFloorDivConst(rewriter, loc, args.lane, numOutRows, anchorOp);
+      Value batch = getBatchLaneIndex(args.lane, numOutRows, numKSlices, numOutHSlices, rewriter, loc);
+      Value sliceLane = affineModConst(rewriter, loc, outerLane, numKSlices * numOutHSlices, anchorOp);
+      Value kSlice = affineModConst(rewriter, loc, sliceLane, numKSlices, anchorOp);
+      Value hSlice = affineFloorDivConst(rewriter, loc, sliceLane, numKSlices, anchorOp);
+      Value kOffset = affineMulConst(rewriter, loc, kSlice, crossbarSize.getValue(), anchorOp);
+      Value hOffset = affineMulConst(rewriter, loc, hSlice, crossbarSize.getValue(), anchorOp);
+
+      auto aTileType =
+        RankedTensorType::get({1, static_cast<int64_t>(crossbarSize.getValue())}, aType.getElementType());
+      auto bTileType = RankedTensorType::get(
+        {static_cast<int64_t>(crossbarSize.getValue()), static_cast<int64_t>(crossbarSize.getValue())},
+        bType.getElementType());
+      auto pieceType =
+        RankedTensorType::get({1, static_cast<int64_t>(crossbarSize.getValue())}, partialPiecesType.getElementType());
+
+      Value aTile =
+        extractBatchedATile(args.inputs.front(), aBatchCount, batch, row, kOffset, aTileType, rewriter, loc);
+      Value bTile =
+        extractBatchedBTile(args.weights.front(), bBatchCount, batch, kOffset, hOffset, bTileType, rewriter, loc);
+      Value piece = spatial::SpatVMMOp::create(rewriter, loc, pieceType, bTile, aTile).getResult();
+
+      SmallVector<OpFoldResult> pieceOffsets {args.lane, rewriter.getIndexAttr(0)};
+      SmallVector<OpFoldResult> pieceSizes {rewriter.getIndexAttr(1), rewriter.getIndexAttr(crossbarSize.getValue())};
+      createParallelInsertSliceIntoBatchOutput(
+        rewriter, loc, piece, args.outputs.front(), pieceOffsets, pieceSizes, getUnitStrides(rewriter, 2));
+    });
+  assert(succeeded(batchOp) && "expected batched MatMul VMM construction to succeed");
+  return *batchOp;
+}
+
+static Value extractDynamicBatchedBColumn(Value matrix,
+                                          int64_t sourceBatchCount,
+                                          Value batch,
+                                          Value column,
+                                          RankedTensorType vectorType,
+                                          PatternRewriter& rewriter,
+                                          Location loc) {
+  auto columnSliceType = RankedTensorType::get({1, vectorType.getDimSize(1), 1}, vectorType.getElementType());
+  SmallVector<OpFoldResult> offsets {sourceBatchCount == 1 ? OpFoldResult(rewriter.getIndexAttr(0))
+                                                           : OpFoldResult(batch),
+                                     rewriter.getIndexAttr(0),
+                                     column};
+  SmallVector<OpFoldResult> sizes {
+    rewriter.getIndexAttr(1), rewriter.getIndexAttr(vectorType.getDimSize(1)), rewriter.getIndexAttr(1)};
+  SmallVector<OpFoldResult> strides {rewriter.getIndexAttr(1), rewriter.getIndexAttr(1), rewriter.getIndexAttr(1)};
+  Value columnSlice = tensor::ExtractSliceOp::create(rewriter, loc, columnSliceType, matrix, offsets, sizes, strides);
+  auto collapsedType = RankedTensorType::get({vectorType.getDimSize(1)}, vectorType.getElementType());
+  Value collapsed = tensor::CollapseShapeOp::create(rewriter,
+                                                    loc,
+                                                    collapsedType,
+                                                    columnSlice,
+                                                    SmallVector<ReassociationIndices> {
+                                                      {0, 1, 2}
+  })
+                      .getResult();
+  return tensor::ExpandShapeOp::create(rewriter,
+                                       loc,
+                                       vectorType,
+                                       collapsed,
+                                       SmallVector<ReassociationIndices> {
+                                         {0, 1}
+  })
+    .getResult();
+}
+
+static Value extractDynamicBatchedBRow(Value matrix,
+                                       int64_t sourceBatchCount,
+                                       Value batch,
+                                       Value row,
+                                       RankedTensorType vectorType,
+                                       PatternRewriter& rewriter,
+                                       Location loc) {
+  auto rowSliceType = RankedTensorType::get({1, 1, vectorType.getDimSize(1)}, vectorType.getElementType());
+  SmallVector<OpFoldResult> offsets {sourceBatchCount == 1 ? OpFoldResult(rewriter.getIndexAttr(0))
+                                                           : OpFoldResult(batch),
+                                     row,
+                                     rewriter.getIndexAttr(0)};
+  SmallVector<OpFoldResult> sizes {
+    rewriter.getIndexAttr(1), rewriter.getIndexAttr(1), rewriter.getIndexAttr(vectorType.getDimSize(1))};
+  auto rowSlice =
+    tensor::ExtractSliceOp::create(rewriter, loc, rowSliceType, matrix, offsets, sizes, getUnitStrides(rewriter, 3));
+  return tensor::CollapseShapeOp::create(rewriter,
+                                         loc,
+                                         vectorType,
+                                         rowSlice,
+                                         SmallVector<ReassociationIndices> {
+                                           {0, 1},
+                                           {2}
+  });
+}
+
+static Value extractDynamicBatchedRowVector(Value matrix,
+                                            int64_t sourceBatchCount,
+                                            Value batch,
+                                            Value row,
+                                            RankedTensorType vectorType,
+                                            PatternRewriter& rewriter,
+                                            Location loc) {
+  auto rowSliceType = RankedTensorType::get({1, 1, vectorType.getDimSize(1)}, vectorType.getElementType());
+  SmallVector<OpFoldResult> offsets {sourceBatchCount == 1 ? OpFoldResult(rewriter.getIndexAttr(0))
+                                                           : OpFoldResult(batch),
+                                     row,
+                                     rewriter.getIndexAttr(0)};
+  SmallVector<OpFoldResult> sizes {
+    rewriter.getIndexAttr(1), rewriter.getIndexAttr(1), rewriter.getIndexAttr(vectorType.getDimSize(1))};
+  auto rowSlice =
+    tensor::ExtractSliceOp::create(rewriter, loc, rowSliceType, matrix, offsets, sizes, getUnitStrides(rewriter, 3));
+  return tensor::CollapseShapeOp::create(rewriter,
+                                         loc,
+                                         vectorType,
+                                         rowSlice,
+                                         SmallVector<ReassociationIndices> {
+                                           {0, 1},
+                                           {2}
+  });
+}
+
+static spatial::SpatComputeBatch createBatchedVvdmulBatch(Value a,
+                                                          int64_t aBatchCount,
+                                                          Value b,
+                                                          int64_t bBatchCount,
+                                                          RankedTensorType aType,
+                                                          RankedTensorType bType,
+                                                          RankedTensorType scalarPiecesType,
+                                                          RankedTensorType outType,
+                                                          bool bAlreadyTransposed,
+                                                          PatternRewriter& rewriter,
+                                                          Location loc) {
+  const int64_t numBatches = outType.getDimSize(0);
+  const int64_t numOutRows = outType.getDimSize(1);
+  const int64_t numOutCols = outType.getDimSize(2);
+  const int64_t reductionSize = aType.getDimSize(2);
+  const int64_t laneCount = numBatches * numOutRows * numOutCols;
+  auto batchOp = createSpatComputeBatch(
+    rewriter,
+    loc,
+    TypeRange {scalarPiecesType},
+    laneCount,
+    ValueRange {},
+    ValueRange {a, b},
+    [&](detail::SpatComputeBatchBodyArgs args) {
+      Operation* anchorOp = rewriter.getInsertionBlock()->getParentOp();
+      Value batch = affineFloorDivConst(rewriter, loc, args.lane, numOutRows * numOutCols, anchorOp);
+      Value batchLane = affineModConst(rewriter, loc, args.lane, numOutRows * numOutCols, anchorOp);
+      Value row = affineFloorDivConst(rewriter, loc, batchLane, numOutCols, anchorOp);
+      Value column = affineModConst(rewriter, loc, batchLane, numOutCols, anchorOp);
+
+      auto vectorType = RankedTensorType::get({1, reductionSize}, aType.getElementType());
+      auto scalarType = RankedTensorType::get({1, 1}, outType.getElementType());
+      Value aVector =
+        extractDynamicBatchedRowVector(args.inputs[0], aBatchCount, batch, row, vectorType, rewriter, loc);
+      Value bVector =
+        bAlreadyTransposed
+          ? extractDynamicBatchedBRow(args.inputs[1], bBatchCount, batch, column, vectorType, rewriter, loc)
+          : extractDynamicBatchedBColumn(args.inputs[1], bBatchCount, batch, column, vectorType, rewriter, loc);
+      Value scalar = spatial::SpatVVDMulOp::create(rewriter, loc, scalarType, aVector, bVector).getResult();
+      SmallVector<OpFoldResult> outputOffsets {args.lane, rewriter.getIndexAttr(0)};
+      SmallVector<OpFoldResult> scalarSizes {rewriter.getIndexAttr(1), rewriter.getIndexAttr(1)};
+      createParallelInsertSliceIntoBatchOutput(
+        rewriter, loc, scalar, args.outputs.front(), outputOffsets, scalarSizes, getUnitStrides(rewriter, 2));
+    });
+  assert(succeeded(batchOp) && "expected batched MatMul VVDMul construction to succeed");
+  return *batchOp;
+}
+
+static Value createBatchedDynamicOutputCompute(Value scalarPieces,
+                                               RankedTensorType scalarPiecesType,
+                                               RankedTensorType outType,
+                                               PatternRewriter& rewriter,
+                                               Location loc) {
+  const int64_t laneCount = scalarPiecesType.getDimSize(0);
+  const int64_t numOutRows = outType.getDimSize(1);
+  const int64_t numOutCols = outType.getDimSize(2);
+  auto scalarType = RankedTensorType::get({1, 1}, outType.getElementType());
+  auto outputScalarType = RankedTensorType::get({1, 1, 1}, outType.getElementType());
+
+  auto computeOp =
+    createSpatCompute<1>(rewriter, loc, TypeRange {outType}, {}, ValueRange {scalarPieces}, [&](Value pieces) {
+      Value outputInit =
+        tensor::EmptyOp::create(rewriter, loc, outType.getShape(), outType.getElementType()).getResult();
+      Value c0 = getOrCreateIndexConstant(rewriter, rewriter.getInsertionBlock()->getParentOp(), 0);
+      Value c1 = getOrCreateIndexConstant(rewriter, rewriter.getInsertionBlock()->getParentOp(), 1);
+      Value cLaneCount = getOrCreateIndexConstant(rewriter, rewriter.getInsertionBlock()->getParentOp(), laneCount);
+      auto loop = scf::ForOp::create(rewriter, loc, c0, cLaneCount, c1, ValueRange {outputInit});
+      rewriter.setInsertionPointToStart(loop.getBody());
+
+      Value lane = loop.getInductionVar();
+      Value outputAcc = loop.getRegionIterArgs().front();
+      Operation* anchorOp = rewriter.getInsertionBlock()->getParentOp();
+      Value batch = affineFloorDivConst(rewriter, loc, lane, numOutRows * numOutCols, anchorOp);
+      Value batchLane = affineModConst(rewriter, loc, lane, numOutRows * numOutCols, anchorOp);
+      Value row = affineFloorDivConst(rewriter, loc, batchLane, numOutCols, anchorOp);
+      Value column = affineModConst(rewriter, loc, batchLane, numOutCols, anchorOp);
+      SmallVector<OpFoldResult> scalarOffsets {lane, rewriter.getIndexAttr(0)};
+      SmallVector<OpFoldResult> scalarSizes {rewriter.getIndexAttr(1), rewriter.getIndexAttr(1)};
+      Value scalar = tensor::ExtractSliceOp::create(
+        rewriter, loc, scalarType, pieces, scalarOffsets, scalarSizes, getUnitStrides(rewriter, 2));
+      Value expanded = tensor::ExpandShapeOp::create(rewriter,
+                                                     loc,
+                                                     outputScalarType,
+                                                     scalar,
+                                                     SmallVector<ReassociationIndices> {
+                                                       {0},
+                                                       {1, 2}
+      });
+      SmallVector<OpFoldResult> outputOffsets {batch, row, column};
+      SmallVector<OpFoldResult> outputSizes {
+        rewriter.getIndexAttr(1), rewriter.getIndexAttr(1), rewriter.getIndexAttr(1)};
+      scf::YieldOp::create(
+        rewriter,
+        loc,
+        tensor::InsertSliceOp::create(
+          rewriter, loc, expanded, outputAcc, outputOffsets, outputSizes, getUnitStrides(rewriter, 3))
+          .getResult());
+
+      rewriter.setInsertionPointAfter(loop);
+      spatial::SpatYieldOp::create(rewriter, loc, loop.getResult(0));
+    });
+  return computeOp.getResult(0);
+}
+
+static Value transposeBatchedOutput(Value value, RankedTensorType outputType, PatternRewriter& rewriter, Location loc) {
+  auto transposeCompute =
+    createSpatCompute<1>(rewriter, loc, TypeRange {outputType}, {}, ValueRange {value}, [&](Value input) {
+      Value transposed = ONNXTransposeOp::create(rewriter, loc, outputType, input, rewriter.getI64ArrayAttr({0, 2, 1}));
+      spatial::SpatYieldOp::create(rewriter, loc, transposed);
+    });
+  return transposeCompute.getResult(0);
+}
+
+static Value extractBatchedReductionPiece(Value partialPiecesArg,
+                                          Value batch,
+                                          Value hSlice,
+                                          int64_t kSlice,
+                                          RankedTensorType pieceType,
+                                          int64_t numKSlices,
+                                          int64_t numOutHSlices,
+                                          int64_t numOutRows,
+                                          PatternRewriter& rewriter,
+                                          Location loc) {
+  Operation* anchorOp = rewriter.getInsertionBlock()->getParentOp();
+  Value batchOffset = affineMulConst(rewriter, loc, batch, numOutRows * numKSlices * numOutHSlices, anchorOp);
+  Value hOffset = affineMulConst(rewriter, loc, hSlice, numKSlices * numOutRows, anchorOp);
+  Value kOffset = getOrCreateIndexConstant(rewriter, rewriter.getInsertionBlock()->getParentOp(), kSlice * numOutRows);
+  Value batchAndHSlice = arith::AddIOp::create(rewriter, loc, batchOffset, hOffset);
+  Value pieceOffset = arith::AddIOp::create(rewriter, loc, batchAndHSlice, kOffset);
+  SmallVector<OpFoldResult> offsets {pieceOffset, rewriter.getIndexAttr(0)};
+  SmallVector<OpFoldResult> sizes {rewriter.getIndexAttr(numOutRows), rewriter.getIndexAttr(crossbarSize.getValue())};
+  return tensor::ExtractSliceOp::create(
+    rewriter, loc, pieceType, partialPiecesArg, offsets, sizes, getUnitStrides(rewriter, 2));
+}
+
+static Value reduceBatchedPartialPiecesForHSlice(Value partialPiecesArg,
+                                                 Value batch,
+                                                 Value hSlice,
+                                                 RankedTensorType pieceType,
+                                                 int64_t numKSlices,
+                                                 int64_t numOutHSlices,
+                                                 int64_t numOutRows,
+                                                 PatternRewriter& rewriter,
+                                                 Location loc) {
+  SmallVector<Value> activePieces;
+  activePieces.reserve(numKSlices);
+  for (int64_t kSlice = 0; kSlice < numKSlices; ++kSlice)
+    activePieces.push_back(extractBatchedReductionPiece(
+      partialPiecesArg, batch, hSlice, kSlice, pieceType, numKSlices, numOutHSlices, numOutRows, rewriter, loc));
+
+  while (activePieces.size() > 1) {
+    SmallVector<Value> nextPieces;
+    nextPieces.reserve((activePieces.size() + 1) / 2);
+    for (size_t pieceIndex = 0; pieceIndex + 1 < activePieces.size(); pieceIndex += 2)
+      nextPieces.push_back(
+        spatial::SpatVAddOp::create(rewriter, loc, pieceType, activePieces[pieceIndex], activePieces[pieceIndex + 1])
+          .getResult());
+    if (activePieces.size() % 2 != 0)
+      nextPieces.push_back(activePieces.back());
+    activePieces = std::move(nextPieces);
+  }
+
+  return activePieces.front();
+}
+
+static Value createBatchedReductionCompute(Value partialPieces,
+                                           RankedTensorType partialPiecesType,
+                                           RankedTensorType outType,
+                                           RankedTensorType paddedOutType,
+                                           int64_t numBatches,
+                                           int64_t numKSlices,
+                                           PatternRewriter& rewriter,
+                                           Location loc) {
+  auto computeOp = createSpatCompute<1>(
+    rewriter, loc, TypeRange {outType}, {}, ValueRange {partialPieces}, [&](Value partialPiecesArg) {
+      const int64_t numOutRows = outType.getDimSize(1);
+      const int64_t numOutHSlices = ceilIntegerDivide(outType.getDimSize(2), crossbarSize.getValue());
+      auto pieceType = RankedTensorType::get({numOutRows, static_cast<int64_t>(crossbarSize.getValue())},
+                                             partialPiecesType.getElementType());
+      auto outputSliceType = RankedTensorType::get({1, numOutRows, static_cast<int64_t>(crossbarSize.getValue())},
+                                                   partialPiecesType.getElementType());
+
+      Value outputInit =
+        tensor::EmptyOp::create(rewriter, loc, paddedOutType.getShape(), paddedOutType.getElementType()).getResult();
+      Value c0 = getOrCreateIndexConstant(rewriter, rewriter.getInsertionBlock()->getParentOp(), 0);
+      Value c1 = getOrCreateIndexConstant(rewriter, rewriter.getInsertionBlock()->getParentOp(), 1);
+      Value cNumBatches = getOrCreateIndexConstant(rewriter, rewriter.getInsertionBlock()->getParentOp(), numBatches);
+      Value cNumOutHSlices =
+        getOrCreateIndexConstant(rewriter, rewriter.getInsertionBlock()->getParentOp(), numOutHSlices);
+
+      auto batchLoop = scf::ForOp::create(rewriter, loc, c0, cNumBatches, c1, ValueRange {outputInit});
+      rewriter.setInsertionPointToStart(batchLoop.getBody());
+      Value batch = batchLoop.getInductionVar();
+      Value batchAcc = batchLoop.getRegionIterArgs().front();
+
+      auto hLoop = scf::ForOp::create(rewriter, loc, c0, cNumOutHSlices, c1, ValueRange {batchAcc});
+      rewriter.setInsertionPointToStart(hLoop.getBody());
+      Value hSlice = hLoop.getInductionVar();
+      Value outputAcc = hLoop.getRegionIterArgs().front();
+
+      Value reduced = reduceBatchedPartialPiecesForHSlice(
+        partialPiecesArg, batch, hSlice, pieceType, numKSlices, numOutHSlices, numOutRows, rewriter, loc);
+      Value expandedReduced = tensor::ExpandShapeOp::create(rewriter,
+                                                            loc,
+                                                            outputSliceType,
+                                                            reduced,
+                                                            SmallVector<ReassociationIndices> {
+                                                              {0, 1},
+                                                              {2}
+      });
+      Value hOffset =
+        affineMulConst(rewriter, loc, hSlice, crossbarSize.getValue(), rewriter.getInsertionBlock()->getParentOp());
+      SmallVector<OpFoldResult> outputOffsets {batch, rewriter.getIndexAttr(0), hOffset};
+      SmallVector<OpFoldResult> outputSizes {
+        rewriter.getIndexAttr(1), rewriter.getIndexAttr(numOutRows), rewriter.getIndexAttr(crossbarSize.getValue())};
+      scf::YieldOp::create(
+        rewriter,
+        loc,
+        tensor::InsertSliceOp::create(
+          rewriter, loc, expandedReduced, outputAcc, outputOffsets, outputSizes, getUnitStrides(rewriter, 3))
+          .getResult());
+
+      rewriter.setInsertionPointAfter(hLoop);
+      scf::YieldOp::create(rewriter, loc, hLoop.getResult(0));
+
+      rewriter.setInsertionPointAfter(batchLoop);
+      Value paddedOutput = batchLoop.getResult(0);
+      Value result = paddedOutput;
+      if (paddedOutType != outType) {
+        SmallVector<OpFoldResult> outputOffsets {
+          rewriter.getIndexAttr(0), rewriter.getIndexAttr(0), rewriter.getIndexAttr(0)};
+        SmallVector<OpFoldResult> outputSizes {rewriter.getIndexAttr(numBatches),
+                                               rewriter.getIndexAttr(outType.getDimSize(1)),
+                                               rewriter.getIndexAttr(outType.getDimSize(2))};
+        result = tensor::ExtractSliceOp::create(
+          rewriter, loc, outType, paddedOutput, outputOffsets, outputSizes, getUnitStrides(rewriter, 3));
+      }
+      spatial::SpatYieldOp::create(rewriter, loc, result);
+    });
+  return computeOp.getResult(0);
+}
+
+struct MatMulShapeInfo {
+  RankedTensorType lhsType;
+  RankedTensorType rhsType;
+  RankedTensorType outType;
+  SmallVector<int64_t> batchShape;
+  int64_t lhsBatch;
+  int64_t rhsBatch;
+  int64_t batch;
+  int64_t m;
+  int64_t k;
+  int64_t n;
+};
+
+static FailureOr<MatMulShapeInfo> analyzeMatMulShape(ONNXMatMulOp matmulOp) {
   auto lhsType = dyn_cast<RankedTensorType>(matmulOp.getA().getType());
   auto rhsType = dyn_cast<RankedTensorType>(matmulOp.getB().getType());
   auto outType = dyn_cast<RankedTensorType>(matmulOp.getY().getType());
@@ -215,8 +686,7 @@ struct MatMulToGemm : OpRewritePattern<ONNXMatMulOp> {
     return failure();
   if (lhsType.getRank() < 2 || rhsType.getRank() < 2 || outType.getRank() < 2)
     return failure();
-    if (!haveStaticPositiveShape(lhsType.getShape()) || !haveStaticPositiveShape(rhsType.getShape())
-        || !haveStaticPositiveShape(outType.getShape()))
+  if (!hasStaticPositiveShape(lhsType) || !hasStaticPositiveShape(rhsType) || !hasStaticPositiveShape(outType))
     return failure();
 
   SmallVector<int64_t> lhsBatchShape(lhsType.getShape().begin(), lhsType.getShape().end() - 2);
@@ -224,10 +694,10 @@ struct MatMulToGemm : OpRewritePattern<ONNXMatMulOp> {
   auto batchShape = inferSupportedBatchShape(lhsBatchShape, rhsBatchShape);
   if (failed(batchShape))
     return failure();
+
   const int64_t lhsBatch = lhsBatchShape.empty() ? 1 : getStaticShapeElementCount(lhsBatchShape);
   const int64_t rhsBatch = rhsBatchShape.empty() ? 1 : getStaticShapeElementCount(rhsBatchShape);
   const int64_t batch = batchShape->empty() ? 1 : getStaticShapeElementCount(*batchShape);
-
   const int64_t m = lhsType.getDimSize(lhsType.getRank() - 2);
   const int64_t k = lhsType.getDimSize(lhsType.getRank() - 1);
   const int64_t rhsK = rhsType.getDimSize(rhsType.getRank() - 2);
@@ -246,30 +716,38 @@ struct MatMulToGemm : OpRewritePattern<ONNXMatMulOp> {
       return failure();
   }
 
-    Location loc = matmulOp.getLoc();
-    bool useTransposedForm = isHostFoldableValue(matmulOp.getA()) && !isHostFoldableValue(matmulOp.getB());
+  return MatMulShapeInfo {lhsType, rhsType, outType, *batchShape, lhsBatch, rhsBatch, batch, m, k, n};
+}
 
-    Value lhs = collapseBatchDims(matmulOp.getA(), lhsBatch, m, k, rewriter, loc);
-    Value rhs = collapseBatchDims(matmulOp.getB(), rhsBatch, k, n, rewriter, loc);
-    int64_t lhsBatchForGemm = lhsBatch;
-    int64_t rhsBatchForGemm = rhsBatch;
-    int64_t gemmM = m;
-    int64_t gemmK = k;
-    int64_t gemmN = n;
+struct MatMulToGemm : OpRewritePattern<ONNXMatMulOp> {
+  using OpRewritePattern::OpRewritePattern;
+
+  LogicalResult matchAndRewrite(ONNXMatMulOp matmulOp, PatternRewriter& rewriter) const override {
+    auto shapeInfo = analyzeMatMulShape(matmulOp);
+    if (failed(shapeInfo) || shapeInfo->outType.getRank() != 2)
+      return failure();
+
+    Location loc = matmulOp.getLoc();
+    bool useTransposedForm = isCompileTimeComputable(matmulOp.getA()) && !isCompileTimeComputable(matmulOp.getB());
+
+    Value lhs = collapseBatchDims(matmulOp.getA(), shapeInfo->lhsBatch, shapeInfo->m, shapeInfo->k, rewriter, loc);
+    Value rhs = collapseBatchDims(matmulOp.getB(), shapeInfo->rhsBatch, shapeInfo->k, shapeInfo->n, rewriter, loc);
+    int64_t lhsBatchForGemm = shapeInfo->lhsBatch;
+    int64_t rhsBatchForGemm = shapeInfo->rhsBatch;
+    int64_t gemmM = shapeInfo->m;
+    int64_t gemmK = shapeInfo->k;
+    int64_t gemmN = shapeInfo->n;
     if (useTransposedForm) {
       lhs = transposeLastTwoDimsInCompute(matmulOp.getB(), rewriter, loc);
-      lhsBatchForGemm = rhsBatch;
+      lhsBatchForGemm = shapeInfo->rhsBatch;
       rhs = transposeLastTwoDims(matmulOp.getA(), rewriter, loc);
-      rhsBatchForGemm = lhsBatch;
-      gemmM = n;
-      gemmN = m;
+      rhsBatchForGemm = shapeInfo->lhsBatch;
+      gemmM = shapeInfo->n;
+      gemmN = shapeInfo->m;
     }
 
-    auto gemmType = RankedTensorType::get({gemmM, gemmN}, outType.getElementType());
-    auto batchedOutType = RankedTensorType::get({1, m, n}, outType.getElementType());
+    auto gemmType = RankedTensorType::get({gemmM, gemmN}, shapeInfo->outType.getElementType());
     Value none = ONNXNoneOp::create(rewriter, loc, rewriter.getNoneType());
-
-    if (outType.getRank() == 2) {
     Value lhsMatrix = extractBatchMatrix(lhs, /*batchIndex=*/0, lhsBatchForGemm, gemmM, gemmK, rewriter, loc);
     Value rhsMatrix = extractBatchMatrix(rhs, /*batchIndex=*/0, rhsBatchForGemm, gemmK, gemmN, rewriter, loc);
     Value gemmResult = ONNXGemmOp::create(rewriter,
@@ -285,8 +763,9 @@ struct MatMulToGemm : OpRewritePattern<ONNXMatMulOp> {
                          .getY();
     if (useTransposedForm) {
       auto transposeCompute =
-          createSpatCompute<1>(rewriter, loc, TypeRange {outType}, {}, gemmResult, [&](Value input) {
-            Value transposed = ONNXTransposeOp::create(rewriter, loc, outType, input, rewriter.getI64ArrayAttr({1, 0}));
+        createSpatCompute<1>(rewriter, loc, TypeRange {shapeInfo->outType}, {}, gemmResult, [&](Value input) {
+          Value transposed =
+            ONNXTransposeOp::create(rewriter, loc, shapeInfo->outType, input, rewriter.getI64ArrayAttr({1, 0}));
           spatial::SpatYieldOp::create(rewriter, loc, transposed);
         });
       gemmResult = transposeCompute.getResult(0);
@@ -294,48 +773,115 @@ struct MatMulToGemm : OpRewritePattern<ONNXMatMulOp> {
     rewriter.replaceOp(matmulOp, gemmResult);
     return success();
   }
+};
 
-    SmallVector<Value> batchResults;
-    batchResults.reserve(batch);
-    for (int64_t batchIdx = 0; batchIdx < batch; batchIdx++) {
-      Value lhsMatrix = extractBatchMatrix(lhs, batchIdx, lhsBatchForGemm, gemmM, gemmK, rewriter, loc);
-      Value rhsMatrix = extractBatchMatrix(rhs, batchIdx, rhsBatchForGemm, gemmK, gemmN, rewriter, loc);
-      Value gemmResult = ONNXGemmOp::create(rewriter,
-                                            loc,
-                                            gemmType,
-                                            lhsMatrix,
-                                            rhsMatrix,
-                                            none,
-                                            rewriter.getF32FloatAttr(1.0f),
-                                            rewriter.getF32FloatAttr(1.0f),
-                                            rewriter.getBoolAttr(false),
-                                            rewriter.getBoolAttr(false))
-                           .getY();
-      auto batchResultCompute =
-        createSpatCompute<1>(rewriter, loc, TypeRange {batchedOutType}, {}, gemmResult, [&](Value input) {
-          Value resultMatrix = input;
+struct MatMulBatchedToSpatialComputes : OpRewritePattern<ONNXMatMulOp> {
+  using OpRewritePattern::OpRewritePattern;
+
+  LogicalResult matchAndRewrite(ONNXMatMulOp matmulOp, PatternRewriter& rewriter) const override {
+    auto shapeInfo = analyzeMatMulShape(matmulOp);
+    if (failed(shapeInfo))
+      return failure();
+    if (shapeInfo->outType.getRank() == 2)
+      return failure();
+
+    Location loc = matmulOp.getLoc();
+    bool useTransposedForm = isCompileTimeComputable(matmulOp.getA()) && !isCompileTimeComputable(matmulOp.getB());
+
+    Value lhs = collapseBatchDims(matmulOp.getA(), shapeInfo->lhsBatch, shapeInfo->m, shapeInfo->k, rewriter, loc);
+    Value rhs = collapseBatchDims(matmulOp.getB(), shapeInfo->rhsBatch, shapeInfo->k, shapeInfo->n, rewriter, loc);
+    int64_t lhsBatchForGemm = shapeInfo->lhsBatch;
+    int64_t rhsBatchForGemm = shapeInfo->rhsBatch;
+    int64_t gemmM = shapeInfo->m;
+    int64_t gemmK = shapeInfo->k;
+    int64_t gemmN = shapeInfo->n;
     if (useTransposedForm) {
-            resultMatrix = ONNXTransposeOp::create(rewriter,
-                                                   loc,
-                                                   RankedTensorType::get({m, n}, outType.getElementType()),
-                                                   input,
-                                                   rewriter.getI64ArrayAttr({1, 0}));
-          }
-          Value expanded = tensor::ExpandShapeOp::create(rewriter,
-                                                         loc,
-                                                         batchedOutType,
-                                                         resultMatrix,
-                                                         SmallVector<ReassociationIndices> {
-                                                           {0, 1},
-                                                           {2}
-          });
-          spatial::SpatYieldOp::create(rewriter, loc, expanded);
-        });
-      batchResults.push_back(batchResultCompute.getResult(0));
+      lhs = transposeLastTwoDimsInCompute(matmulOp.getB(), rewriter, loc);
+      lhsBatchForGemm = shapeInfo->rhsBatch;
+      rhs = transposeLastTwoDims(matmulOp.getA(), rewriter, loc);
+      rhsBatchForGemm = shapeInfo->lhsBatch;
+      gemmM = shapeInfo->n;
+      gemmN = shapeInfo->m;
     }
 
-    Value result = concatValues(batchResults, /*axis=*/0, rewriter, loc);
-    result = expandBatchDims(result, outType, batchShape->size(), rewriter, loc);
+    lhs = ensureBatchedTensor(lhs, lhsBatchForGemm, gemmM, gemmK, rewriter, loc);
+    rhs = ensureBatchedTensor(rhs, rhsBatchForGemm, gemmK, gemmN, rewriter, loc);
+    auto lhsBatchedType = cast<RankedTensorType>(lhs.getType());
+    auto rhsBatchedType = cast<RankedTensorType>(rhs.getType());
+    auto directOutType = RankedTensorType::get({shapeInfo->batch, gemmM, gemmN}, shapeInfo->outType.getElementType());
+
+    if (isCompileTimeComputable(rhs)) {
+      const int64_t numKSlices = ceilIntegerDivide(gemmK, crossbarSize.getValue());
+      const int64_t numOutHSlices = ceilIntegerDivide(gemmN, crossbarSize.getValue());
+      const int64_t paddedReductionSize = numKSlices * static_cast<int64_t>(crossbarSize.getValue());
+      const int64_t paddedOutCols = numOutHSlices * static_cast<int64_t>(crossbarSize.getValue());
+      auto paddedLhsType = RankedTensorType::get(
+        {lhsBatchForGemm, gemmM, paddedReductionSize}, lhsBatchedType.getElementType(), lhsBatchedType.getEncoding());
+      auto paddedRhsType = RankedTensorType::get({shapeInfo->batch, paddedReductionSize, paddedOutCols},
+                                                 rhsBatchedType.getElementType(),
+                                                 rhsBatchedType.getEncoding());
+      auto paddedOutType =
+        RankedTensorType::get({shapeInfo->batch, gemmM, paddedOutCols}, shapeInfo->outType.getElementType());
+
+      auto paddedRhs = materializePaddedBatchedWeight(rhs, rhsBatchForGemm, shapeInfo->batch, paddedRhsType, rewriter);
+      if (succeeded(paddedRhs)) {
+        Value paddedLhs = createPaddedBatchedInputCompute(lhs, paddedLhsType, rewriter, loc);
+        const int64_t laneCount = shapeInfo->batch * gemmM * numKSlices * numOutHSlices;
+        auto partialPiecesType = RankedTensorType::get({laneCount, static_cast<int64_t>(crossbarSize.getValue())},
+                                                       shapeInfo->outType.getElementType());
+        auto batchOp = createBatchedVmmBatch(paddedLhs,
+                                             *paddedRhs,
+                                             paddedLhsType,
+                                             lhsBatchForGemm,
+                                             paddedRhsType,
+                                             rhsBatchForGemm,
+                                             partialPiecesType,
+                                             gemmM,
+                                             numKSlices,
+                                             numOutHSlices,
+                                             rewriter,
+                                             loc);
+        Value result = createBatchedReductionCompute(batchOp.getResult(0),
+                                                     partialPiecesType,
+                                                     directOutType,
+                                                     paddedOutType,
+                                                     shapeInfo->batch,
+                                                     numKSlices,
+                                                     rewriter,
+                                                     loc);
+        if (useTransposedForm)
+          result = transposeBatchedOutput(
+            result,
+            RankedTensorType::get({shapeInfo->batch, shapeInfo->m, shapeInfo->n}, shapeInfo->outType.getElementType()),
+            rewriter,
+            loc);
+        result = expandBatchDims(result, shapeInfo->outType, shapeInfo->batchShape.size(), rewriter, loc);
+        rewriter.replaceOp(matmulOp, result);
+        return success();
+      }
+    }
+    const int64_t laneCount = shapeInfo->batch * gemmM * gemmN;
+    auto scalarPiecesType = RankedTensorType::get({laneCount, 1}, shapeInfo->outType.getElementType());
+    auto batchOp = createBatchedVvdmulBatch(lhs,
+                                            lhsBatchForGemm,
+                                            rhs,
+                                            rhsBatchForGemm,
+                                            lhsBatchedType,
+                                            rhsBatchedType,
+                                            scalarPiecesType,
+                                            directOutType,
+                                            false,
+                                            rewriter,
+                                            loc);
+    Value result =
+      createBatchedDynamicOutputCompute(batchOp.getResult(0), scalarPiecesType, directOutType, rewriter, loc);
+    if (useTransposedForm)
+      result = transposeBatchedOutput(
+        result,
+        RankedTensorType::get({shapeInfo->batch, shapeInfo->m, shapeInfo->n}, shapeInfo->outType.getElementType()),
+        rewriter,
+        loc);
+    result = expandBatchDims(result, shapeInfo->outType, shapeInfo->batchShape.size(), rewriter, loc);
     rewriter.replaceOp(matmulOp, result);
     return success();
   }
@@ -344,7 +890,7 @@ struct MatMulToGemm : OpRewritePattern<ONNXMatMulOp> {
 } // namespace
 
 void populateMatMulRewritePatterns(RewritePatternSet& patterns, MLIRContext* ctx) {
-  patterns.insert<MatMulToGemm>(ctx);
+  patterns.insert<MatMulToGemm, MatMulBatchedToSpatialComputes>(ctx);
 }
 
 } // namespace onnx_mlir

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

@@ -1,13 +1,16 @@
+#include "mlir/Dialect/Affine/IR/AffineOps.h"
 #include "mlir/Dialect/Tensor/IR/Tensor.h"
 #include "mlir/Transforms/DialectConversion.h"
 
 #include "llvm/ADT/SmallVector.h"
 
 #include <algorithm>
+#include <numeric>
 
+#include "src/Accelerators/PIM/Common/IR/AffineUtils.hpp"
 #include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Common/Common.hpp"
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/ConversionPatterns.hpp"
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/HostFoldability.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/CompileTime.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Patterns.hpp"
 #include "src/Accelerators/PIM/Dialect/Spatial/SpatialOps.hpp"
 #include "src/Dialect/ONNX/ONNXOps.hpp"
 
@@ -16,26 +19,6 @@ using namespace mlir;
 namespace onnx_mlir {
 namespace {
 
-static SmallVector<int64_t> normalizeAxes(ArrayAttr axesAttr, int64_t rank) {
-  SmallVector<int64_t> normalizedAxes;
-  if (!axesAttr) {
-    normalizedAxes.reserve(rank);
-    for (int64_t axis = 0; axis < rank; axis++)
-      normalizedAxes.push_back(axis);
-    return normalizedAxes;
-  }
-
-  normalizedAxes.reserve(axesAttr.size());
-  for (Attribute attr : axesAttr) {
-    int64_t axis = cast<IntegerAttr>(attr).getInt();
-    normalizedAxes.push_back(axis >= 0 ? axis : rank + axis);
-  }
-
-  llvm::sort(normalizedAxes);
-  normalizedAxes.erase(std::unique(normalizedAxes.begin(), normalizedAxes.end()), normalizedAxes.end());
-  return normalizedAxes;
-}
-
 static SmallVector<bool> buildReducedAxesMask(ArrayRef<int64_t> axes, int64_t rank) {
   SmallVector<bool> reducedAxes(rank, false);
   for (int64_t axis : axes) {
@@ -50,6 +33,184 @@ static RankedTensorType getAllOnesType(RankedTensorType inputType, Type elementT
   return RankedTensorType::get(SmallVector<int64_t>(inputType.getRank(), 1), elementType);
 }
 
+static RankedTensorType getKeepdimsType(RankedTensorType inputType, Type elementType, ArrayRef<bool> reducedAxes) {
+  SmallVector<int64_t> shape;
+  shape.reserve(inputType.getRank());
+  for (auto [dim, isReduced] : llvm::zip_equal(inputType.getShape(), reducedAxes))
+    shape.push_back(isReduced ? 1 : dim);
+  return RankedTensorType::get(shape, elementType, inputType.getEncoding());
+}
+
+static RankedTensorType getCompactKeptType(RankedTensorType inputType, Type elementType, ArrayRef<bool> reducedAxes) {
+  SmallVector<int64_t> shape;
+  for (auto [dim, isReduced] : llvm::zip_equal(inputType.getShape(), reducedAxes))
+    if (!isReduced)
+      shape.push_back(dim);
+  return RankedTensorType::get(shape, elementType, inputType.getEncoding());
+}
+
+static RankedTensorType getReducedSliceType(RankedTensorType inputType, ArrayRef<bool> reducedAxes) {
+  SmallVector<int64_t> shape;
+  shape.reserve(inputType.getRank());
+  for (auto [dim, isReduced] : llvm::zip_equal(inputType.getShape(), reducedAxes))
+    shape.push_back(isReduced ? dim : 1);
+  return RankedTensorType::get(shape, inputType.getElementType(), inputType.getEncoding());
+}
+
+static RankedTensorType getLanePackedKeepdimsType(int64_t laneCount, RankedTensorType leafType) {
+  SmallVector<int64_t> shape(leafType.getShape().begin(), leafType.getShape().end());
+  shape.front() = laneCount;
+  return RankedTensorType::get(shape, leafType.getElementType(), leafType.getEncoding());
+}
+
+static SmallVector<int64_t> getKeptAxes(ArrayRef<bool> reducedAxes) {
+  SmallVector<int64_t> keptAxes;
+  for (auto [axis, isReduced] : llvm::enumerate(reducedAxes))
+    if (!isReduced)
+      keptAxes.push_back(static_cast<int64_t>(axis));
+  return keptAxes;
+}
+
+static Value
+computeLaneIndex(Value lane, int64_t stride, int64_t dimSize, ConversionPatternRewriter& rewriter, Location loc) {
+  if (dimSize == 1)
+    return getOrCreateIndexConstant(rewriter, rewriter.getInsertionBlock()->getParentOp(), 0);
+
+  MLIRContext* context = rewriter.getContext();
+  AffineExpr d0 = getAffineDimExpr(0, context);
+  AffineExpr expr = d0;
+  if (stride != 1)
+    expr = expr.floorDiv(stride);
+  if (dimSize != 1)
+    expr = expr % dimSize;
+  return createOrFoldAffineApply(rewriter, loc, expr, ValueRange {lane}, rewriter.getInsertionBlock()->getParentOp());
+}
+
+static FailureOr<Value> buildReduceMeanKeepdimsBatch(Value input,
+                                                     ArrayRef<bool> reducedAxes,
+                                                     RankedTensorType batchType,
+                                                     RankedTensorType leafType,
+                                                     ConversionPatternRewriter& rewriter,
+                                                     Location loc) {
+  auto inputType = cast<RankedTensorType>(input.getType());
+  auto sliceType = getReducedSliceType(inputType, reducedAxes);
+  SmallVector<int64_t> keptAxes = getKeptAxes(reducedAxes);
+
+  int64_t laneCount = 1;
+  SmallVector<int64_t> keptAxisStrides(keptAxes.size(), 1);
+  for (int64_t index = static_cast<int64_t>(keptAxes.size()) - 1; index >= 0; --index) {
+    keptAxisStrides[index] = laneCount;
+    int64_t dimSize = inputType.getDimSize(keptAxes[index]);
+    if (dimSize <= 0)
+      return failure();
+    if (laneCount > std::numeric_limits<int32_t>::max() / dimSize)
+      return failure();
+    laneCount *= dimSize;
+  }
+
+  SmallVector<OpFoldResult> sliceOffsets;
+  SmallVector<OpFoldResult> sliceSizes;
+  SmallVector<OpFoldResult> insertOffsets;
+  SmallVector<OpFoldResult> insertSizes(inputType.getRank(), rewriter.getIndexAttr(1));
+  SmallVector<OpFoldResult> unitStrides = getUnitStrides(rewriter, inputType.getRank());
+  sliceOffsets.reserve(inputType.getRank());
+  sliceSizes.reserve(inputType.getRank());
+  insertOffsets.reserve(inputType.getRank());
+
+  auto batchOp =
+    createSpatComputeBatch(rewriter,
+                           loc,
+                           TypeRange {batchType},
+                           laneCount,
+                           {},
+                           ValueRange {input},
+                           [&](detail::SpatComputeBatchBodyArgs args) {
+                             size_t keptAxisIndex = 0;
+                             sliceOffsets.clear();
+                             sliceSizes.clear();
+                             insertOffsets.clear();
+                             for (auto [axis, isReduced] : llvm::enumerate(reducedAxes)) {
+                               if (isReduced) {
+                                 sliceOffsets.push_back(rewriter.getIndexAttr(0));
+                                 sliceSizes.push_back(rewriter.getIndexAttr(inputType.getDimSize(axis)));
+                                 continue;
+                               }
+
+                               Value axisIndex = computeLaneIndex(
+                                 args.lane, keptAxisStrides[keptAxisIndex], inputType.getDimSize(axis), rewriter, loc);
+                               ++keptAxisIndex;
+                               sliceOffsets.push_back(axisIndex);
+                               sliceSizes.push_back(rewriter.getIndexAttr(1));
+                             }
+
+                             insertOffsets.push_back(args.lane);
+                             insertOffsets.append(inputType.getRank() - 1, rewriter.getIndexAttr(0));
+
+                             Value slice = tensor::ExtractSliceOp::create(
+                               rewriter, loc, sliceType, args.inputs.front(), sliceOffsets, sliceSizes, unitStrides);
+                             Value reduced = spatial::SpatVAvgOp::create(rewriter, loc, leafType, slice).getResult();
+                             createParallelInsertSliceIntoBatchOutput(
+                               rewriter, loc, reduced, args.outputs.front(), insertOffsets, insertSizes, unitStrides);
+                           });
+  if (failed(batchOp))
+    return failure();
+  return (*batchOp).getResult(0);
+}
+
+static Value buildKeepdimsFromLanePackedBatch(Value batchValue,
+                                              RankedTensorType keepdimsType,
+                                              RankedTensorType compactKeptType,
+                                              ArrayRef<bool> reducedAxes,
+                                              ConversionPatternRewriter& rewriter,
+                                              Location loc) {
+  auto batchType = cast<RankedTensorType>(batchValue.getType());
+  if (batchType == keepdimsType)
+    return batchValue;
+
+  SmallVector<ReassociationIndices> collapseToFlat {{}};
+  for (int64_t axis = 0; axis < batchType.getRank(); ++axis)
+    collapseToFlat.front().push_back(axis);
+
+  SmallVector<ReassociationIndices> expandFlatToCompact(1);
+  for (int64_t axis = 0; axis < compactKeptType.getRank(); ++axis)
+    expandFlatToCompact.front().push_back(axis);
+
+  SmallVector<ReassociationIndices> expandCompactToKeepdims;
+  ReassociationIndices pendingLeadingReducedAxes;
+  for (auto [axis, isReduced] : llvm::enumerate(reducedAxes)) {
+    if (isReduced) {
+      if (expandCompactToKeepdims.empty())
+        pendingLeadingReducedAxes.push_back(axis);
+      else
+        expandCompactToKeepdims.back().push_back(axis);
+      continue;
+    }
+
+    expandCompactToKeepdims.emplace_back();
+    auto& group = expandCompactToKeepdims.back();
+    group.append(pendingLeadingReducedAxes.begin(), pendingLeadingReducedAxes.end());
+    pendingLeadingReducedAxes.clear();
+    group.push_back(axis);
+  }
+  if (!pendingLeadingReducedAxes.empty())
+    expandCompactToKeepdims.back().append(pendingLeadingReducedAxes.begin(), pendingLeadingReducedAxes.end());
+
+  auto reshapeCompute =
+    createSpatCompute<1>(rewriter, loc, TypeRange {keepdimsType}, {}, ValueRange {batchValue}, [&](Value input) {
+      auto flatType =
+        RankedTensorType::get({batchType.getDimSize(0)}, batchType.getElementType(), batchType.getEncoding());
+      Value flat = tensor::CollapseShapeOp::create(rewriter, loc, flatType, input, collapseToFlat);
+      Value compact = flat;
+      if (compactKeptType != flatType)
+        compact = tensor::ExpandShapeOp::create(rewriter, loc, compactKeptType, flat, expandFlatToCompact);
+      Value keepdims = compact;
+      if (keepdimsType != compactKeptType)
+        keepdims = tensor::ExpandShapeOp::create(rewriter, loc, keepdimsType, compact, expandCompactToKeepdims);
+      spatial::SpatYieldOp::create(rewriter, loc, keepdims);
+    });
+  return reshapeCompute.getResult(0);
+}
+
 static SmallVector<ReassociationIndices> buildCollapseReassociation(ArrayRef<bool> reducedAxes) {
   SmallVector<ReassociationIndices> reassociation;
   ReassociationIndices currentGroup;
@@ -72,56 +233,6 @@ static SmallVector<ReassociationIndices> buildCollapseReassociation(ArrayRef<boo
   return reassociation;
 }
 
-static Value
-createAverageCompute(Value input, RankedTensorType resultType, ConversionPatternRewriter& rewriter, Location loc) {
-  constexpr size_t numInputs = 1;
-  auto computeOp = createSpatCompute<numInputs>(rewriter, loc, resultType, {}, ValueRange {input}, [&](Value x) {
-    auto avgOp = spatial::SpatVAvgOp::create(rewriter, loc, resultType, x);
-    spatial::SpatYieldOp::create(rewriter, loc, avgOp.getResult());
-  });
-  return computeOp.getResult(0);
-}
-
-static Value concatValues(ValueRange inputs, int64_t axis, ConversionPatternRewriter& rewriter, Location loc) {
-  auto firstType = cast<RankedTensorType>(inputs.front().getType());
-  SmallVector<int64_t> outputShape(firstType.getShape().begin(), firstType.getShape().end());
-  int64_t concatDimSize = 0;
-  for (Value input : inputs)
-    concatDimSize += cast<RankedTensorType>(input.getType()).getDimSize(axis);
-  outputShape[axis] = concatDimSize;
-  auto resultType = RankedTensorType::get(outputShape, firstType.getElementType(), firstType.getEncoding());
-
-  if (llvm::all_of(inputs, isHostFoldableValue))
-    return createSpatConcat(rewriter, loc, axis, inputs);
-
-  auto concatCompute = createSpatCompute(rewriter, loc, TypeRange {resultType}, {}, inputs, [&](ValueRange args) {
-    spatial::SpatYieldOp::create(rewriter, loc, createSpatConcat(rewriter, loc, axis, args));
-  });
-  return concatCompute.getResult(0);
-}
-
-static Value buildReduceMeanKeepdims(Value input,
-                                     ArrayRef<bool> reducedAxes,
-                                     int64_t axis,
-                                     RankedTensorType leafType,
-                                     ConversionPatternRewriter& rewriter,
-                                     Location loc) {
-  int64_t rank = cast<RankedTensorType>(input.getType()).getRank();
-  if (axis == rank)
-    return createAverageCompute(input, leafType, rewriter, loc);
-
-  if (reducedAxes[axis])
-    return buildReduceMeanKeepdims(input, reducedAxes, axis + 1, leafType, rewriter, loc);
-
-  SmallVector<Value> slices = sliceTensor(input, axis, /*sliceSize=*/1, rewriter, loc);
-  SmallVector<Value> reducedSlices;
-  reducedSlices.reserve(slices.size());
-  for (Value slice : slices)
-    reducedSlices.push_back(buildReduceMeanKeepdims(slice, reducedAxes, axis + 1, leafType, rewriter, loc));
-
-  return concatValues(reducedSlices, axis, rewriter, loc);
-}
-
 static Value squeezeReducedAxes(Value keepdimsValue,
                                 RankedTensorType resultType,
                                 ArrayRef<bool> reducedAxes,
@@ -129,13 +240,13 @@ static Value squeezeReducedAxes(Value keepdimsValue,
                                 Location loc) {
   if (resultType.getRank() == 0) {
     SmallVector<Value> indices(cast<RankedTensorType>(keepdimsValue.getType()).getRank(),
-                               arith::ConstantIndexOp::create(rewriter, loc, 0));
+                               getOrCreateIndexConstant(rewriter, rewriter.getInsertionBlock()->getParentOp(), 0));
     Value element = tensor::ExtractOp::create(rewriter, loc, keepdimsValue, indices);
     return tensor::FromElementsOp::create(rewriter, loc, resultType, ValueRange {element});
   }
 
   auto reassociation = buildCollapseReassociation(reducedAxes);
-  if (isHostFoldableValue(keepdimsValue))
+  if (isCompileTimeComputable(keepdimsValue))
     return tensor::CollapseShapeOp::create(rewriter, loc, resultType, keepdimsValue, reassociation).getResult();
 
   auto squeezeCompute =
@@ -156,16 +267,33 @@ struct ReduceMeanToSpatialCompute : OpConversionPattern<ONNXReduceMeanV13Op> {
     auto resultType = dyn_cast<RankedTensorType>(reduceMeanOp.getReduced().getType());
     if (!inputType || !resultType || !inputType.hasStaticShape() || !resultType.hasStaticShape())
       return failure();
+    if (inputType.getRank() == 0) {
+      rewriter.replaceOp(reduceMeanOp, adaptor.getData());
+      return success();
+    }
 
-    SmallVector<int64_t> axes = normalizeAxes(reduceMeanOp.getAxesAttr(), inputType.getRank());
-    SmallVector<bool> reducedAxes = buildReducedAxesMask(axes, inputType.getRank());
+    auto axes = normalizeAxesChecked(std::optional<ArrayAttr>(reduceMeanOp.getAxesAttr()), inputType.getRank());
+    if (failed(axes))
+      return failure();
+    SmallVector<bool> reducedAxes = buildReducedAxesMask(*axes, inputType.getRank());
     if (reducedAxes.empty() && inputType.getRank() != 0)
       return failure();
 
     Location loc = reduceMeanOp.getLoc();
     RankedTensorType leafType = getAllOnesType(inputType, resultType.getElementType());
+    RankedTensorType compactKeptType = getCompactKeptType(inputType, resultType.getElementType(), reducedAxes);
+    RankedTensorType keepdimsType = getKeepdimsType(inputType, resultType.getElementType(), reducedAxes);
+    int64_t laneCount = 1;
+    for (int64_t dim : compactKeptType.getShape())
+      laneCount *= dim;
+    RankedTensorType batchType = getLanePackedKeepdimsType(laneCount, leafType);
+
+    auto lanePackedKeepdims =
+      buildReduceMeanKeepdimsBatch(adaptor.getData(), reducedAxes, batchType, leafType, rewriter, loc);
+    if (failed(lanePackedKeepdims))
+      return failure();
     Value reducedKeepdims =
-      buildReduceMeanKeepdims(adaptor.getData(), reducedAxes, /*axis=*/0, leafType, rewriter, loc);
+      buildKeepdimsFromLanePackedBatch(*lanePackedKeepdims, keepdimsType, compactKeptType, reducedAxes, rewriter, loc);
 
     if (reduceMeanOp.getKeepdims() != 0) {
       rewriter.replaceOp(reduceMeanOp, reducedKeepdims);

src/PIM/Conversion/ONNXToSpatial/Patterns/NN/Pool.cpp

@@ -23,43 +23,26 @@ using namespace mlir;
 namespace onnx_mlir {
 namespace {
 
-template <typename ArrayAttrT>
-static int64_t getI64(ArrayAttrT arrayAttr, size_t index) {
-  return cast<IntegerAttr>(arrayAttr[index]).getInt();
-}
-
-template <typename ArrayAttrT>
-static int64_t getOptionalI64(std::optional<ArrayAttrT> arrayAttr, size_t index, int64_t defaultValue) {
-  return arrayAttr ? getI64(*arrayAttr, index) : defaultValue;
-}
-
-static Value materializeContiguousTile(ConversionPatternRewriter& rewriter, Location loc, Value tile) {
+static Value materializeTileTensor(ConversionPatternRewriter& rewriter, Location loc, Value tile) {
   auto tileType = cast<RankedTensorType>(tile.getType());
   Value empty = tensor::EmptyOp::create(rewriter, loc, tileType.getShape(), tileType.getElementType());
-
-  SmallVector<OpFoldResult> offsets(tileType.getRank(), rewriter.getIndexAttr(0));
-  SmallVector<OpFoldResult> sizes;
-  sizes.reserve(tileType.getRank());
-  for (int64_t dimSize : tileType.getShape())
-    sizes.push_back(rewriter.getIndexAttr(dimSize));
-  SmallVector<OpFoldResult> strides(tileType.getRank(), rewriter.getIndexAttr(1));
-
-  return tensor::InsertSliceOp::create(rewriter, loc, tile, empty, offsets, sizes, strides);
+  return insertStaticSlice(rewriter, loc, tile, empty, getZeroOffsets(rewriter, tileType.getRank()));
 }
 
 static Value
 createPoolFillElement(ConversionPatternRewriter& rewriter, Location loc, Type elementType, bool useMinimumValue) {
+  Operation* anchorOp = rewriter.getInsertionBlock()->getParentOp();
   if (!useMinimumValue)
-    return arith::ConstantOp::create(rewriter, loc, elementType, rewriter.getZeroAttr(elementType));
+    return getOrCreateConstant(rewriter, anchorOp, rewriter.getZeroAttr(elementType), elementType);
 
   if (auto floatType = dyn_cast<FloatType>(elementType)) {
     auto minValue = llvm::APFloat::getInf(floatType.getFloatSemantics(), /*Negative=*/true);
-    return arith::ConstantOp::create(rewriter, loc, elementType, rewriter.getFloatAttr(floatType, minValue));
+    return getOrCreateConstant(rewriter, anchorOp, rewriter.getFloatAttr(floatType, minValue), elementType);
   }
 
   if (auto integerType = dyn_cast<IntegerType>(elementType)) {
     auto minValue = llvm::APInt::getSignedMinValue(integerType.getWidth());
-    return arith::ConstantOp::create(rewriter, loc, elementType, rewriter.getIntegerAttr(integerType, minValue));
+    return getOrCreateConstant(rewriter, anchorOp, rewriter.getIntegerAttr(integerType, minValue), elementType);
   }
 
   llvm_unreachable("unsupported pool element type");
@@ -166,7 +149,7 @@ static FailureOr<Value> createAverageScaleTensor(ConversionPatternRewriter& rewr
   }
 
   auto scaleAttr = DenseElementsAttr::get(scaleType, scaleValues);
-  return arith::ConstantOp::create(rewriter, loc, scaleType, scaleAttr).getResult();
+  return getOrCreateConstant(rewriter, rewriter.getInsertionBlock()->getParentOp(), scaleAttr, scaleType);
 }
 
 template <typename PoolOp>
@@ -197,12 +180,12 @@ struct PoolToSpatialComputeBase : public OpConversionPattern<PoolOp> {
     const int64_t inputWidth = xType.getDimSize(3);
     const int64_t outputHeight = outType.getDimSize(2);
     const int64_t outputWidth = outType.getDimSize(3);
-    const int64_t kernelHeight = getI64(kernelAttr, 0);
-    const int64_t kernelWidth = getI64(kernelAttr, 1);
-    const int64_t strideHeight = getOptionalI64(poolOp.getStrides(), 0, 1);
-    const int64_t strideWidth = getOptionalI64(poolOp.getStrides(), 1, 1);
-    const int64_t dilationHeight = getOptionalI64(poolOp.getDilations(), 0, 1);
-    const int64_t dilationWidth = getOptionalI64(poolOp.getDilations(), 1, 1);
+    const int64_t kernelHeight = getI64Attr(kernelAttr, 0);
+    const int64_t kernelWidth = getI64Attr(kernelAttr, 1);
+    const int64_t strideHeight = getOptionalI64Attr(poolOp.getStrides(), 0, 1);
+    const int64_t strideWidth = getOptionalI64Attr(poolOp.getStrides(), 1, 1);
+    const int64_t dilationHeight = getOptionalI64Attr(poolOp.getDilations(), 0, 1);
+    const int64_t dilationWidth = getOptionalI64Attr(poolOp.getDilations(), 1, 1);
 
     int64_t padTop = 0;
     int64_t padLeft = 0;
@@ -212,10 +195,10 @@ struct PoolToSpatialComputeBase : public OpConversionPattern<PoolOp> {
     if (auto padsAttr = poolOp.getPads()) {
       if (padsAttr->size() != 4)
         return rewriter.notifyMatchFailure(poolOp, "pads must have four elements.");
-      padTop = getI64(*padsAttr, 0);
-      padLeft = getI64(*padsAttr, 1);
-      padBottom = getI64(*padsAttr, 2);
-      padRight = getI64(*padsAttr, 3);
+      padTop = getI64Attr(*padsAttr, 0);
+      padLeft = getI64Attr(*padsAttr, 1);
+      padBottom = getI64Attr(*padsAttr, 2);
+      padRight = getI64Attr(*padsAttr, 3);
     }
     else {
       StringRef autoPad = poolOp.getAutoPad();
@@ -283,13 +266,14 @@ struct PoolToSpatialComputeBase : public OpConversionPattern<PoolOp> {
           createPaddedPoolInput(rewriter, loc, poolOp, xArg, xType, padTop, padLeft, padBottom, padRight);
         Value pooledOutputInit = tensor::EmptyOp::create(rewriter, loc, outType.getShape(), outType.getElementType());
 
-        Value c0 = arith::ConstantIndexOp::create(rewriter, loc, 0);
-        Value c1 = arith::ConstantIndexOp::create(rewriter, loc, 1);
-        Value cOutputPatchCount = arith::ConstantIndexOp::create(rewriter, loc, outputPatchCount);
-        Value cOutputPixelsPerBatch = arith::ConstantIndexOp::create(rewriter, loc, outputHeight * outputWidth);
-        Value cOutputWidth = arith::ConstantIndexOp::create(rewriter, loc, outputWidth);
-        Value cStrideHeight = arith::ConstantIndexOp::create(rewriter, loc, strideHeight);
-        Value cStrideWidth = arith::ConstantIndexOp::create(rewriter, loc, strideWidth);
+        Operation* anchorOp = rewriter.getInsertionBlock()->getParentOp();
+        Value c0 = getOrCreateIndexConstant(rewriter, anchorOp, 0);
+        Value c1 = getOrCreateIndexConstant(rewriter, anchorOp, 1);
+        Value cOutputPatchCount = getOrCreateIndexConstant(rewriter, anchorOp, outputPatchCount);
+        Value cOutputPixelsPerBatch = getOrCreateIndexConstant(rewriter, anchorOp, outputHeight * outputWidth);
+        Value cOutputWidth = getOrCreateIndexConstant(rewriter, anchorOp, outputWidth);
+        Value cStrideHeight = getOrCreateIndexConstant(rewriter, anchorOp, strideHeight);
+        Value cStrideWidth = getOrCreateIndexConstant(rewriter, anchorOp, strideWidth);
 
         auto outputLoop = scf::ForOp::create(rewriter, loc, c0, cOutputPatchCount, c1, ValueRange {pooledOutputInit});
         rewriter.setInsertionPointToStart(outputLoop.getBody());
@@ -314,14 +298,14 @@ struct PoolToSpatialComputeBase : public OpConversionPattern<PoolOp> {
           for (int64_t kernelH = 0; kernelH < kernelHeight; ++kernelH) {
             Value paddedInH = windowBaseH;
             if (kernelH * dilationHeight != 0) {
-              Value kernelHOffset = arith::ConstantIndexOp::create(rewriter, loc, kernelH * dilationHeight);
+              Value kernelHOffset = getOrCreateIndexConstant(rewriter, anchorOp, kernelH * dilationHeight);
               paddedInH = arith::AddIOp::create(rewriter, loc, paddedInH, kernelHOffset);
             }
 
             for (int64_t kernelW = 0; kernelW < kernelWidth; ++kernelW) {
               Value paddedInW = windowBaseW;
               if (kernelW * dilationWidth != 0) {
-                Value kernelWOffset = arith::ConstantIndexOp::create(rewriter, loc, kernelW * dilationWidth);
+                Value kernelWOffset = getOrCreateIndexConstant(rewriter, anchorOp, kernelW * dilationWidth);
                 paddedInW = arith::AddIOp::create(rewriter, loc, paddedInW, kernelWOffset);
               }
 
@@ -335,7 +319,7 @@ struct PoolToSpatialComputeBase : public OpConversionPattern<PoolOp> {
                 rewriter.getIndexAttr(1), rewriter.getIndexAttr(1), rewriter.getIndexAttr(1), rewriter.getIndexAttr(1)};
               Value windowValue =
                 tensor::ExtractSliceOp::create(rewriter, loc, tileType, paddedInput, offsets, sizes, strides);
-              windowValue = materializeContiguousTile(rewriter, loc, windowValue);
+              windowValue = materializeTileTensor(rewriter, loc, windowValue);
               reducedWindow = ReduceOp::create(rewriter, loc, tileType, reducedWindow, windowValue);
             }
           }
@@ -351,7 +335,7 @@ struct PoolToSpatialComputeBase : public OpConversionPattern<PoolOp> {
               rewriter.getIndexAttr(1), rewriter.getIndexAttr(1), rewriter.getIndexAttr(1), rewriter.getIndexAttr(1)};
             Value scaleSlice = tensor::ExtractSliceOp::create(
               rewriter, loc, tileType, averageScaleTensor, scaleOffsets, scaleSizes, scaleStrides);
-            scaleSlice = materializeContiguousTile(rewriter, loc, scaleSlice);
+            scaleSlice = materializeTileTensor(rewriter, loc, scaleSlice);
             reducedWindow = spatial::SpatVMulOp::create(rewriter, loc, tileType, reducedWindow, scaleSlice);
           }
 

src/PIM/Conversion/ONNXToSpatial/Patterns/NN/Softmax.cpp

@@ -4,7 +4,7 @@
 #include "mlir/Transforms/DialectConversion.h"
 
 #include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Common/Common.hpp"
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/ConversionPatterns.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Patterns.hpp"
 #include "src/Accelerators/PIM/Dialect/Spatial/SpatialOps.hpp"
 #include "src/Dialect/ONNX/ONNXOps.hpp"
 
@@ -13,16 +13,6 @@ using namespace mlir;
 namespace onnx_mlir {
 namespace {
 
-static int64_t normalizeAxis(int64_t axis, int64_t rank) { return axis >= 0 ? axis : rank + axis; }
-
-static SmallVector<int64_t> permuteShape(ArrayRef<int64_t> shape, ArrayRef<int64_t> permutation) {
-  SmallVector<int64_t> permutedShape;
-  permutedShape.reserve(permutation.size());
-  for (int64_t axis : permutation)
-    permutedShape.push_back(shape[axis]);
-  return permutedShape;
-}
-
 static Value buildLoopSoftmaxSlice(Value input,
                                    Value accumulator,
                                    RankedTensorType inputType,
@@ -36,7 +26,7 @@ static Value buildLoopSoftmaxSlice(Value input,
 
   SmallVector<OpFoldResult> offsets;
   SmallVector<OpFoldResult> sizes;
-  SmallVector<OpFoldResult> strides(rank, rewriter.getIndexAttr(1));
+  SmallVector<OpFoldResult> strides = getUnitStrides(rewriter, rank);
   offsets.reserve(rank);
   sizes.reserve(rank);
 
@@ -62,9 +52,10 @@ static Value buildLoopSoftmaxNest(Value input,
   if (axis == inputType.getRank() - 1)
     return buildLoopSoftmaxSlice(input, accumulator, inputType, outerIndices, rewriter, loc);
 
-  Value c0 = arith::ConstantIndexOp::create(rewriter, loc, 0);
-  Value c1 = arith::ConstantIndexOp::create(rewriter, loc, 1);
-  Value cUpper = arith::ConstantIndexOp::create(rewriter, loc, inputType.getDimSize(axis));
+  Operation* anchorOp = rewriter.getInsertionBlock()->getParentOp();
+  Value c0 = getOrCreateIndexConstant(rewriter, anchorOp, 0);
+  Value c1 = getOrCreateIndexConstant(rewriter, anchorOp, 1);
+  Value cUpper = getOrCreateIndexConstant(rewriter, anchorOp, inputType.getDimSize(axis));
 
   auto loop = scf::ForOp::create(rewriter, loc, c0, cUpper, c1, ValueRange {accumulator});
   rewriter.setInsertionPointToStart(loop.getBody());
@@ -110,44 +101,29 @@ struct SoftmaxToSpatialCompute : OpConversionPattern<ONNXSoftmaxOp> {
     if (!inputType || !inputType.hasStaticShape())
       return failure();
 
-    int64_t axis = normalizeAxis(softmaxOp.getAxis(), inputType.getRank());
-    if (axis < 0 || axis >= inputType.getRank())
+    auto axis = normalizeAxisChecked(softmaxOp.getAxis(), inputType.getRank());
+    if (failed(axis))
       return failure();
 
     Value input = adaptor.getInput();
     Value result;
-    if (axis == inputType.getRank() - 1) {
+    if (*axis == inputType.getRank() - 1) {
       result = createLoopSoftmaxCompute(input, rewriter, softmaxOp.getLoc());
     }
     else {
       SmallVector<int64_t> permutation;
       permutation.reserve(inputType.getRank());
       for (int64_t dim = 0; dim < inputType.getRank(); ++dim)
-        if (dim != axis)
+        if (dim != *axis)
           permutation.push_back(dim);
-      permutation.push_back(axis);
-
-      SmallVector<int64_t> inversePermutation(inputType.getRank());
-      for (auto [newIndex, oldIndex] : llvm::enumerate(permutation))
-        inversePermutation[oldIndex] = static_cast<int64_t>(newIndex);
+      permutation.push_back(*axis);
+      SmallVector<int64_t> inversePermutation = invertPermutation(permutation);
 
       auto transposedType = RankedTensorType::get(
         permuteShape(inputType.getShape(), permutation), inputType.getElementType(), inputType.getEncoding());
-      auto preTransposeCompute =
-        createSpatCompute<1>(rewriter, softmaxOp.getLoc(), TypeRange {transposedType}, {}, input, [&](Value x) {
-          Value transposed = ONNXTransposeOp::create(
-            rewriter, softmaxOp.getLoc(), transposedType, x, rewriter.getI64ArrayAttr(permutation));
-          spatial::SpatYieldOp::create(rewriter, softmaxOp.getLoc(), transposed);
-        });
-      Value transposedInput = preTransposeCompute.getResult(0);
+      Value transposedInput = transposeMaybeInCompute(input, transposedType, permutation, rewriter, softmaxOp.getLoc());
       Value transposedResult = createLoopSoftmaxCompute(transposedInput, rewriter, softmaxOp.getLoc());
-      auto postTransposeCompute =
-        createSpatCompute<1>(rewriter, softmaxOp.getLoc(), TypeRange {inputType}, {}, transposedResult, [&](Value x) {
-          Value transposed = ONNXTransposeOp::create(
-            rewriter, softmaxOp.getLoc(), inputType, x, rewriter.getI64ArrayAttr(inversePermutation));
-          spatial::SpatYieldOp::create(rewriter, softmaxOp.getLoc(), transposed);
-        });
-      result = postTransposeCompute.getResult(0);
+      result = transposeMaybeInCompute(transposedResult, inputType, inversePermutation, rewriter, softmaxOp.getLoc());
     }
 
     rewriter.replaceOp(softmaxOp, result);

src/PIM/Conversion/ONNXToSpatial/Patterns/Post.cpp

@@ -0,0 +1,288 @@
+#include "mlir/Dialect/Arith/IR/Arith.h"
+#include "mlir/Dialect/Func/IR/FuncOps.h"
+#include "mlir/Dialect/Linalg/IR/Linalg.h"
+#include "mlir/Dialect/Tensor/IR/Tensor.h"
+#include "mlir/IR/IRMapping.h"
+#include "mlir/IR/PatternMatch.h"
+
+#include "llvm/ADT/STLExtras.h"
+#include "llvm/ADT/SmallVector.h"
+
+#include "src/Accelerators/PIM/Common/IR/WeightUtils.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Common/WeightMaterialization.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Patterns.hpp"
+#include "src/Accelerators/PIM/Dialect/Spatial/SpatialOps.hpp"
+#include "src/Dialect/ONNX/ONNXOps.hpp"
+
+using namespace mlir;
+
+namespace onnx_mlir {
+
+namespace {
+
+static bool isWeightMaterializationHelperUser(Operation* op) {
+  return isa<tensor::ExtractSliceOp, tensor::ExpandShapeOp, tensor::CollapseShapeOp, linalg::TransposeOp>(op);
+}
+
+static bool canPromoteInputBlockArgument(BlockArgument arg) {
+  return !arg.use_empty() && llvm::all_of(arg.getUsers(), isWeightMaterializationHelperUser);
+}
+
+static bool canPromoteInputBlockArgument(std::optional<BlockArgument> arg) {
+  return arg && canPromoteInputBlockArgument(*arg);
+}
+
+static bool isDirectConstantValue(Value value) {
+  return isa_and_nonnull<arith::ConstantOp, ONNXConstantOp>(value.getDefiningOp());
+}
+
+struct PromotedOperands {
+  SmallVector<bool> promoteInput;
+  SmallVector<Value> newWeights;
+  SmallVector<Value> newInputs;
+  SmallVector<Type> newInputTypes;
+  SmallVector<Location> newInputLocs;
+};
+
+template <typename ComputeOpTy>
+static bool hasPromotableWeightLikeInputs(ComputeOpTy compute) {
+  for (auto [inputIdx, input] : llvm::enumerate(compute.getInputs())) {
+    if (!isWeightLikeComputeOperand(input))
+      continue;
+    if (isDirectConstantValue(input) && !canPromoteInputBlockArgument(compute.getInputArgument(inputIdx)))
+      continue;
+    return true;
+  }
+  return false;
+}
+
+template <typename ComputeOpTy>
+static FailureOr<PromotedOperands> computePromotedOperands(ComputeOpTy compute) {
+  PromotedOperands promoted;
+  promoted.promoteInput.assign(compute.getInputs().size(), false);
+  promoted.newWeights.append(compute.getWeights().begin(), compute.getWeights().end());
+  promoted.newWeights.reserve(compute.getWeights().size() + compute.getInputs().size());
+  promoted.newInputs.reserve(compute.getInputs().size());
+  promoted.newInputTypes.reserve(compute.getInputs().size());
+  promoted.newInputLocs.reserve(compute.getInputs().size());
+
+  bool needsRewrite = false;
+  for (auto [inputIdx, input] : llvm::enumerate(compute.getInputs())) {
+    if (!isWeightLikeComputeOperand(input))
+      goto keep_input;
+    if (isDirectConstantValue(input) && !canPromoteInputBlockArgument(compute.getInputArgument(inputIdx)))
+      goto keep_input;
+    promoted.promoteInput[inputIdx] = true;
+    promoted.newWeights.push_back(input);
+    needsRewrite = true;
+    continue;
+
+keep_input:
+    promoted.newInputs.push_back(input);
+    promoted.newInputTypes.push_back(input.getType());
+    promoted.newInputLocs.push_back(input.getLoc());
+  }
+
+  if (!needsRewrite)
+    return failure();
+  return promoted;
+}
+
+template <typename ComputeOpTy>
+static LogicalResult mapPromotedInputArguments(ComputeOpTy compute,
+                                               const PromotedOperands& promoted,
+                                               IRRewriter& bodyRewriter,
+                                               IRMapping& mapper,
+                                               std::function<std::optional<BlockArgument>(size_t)> getNewInputArg,
+                                               PatternRewriter& rewriter) {
+  size_t newInputIdx = 0;
+  for (auto [oldInputIdx, input] : llvm::enumerate(compute.getInputs())) {
+    auto oldArg = compute.getInputArgument(oldInputIdx);
+    if (!oldArg)
+      return rewriter.notifyMatchFailure(compute, "missing input block argument during rewrite");
+    if (!promoted.promoteInput[oldInputIdx]) {
+      auto newInputArg = getNewInputArg(newInputIdx++);
+      if (!newInputArg)
+        return rewriter.notifyMatchFailure(compute, "missing rewritten input block argument");
+      mapper.map(*oldArg, *newInputArg);
+      continue;
+    }
+
+    auto clonedValue = materializeWeightLikeValueInBlock(input, bodyRewriter, mapper);
+    if (failed(clonedValue))
+      return rewriter.notifyMatchFailure(compute, "failed to materialize promoted weight-like operand");
+    mapper.map(*oldArg, *clonedValue);
+  }
+  return success();
+}
+
+// Promotes foldable helper chains from runtime inputs to weights to avoid artificial compute inputs.
+struct PromoteWeightLikeComputeInputsPattern : OpRewritePattern<spatial::SpatCompute> {
+  using OpRewritePattern<spatial::SpatCompute>::OpRewritePattern;
+
+  LogicalResult matchAndRewrite(spatial::SpatCompute compute, PatternRewriter& rewriter) const override {
+    auto promoted = computePromotedOperands(compute);
+    if (failed(promoted))
+      return rewriter.notifyMatchFailure(compute, "no weight-like inputs to promote");
+    Block& oldBlock = compute.getBody().front();
+
+    rewriter.setInsertionPointAfter(compute);
+    auto newCompute = spatial::SpatCompute::create(
+      rewriter, compute.getLoc(), compute.getResultTypes(), promoted->newWeights, promoted->newInputs);
+    SmallVector<Type> newBlockArgTypes;
+    SmallVector<Location> newBlockArgLocs;
+    for (Value weight : promoted->newWeights) {
+      newBlockArgTypes.push_back(weight.getType());
+      newBlockArgLocs.push_back(weight.getLoc());
+    }
+    llvm::append_range(newBlockArgTypes, promoted->newInputTypes);
+    llvm::append_range(newBlockArgLocs, promoted->newInputLocs);
+    auto* newBlock = rewriter.createBlock(
+      &newCompute.getBody(), newCompute.getBody().end(), TypeRange(newBlockArgTypes), newBlockArgLocs);
+    newCompute.getProperties().setOperandSegmentSizes(
+      {static_cast<int>(promoted->newWeights.size()), static_cast<int>(promoted->newInputs.size())});
+    rewriter.setInsertionPointToStart(newBlock);
+
+    IRRewriter bodyRewriter(rewriter.getContext());
+    bodyRewriter.setInsertionPointToStart(newBlock);
+
+    IRMapping mapper;
+    for (auto [weightIndex, weight] : llvm::enumerate(compute.getWeights())) {
+      auto oldWeightArg = compute.getWeightArgument(weightIndex);
+      auto newWeightArg = newCompute.getWeightArgument(weightIndex);
+      if (!oldWeightArg || !newWeightArg)
+        return rewriter.notifyMatchFailure(compute, "missing compute weight block argument during rewrite");
+      mapper.map(*oldWeightArg, *newWeightArg);
+    }
+    if (failed(mapPromotedInputArguments(
+          compute,
+          *promoted,
+          bodyRewriter,
+          mapper,
+          [&](size_t index) { return newCompute.getInputArgument(index); },
+          rewriter)))
+      return failure();
+
+    for (Operation& op : oldBlock.without_terminator())
+      rewriter.clone(op, mapper);
+
+    auto oldYield = cast<spatial::SpatYieldOp>(oldBlock.getTerminator());
+    SmallVector<Value> newYieldOperands;
+    newYieldOperands.reserve(oldYield.getOutputs().size());
+    for (Value operand : oldYield.getOutputs()) {
+      auto mapped = mapper.lookupOrNull(operand);
+      newYieldOperands.push_back(mapped ? cast<Value>(mapped) : operand);
+    }
+    spatial::SpatYieldOp::create(rewriter, oldYield.getLoc(), newYieldOperands);
+
+    rewriter.replaceOp(compute, newCompute.getResults());
+    return success();
+  }
+};
+
+// Promotes foldable batch helper chains to weights while preserving compact compute_batch IR.
+struct PromoteWeightLikeComputeBatchInputsPattern : OpRewritePattern<spatial::SpatComputeBatch> {
+  using OpRewritePattern<spatial::SpatComputeBatch>::OpRewritePattern;
+
+  LogicalResult matchAndRewrite(spatial::SpatComputeBatch compute, PatternRewriter& rewriter) const override {
+    auto promoted = computePromotedOperands(compute);
+    if (failed(promoted))
+      return rewriter.notifyMatchFailure(compute, "no weight-like batch inputs to promote");
+    Block& oldBlock = compute.getBody().front();
+
+    rewriter.setInsertionPointAfter(compute);
+
+    auto newCompute =
+      spatial::SpatComputeBatch::create(rewriter,
+                                        compute.getLoc(),
+                                        compute.getResultTypes(),
+                                        rewriter.getI32IntegerAttr(static_cast<int32_t>(compute.getLaneCount())),
+                                        promoted->newWeights,
+                                        promoted->newInputs);
+    auto laneArg = compute.getLaneArgument();
+    if (!laneArg)
+      return rewriter.notifyMatchFailure(compute, "missing compute_batch lane block argument");
+    SmallVector<Type> newBlockArgTypes;
+    SmallVector<Location> newBlockArgLocs;
+    newBlockArgTypes.reserve(1 + promoted->newWeights.size() + promoted->newInputTypes.size()
+                             + compute.getNumResults());
+    newBlockArgLocs.reserve(1 + promoted->newWeights.size() + promoted->newInputLocs.size() + compute.getNumResults());
+    newBlockArgTypes.push_back(laneArg->getType());
+    newBlockArgLocs.push_back(laneArg->getLoc());
+    for (Value weight : promoted->newWeights) {
+      newBlockArgTypes.push_back(weight.getType());
+      newBlockArgLocs.push_back(weight.getLoc());
+    }
+    llvm::append_range(newBlockArgTypes, promoted->newInputTypes);
+    llvm::append_range(newBlockArgLocs, promoted->newInputLocs);
+    for (auto [resultIndex, resultType] : llvm::enumerate(compute.getResultTypes())) {
+      auto outputArg = compute.getOutputArgument(resultIndex);
+      if (!outputArg)
+        return rewriter.notifyMatchFailure(compute, "missing compute_batch output block argument");
+      newBlockArgTypes.push_back(resultType);
+      newBlockArgLocs.push_back(outputArg->getLoc());
+    }
+
+    auto* newBlock = rewriter.createBlock(
+      &newCompute.getBody(), newCompute.getBody().end(), TypeRange(newBlockArgTypes), newBlockArgLocs);
+    newCompute.getProperties().setOperandSegmentSizes(
+      {static_cast<int>(promoted->newWeights.size()), static_cast<int>(promoted->newInputs.size())});
+    rewriter.setInsertionPointToStart(newBlock);
+
+    IRRewriter bodyRewriter(rewriter.getContext());
+    bodyRewriter.setInsertionPointToStart(newBlock);
+
+    IRMapping mapper;
+    auto newLaneArg = newCompute.getLaneArgument();
+    if (!newLaneArg)
+      return rewriter.notifyMatchFailure(compute, "missing rewritten compute_batch lane block argument");
+    mapper.map(*laneArg, *newLaneArg);
+    for (auto [weightIndex, weight] : llvm::enumerate(compute.getWeights())) {
+      auto oldWeightArg = compute.getWeightArgument(weightIndex);
+      auto newWeightArg = newCompute.getWeightArgument(weightIndex);
+      if (!oldWeightArg || !newWeightArg)
+        return rewriter.notifyMatchFailure(compute, "missing compute_batch weight block argument during rewrite");
+      mapper.map(*oldWeightArg, *newWeightArg);
+    }
+    if (failed(mapPromotedInputArguments(
+          compute,
+          *promoted,
+          bodyRewriter,
+          mapper,
+          [&](size_t index) { return newCompute.getInputArgument(index); },
+          rewriter)))
+      return failure();
+    for (auto resultIndex : llvm::seq<size_t>(0, compute.getNumResults())) {
+      auto outputArg = compute.getOutputArgument(resultIndex);
+      if (!outputArg)
+        return rewriter.notifyMatchFailure(compute, "missing compute_batch output block argument during rewrite");
+      mapper.map(*outputArg,
+                 newBlock->getArgument(1 + promoted->newWeights.size() + promoted->newInputs.size() + resultIndex));
+    }
+
+    for (Operation& op : oldBlock)
+      rewriter.clone(op, mapper);
+
+    rewriter.replaceOp(compute, newCompute.getResults());
+    return success();
+  }
+};
+
+} // namespace
+
+void populateWeightPromotionPatterns(RewritePatternSet& patterns, MLIRContext* ctx) {
+  patterns.add<PromoteWeightLikeComputeInputsPattern, PromoteWeightLikeComputeBatchInputsPattern>(ctx);
+}
+
+void annotateWeightsConstants(func::FuncOp funcOp) {
+  funcOp.walk([&](arith::ConstantOp constantOp) {
+    if (hasOnlySpatialMvmVmmWeightUses(constantOp.getResult()))
+      markWeightAlways(constantOp);
+  });
+}
+
+bool requiresPostRewrite(spatial::SpatCompute computeOp) { return hasPromotableWeightLikeInputs(computeOp); }
+
+bool requiresPostRewrite(spatial::SpatComputeBatch computeOp) { return hasPromotableWeightLikeInputs(computeOp); }
+
+} // namespace onnx_mlir

src/PIM/Conversion/ONNXToSpatial/Patterns/Pre.cpp

@@ -1,6 +1,5 @@
 #include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Common/Common.hpp"
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/ConversionPatterns.hpp"
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/PrePatterns.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Patterns.hpp"
 
 using namespace mlir;
 
@@ -12,7 +11,7 @@ namespace {
 
 } // namespace
 
-void populatePrePatterns(mlir::RewritePatternSet& patterns, mlir::MLIRContext* ctx) {
+void populateGeneratedPrePatterns(mlir::RewritePatternSet& patterns, mlir::MLIRContext* ctx) {
   patterns.add<onnxToArithConstant>(ctx);
   patterns.add<convAddToConvWithBiasLeft>(ctx);
   patterns.add<convAddToConvWithBiasRight>(ctx);

src/PIM/Conversion/ONNXToSpatial/Patterns/Tensor/Concat.cpp

@@ -3,7 +3,7 @@
 
 #include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Common/Common.hpp"
 #include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Common/ComputeRegionBuilder.hpp"
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/HostFoldability.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/CompileTime.hpp"
 #include "src/Accelerators/PIM/Dialect/Spatial/SpatialOps.hpp"
 #include "src/Dialect/ONNX/ONNXOps.hpp"
 
@@ -20,7 +20,7 @@ struct Concat : public OpConversionPattern<ONNXConcatOp> {
     auto inputs = adaptor.getInputs();
     int64_t axis = adaptor.getAxis();
 
-    if (llvm::all_of(inputs, isHostFoldableValue)) {
+    if (llvm::all_of(inputs, isCompileTimeComputable)) {
       rewriter.replaceOp(maxpoolOp, createSpatConcat(rewriter, maxpoolOp.getLoc(), axis, inputs));
       return success();
     }

src/PIM/Conversion/ONNXToSpatial/Patterns/Tensor/Gather.cpp

@@ -6,7 +6,7 @@
 #include "llvm/ADT/SmallVector.h"
 
 #include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Common/Common.hpp"
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/ConversionPatterns.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Patterns.hpp"
 #include "src/Accelerators/PIM/Dialect/Spatial/SpatialOps.hpp"
 #include "src/Dialect/ONNX/ONNXOps.hpp"
 
@@ -15,24 +15,6 @@ using namespace mlir;
 namespace onnx_mlir {
 namespace {
 
-static int64_t normalizeAxis(int64_t axis, int64_t rank) { return axis >= 0 ? axis : rank + axis; }
-
-static int64_t normalizeIndex(int64_t index, int64_t dimSize) { return index >= 0 ? index : dimSize + index; }
-
-static Value
-extractSliceAt(Value input, int64_t axis, int64_t offset, ConversionPatternRewriter& rewriter, Location loc) {
-  auto inputType = cast<RankedTensorType>(input.getType());
-  SmallVector<OpFoldResult> offsets(inputType.getRank(), rewriter.getIndexAttr(0));
-  SmallVector<OpFoldResult> sizes;
-  SmallVector<OpFoldResult> strides(inputType.getRank(), rewriter.getIndexAttr(1));
-  sizes.reserve(inputType.getRank());
-  for (int64_t dim : inputType.getShape())
-    sizes.push_back(rewriter.getIndexAttr(dim));
-  offsets[axis] = rewriter.getIndexAttr(offset);
-  sizes[axis] = rewriter.getIndexAttr(1);
-  return tensor::ExtractSliceOp::create(rewriter, loc, input, offsets, sizes, strides);
-}
-
 static Value concatGatherSlices(Value data,
                                 int64_t axis,
                                 ArrayRef<int64_t> indices,
@@ -45,7 +27,7 @@ static Value concatGatherSlices(Value data,
     int64_t normalizedIndex = normalizeIndex(index, axisDim);
     if (normalizedIndex < 0 || normalizedIndex >= axisDim)
       return {};
-    slices.push_back(extractSliceAt(data, axis, normalizedIndex, rewriter, loc));
+    slices.push_back(extractAxisSlice(rewriter, loc, data, axis, normalizedIndex, /*size=*/1));
   }
   if (slices.empty())
     return {};
@@ -96,11 +78,11 @@ struct Gather : OpConversionPattern<ONNXGatherOp> {
       return failure();
 
     int64_t rank = dataType.getRank();
-    int64_t axis = normalizeAxis(gatherOp.getAxis(), rank);
-    if (axis < 0 || axis >= rank)
+    auto axis = normalizeAxisChecked(gatherOp.getAxis(), rank);
+    if (failed(axis))
       return failure();
 
-    int64_t axisDim = dataType.getShape()[axis];
+    int64_t axisDim = dataType.getShape()[*axis];
     if (axisDim <= 0)
       return failure();
 
@@ -116,7 +98,7 @@ struct Gather : OpConversionPattern<ONNXGatherOp> {
                            [&](Value data) -> LogicalResult {
                              Value result;
                              if (indicesType.getRank() == 1) {
-                               result = concatGatherSlices(data, axis, flatIndices, axisDim, rewriter, loc);
+                               result = concatGatherSlices(data, *axis, flatIndices, axisDim, rewriter, loc);
                              }
                              else if (indicesType.getRank() == 2) {
                                int64_t rowCount = indicesType.getShape()[0];
@@ -125,12 +107,13 @@ struct Gather : OpConversionPattern<ONNXGatherOp> {
                                rows.reserve(rowCount);
                                for (int64_t row = 0; row < rowCount; ++row) {
                                  ArrayRef<int64_t> rowIndices(flatIndices.data() + row * rowWidth, rowWidth);
-                                 Value gatheredRow = concatGatherSlices(data, axis, rowIndices, axisDim, rewriter, loc);
+                                 Value gatheredRow =
+                                   concatGatherSlices(data, *axis, rowIndices, axisDim, rewriter, loc);
                                  if (!gatheredRow)
                                    return failure();
-                                 rows.push_back(addLeadingGatherDim(gatheredRow, axis, rewriter, loc));
+                                 rows.push_back(addLeadingGatherDim(gatheredRow, *axis, rewriter, loc));
                                }
-                               result = createSpatConcat(rewriter, loc, /*axis=*/axis, rows);
+                               result = createSpatConcat(rewriter, loc, /*axis=*/*axis, rows);
                              }
                              else {
                                return failure();

src/PIM/Conversion/ONNXToSpatial/Patterns/Tensor/Reshape.cpp

@@ -4,8 +4,8 @@
 #include "llvm/ADT/SmallVector.h"
 
 #include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Common/Common.hpp"
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/ConversionPatterns.hpp"
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/HostFoldability.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/CompileTime.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Patterns.hpp"
 #include "src/Accelerators/PIM/Dialect/Spatial/SpatialOps.hpp"
 #include "src/Dialect/ONNX/ONNXOps.hpp"
 
@@ -14,10 +14,6 @@ using namespace mlir;
 namespace onnx_mlir {
 namespace {
 
-static bool haveStaticPositiveShape(ArrayRef<int64_t> shape) {
-  return llvm::all_of(shape, [](int64_t dim) { return dim > 0; });
-}
-
 static bool inferCollapseReassociation(ArrayRef<int64_t> sourceShape,
                                        ArrayRef<int64_t> resultShape,
                                        SmallVector<ReassociationIndices>& reassociation) {
@@ -106,7 +102,7 @@ struct Reshape : OpConversionPattern<ONNXReshapeOp> {
     auto resultType = dyn_cast<RankedTensorType>(reshapeOp.getReshaped().getType());
     if (!sourceType || !resultType || !sourceType.hasStaticShape() || !resultType.hasStaticShape())
       return failure();
-    if (!haveStaticPositiveShape(sourceType.getShape()) || !haveStaticPositiveShape(resultType.getShape()))
+    if (!hasStaticPositiveShape(sourceType) || !hasStaticPositiveShape(resultType))
       return failure();
 
     if (sourceType == resultType) {
@@ -115,17 +111,9 @@ struct Reshape : OpConversionPattern<ONNXReshapeOp> {
     }
 
     auto replaceWithReshape = [&](auto buildReshape) -> LogicalResult {
-      if (isHostFoldableValue(adaptor.getData())) {
-        rewriter.replaceOp(reshapeOp, buildReshape(adaptor.getData()));
-        return success();
-      }
-
-      auto computeOp = createSpatCompute<1>(
-        rewriter, reshapeOp.getLoc(), TypeRange {resultType}, {}, adaptor.getData(), [&](Value data) {
-          Value reshaped = buildReshape(data);
-          spatial::SpatYieldOp::create(rewriter, reshapeOp.getLoc(), reshaped);
-        });
-      rewriter.replaceOp(reshapeOp, computeOp.getResults());
+      Value reshaped =
+        materializeOrComputeUnary(adaptor.getData(), resultType, rewriter, reshapeOp.getLoc(), buildReshape);
+      rewriter.replaceOp(reshapeOp, reshaped);
       return success();
     };
 

src/PIM/Conversion/ONNXToSpatial/Patterns/Tensor/Resize.cpp

@@ -6,7 +6,7 @@
 #include "llvm/ADT/STLExtras.h"
 
 #include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Common/Common.hpp"
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/ConversionPatterns.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Patterns.hpp"
 #include "src/Accelerators/PIM/Dialect/Spatial/SpatialOps.hpp"
 #include "src/Dialect/ONNX/ONNXOps.hpp"
 
@@ -17,9 +17,10 @@ namespace {
 
 static Value buildNearestAsymmetricIndex(
   Value outputIndex, int64_t inputDim, int64_t outputDim, ConversionPatternRewriter& rewriter, Location loc) {
-  Value cInputDim = arith::ConstantIndexOp::create(rewriter, loc, inputDim);
-  Value cOutputDim = arith::ConstantIndexOp::create(rewriter, loc, outputDim);
-  Value cInputDimLast = arith::ConstantIndexOp::create(rewriter, loc, inputDim - 1);
+  Operation* anchorOp = rewriter.getInsertionBlock()->getParentOp();
+  Value cInputDim = getOrCreateIndexConstant(rewriter, anchorOp, inputDim);
+  Value cOutputDim = getOrCreateIndexConstant(rewriter, anchorOp, outputDim);
+  Value cInputDimLast = getOrCreateIndexConstant(rewriter, anchorOp, inputDim - 1);
   Value scaledIndex = arith::MulIOp::create(rewriter, loc, outputIndex, cInputDim);
   Value inputIndex = arith::DivUIOp::create(rewriter, loc, scaledIndex, cOutputDim);
   return arith::MinUIOp::create(rewriter, loc, inputIndex, cInputDimLast);
@@ -37,12 +38,13 @@ static Value buildNearestResizeLoop(Value input,
   SmallVector<OpFoldResult> unitSizes(resultType.getRank(), rewriter.getIndexAttr(1));
   SmallVector<OpFoldResult> unitStrides(resultType.getRank(), rewriter.getIndexAttr(1));
 
-  Value c0 = arith::ConstantIndexOp::create(rewriter, loc, 0);
-  Value c1 = arith::ConstantIndexOp::create(rewriter, loc, 1);
-  Value cOutputN = arith::ConstantIndexOp::create(rewriter, loc, resultType.getDimSize(0));
-  Value cOutputC = arith::ConstantIndexOp::create(rewriter, loc, resultType.getDimSize(1));
-  Value cOutputH = arith::ConstantIndexOp::create(rewriter, loc, resultType.getDimSize(2));
-  Value cOutputW = arith::ConstantIndexOp::create(rewriter, loc, resultType.getDimSize(3));
+  Operation* anchorOp = rewriter.getInsertionBlock()->getParentOp();
+  Value c0 = getOrCreateIndexConstant(rewriter, anchorOp, 0);
+  Value c1 = getOrCreateIndexConstant(rewriter, anchorOp, 1);
+  Value cOutputN = getOrCreateIndexConstant(rewriter, anchorOp, resultType.getDimSize(0));
+  Value cOutputC = getOrCreateIndexConstant(rewriter, anchorOp, resultType.getDimSize(1));
+  Value cOutputH = getOrCreateIndexConstant(rewriter, anchorOp, resultType.getDimSize(2));
+  Value cOutputW = getOrCreateIndexConstant(rewriter, anchorOp, resultType.getDimSize(3));
 
   Value outputInit = tensor::EmptyOp::create(rewriter, loc, resultType.getShape(), elemType);
 
@@ -58,24 +60,21 @@ static Value buildNearestResizeLoop(Value input,
 
   Value outputC = channelLoop.getInductionVar();
   Value outputChannelAcc = channelLoop.getRegionIterArgs().front();
-  Value inputC =
-    buildNearestAsymmetricIndex(outputC, inputType.getDimSize(1), resultType.getDimSize(1), rewriter, loc);
+  Value inputC = buildNearestAsymmetricIndex(outputC, inputType.getDimSize(1), resultType.getDimSize(1), rewriter, loc);
 
   auto heightLoop = scf::ForOp::create(rewriter, loc, c0, cOutputH, c1, ValueRange {outputChannelAcc});
   rewriter.setInsertionPointToStart(heightLoop.getBody());
 
   Value outputH = heightLoop.getInductionVar();
   Value outputHeightAcc = heightLoop.getRegionIterArgs().front();
-  Value inputH =
-    buildNearestAsymmetricIndex(outputH, inputType.getDimSize(2), resultType.getDimSize(2), rewriter, loc);
+  Value inputH = buildNearestAsymmetricIndex(outputH, inputType.getDimSize(2), resultType.getDimSize(2), rewriter, loc);
 
   auto widthLoop = scf::ForOp::create(rewriter, loc, c0, cOutputW, c1, ValueRange {outputHeightAcc});
   rewriter.setInsertionPointToStart(widthLoop.getBody());
 
   Value outputW = widthLoop.getInductionVar();
   Value outputWidthAcc = widthLoop.getRegionIterArgs().front();
-  Value inputW =
-    buildNearestAsymmetricIndex(outputW, inputType.getDimSize(3), resultType.getDimSize(3), rewriter, loc);
+  Value inputW = buildNearestAsymmetricIndex(outputW, inputType.getDimSize(3), resultType.getDimSize(3), rewriter, loc);
 
   SmallVector<OpFoldResult> inputOffsets = {inputN, inputC, inputH, inputW};
   Value inputSlice =
@@ -114,8 +113,8 @@ struct Resize : OpConversionPattern<ONNXResizeOp> {
 
     if (resizeOp.getMode() != "nearest" || resizeOp.getCoordinateTransformationMode() != "asymmetric"
         || resizeOp.getNearestMode() != "floor")
-      return rewriter.notifyMatchFailure(
-        resizeOp, "resize lowering currently supports only nearest + asymmetric + floor.");
+      return rewriter.notifyMatchFailure(resizeOp,
+                                         "resize lowering currently supports only nearest + asymmetric + floor.");
 
     if (llvm::any_of(inputType.getShape(), [](int64_t dim) { return dim <= 0; })
         || llvm::any_of(resultType.getShape(), [](int64_t dim) { return dim <= 0; }))

src/PIM/Conversion/ONNXToSpatial/Patterns/Tensor/Split.cpp

@@ -2,8 +2,8 @@
 #include "mlir/Transforms/DialectConversion.h"
 
 #include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Common/Common.hpp"
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/ConversionPatterns.hpp"
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/HostFoldability.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/CompileTime.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Patterns.hpp"
 #include "src/Accelerators/PIM/Dialect/Spatial/SpatialOps.hpp"
 #include "src/Dialect/ONNX/ONNXOps.hpp"
 
@@ -12,25 +12,6 @@ using namespace mlir;
 namespace onnx_mlir {
 namespace {
 
-static int64_t normalizeAxis(int64_t axis, int64_t rank) { return axis >= 0 ? axis : rank + axis; }
-
-static Value extractSliceAt(
-  Value input, int64_t axis, int64_t offset, int64_t size, ConversionPatternRewriter& rewriter, Location loc) {
-  auto inputType = cast<RankedTensorType>(input.getType());
-  SmallVector<OpFoldResult> offsets(inputType.getRank(), rewriter.getIndexAttr(0));
-  SmallVector<OpFoldResult> sizes;
-  SmallVector<OpFoldResult> strides(inputType.getRank(), rewriter.getIndexAttr(1));
-  sizes.reserve(inputType.getRank());
-  for (int64_t dim : inputType.getShape())
-    sizes.push_back(rewriter.getIndexAttr(dim));
-  offsets[axis] = rewriter.getIndexAttr(offset);
-  sizes[axis] = rewriter.getIndexAttr(size);
-  SmallVector<int64_t> resultShape(inputType.getShape());
-  resultShape[axis] = size;
-  auto resultType = RankedTensorType::get(resultShape, inputType.getElementType());
-  return tensor::ExtractSliceOp::create(rewriter, loc, resultType, input, offsets, sizes, strides);
-}
-
 struct Split : OpConversionPattern<ONNXSplitOp> {
   using OpConversionPattern::OpConversionPattern;
 
@@ -41,8 +22,8 @@ struct Split : OpConversionPattern<ONNXSplitOp> {
       return failure();
 
     int64_t rank = inputType.getRank();
-    int64_t axis = normalizeAxis(splitOp.getAxis(), rank);
-    if (axis < 0 || axis >= rank)
+    auto axis = normalizeAxisChecked(splitOp.getAxis(), rank);
+    if (failed(axis))
       return failure();
 
     SmallVector<Value> outputs;
@@ -58,12 +39,12 @@ struct Split : OpConversionPattern<ONNXSplitOp> {
       if (!resultType || !resultType.hasStaticShape())
         return failure();
       resultTypes.push_back(resultType);
-      sliceSizes.push_back(resultType.getShape()[axis]);
+      sliceSizes.push_back(resultType.getShape()[*axis]);
     }
 
-    if (isHostFoldableValue(adaptor.getInput())) {
+    if (isCompileTimeComputable(adaptor.getInput())) {
       for (int64_t sliceSize : sliceSizes) {
-        outputs.push_back(extractSliceAt(adaptor.getInput(), axis, offset, sliceSize, rewriter, splitOp.getLoc()));
+        outputs.push_back(extractAxisSlice(rewriter, splitOp.getLoc(), adaptor.getInput(), *axis, offset, sliceSize));
         offset += sliceSize;
       }
       rewriter.replaceOp(splitOp, outputs);
@@ -76,7 +57,8 @@ struct Split : OpConversionPattern<ONNXSplitOp> {
         runtimeOutputs.reserve(resultTypes.size());
         int64_t runtimeOffset = 0;
         for (int64_t sliceSize : sliceSizes) {
-          runtimeOutputs.push_back(extractSliceAt(input, axis, runtimeOffset, sliceSize, rewriter, splitOp.getLoc()));
+          runtimeOutputs.push_back(
+            extractAxisSlice(rewriter, splitOp.getLoc(), input, *axis, runtimeOffset, sliceSize));
           runtimeOffset += sliceSize;
         }
         spatial::SpatYieldOp::create(rewriter, splitOp.getLoc(), runtimeOutputs);

src/PIM/Conversion/ONNXToSpatial/Patterns/Tensor/Transpose.cpp

@@ -0,0 +1,119 @@
+#include "mlir/Dialect/Arith/IR/Arith.h"
+#include "mlir/Dialect/Linalg/IR/Linalg.h"
+#include "mlir/Dialect/Tensor/IR/Tensor.h"
+#include "mlir/Transforms/DialectConversion.h"
+
+#include "llvm/ADT/SmallVector.h"
+
+#include "src/Accelerators/PIM/Common/IR/ShapeUtils.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Common/Common.hpp"
+#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Patterns.hpp"
+#include "src/Dialect/ONNX/ONNXOps.hpp"
+
+using namespace mlir;
+
+namespace onnx_mlir {
+namespace {
+
+static Value createTransposeInit(Value input,
+                                 RankedTensorType resultType,
+                                 ArrayRef<int64_t> permutation,
+                                 ConversionPatternRewriter& rewriter,
+                                 Location loc) {
+  SmallVector<OpFoldResult> sizes;
+  sizes.reserve(resultType.getRank());
+  for (auto [resultDim, sourceDim] : llvm::zip_equal(resultType.getShape(), permutation)) {
+    if (!ShapedType::isDynamic(resultDim)) {
+      sizes.push_back(rewriter.getIndexAttr(resultDim));
+      continue;
+    }
+    sizes.push_back(tensor::DimOp::create(rewriter, loc, input, sourceDim).getResult());
+  }
+  return tensor::EmptyOp::create(rewriter, loc, sizes, resultType.getElementType()).getResult();
+}
+
+static FailureOr<Value> materializeTransposedConstant(Value input,
+                                                      RankedTensorType resultType,
+                                                      ArrayRef<int64_t> permutation,
+                                                      ConversionPatternRewriter& rewriter,
+                                                      Location loc) {
+  auto denseAttr = getHostConstDenseElementsAttr(input);
+  if (!denseAttr)
+    return failure();
+
+  auto inputType = dyn_cast<RankedTensorType>(denseAttr.getType());
+  if (!inputType || !inputType.hasStaticShape() || !resultType.hasStaticShape()
+      || inputType.getRank() != resultType.getRank()
+      || static_cast<int64_t>(permutation.size()) != inputType.getRank()) {
+    return failure();
+  }
+
+  if (denseAttr.isSplat())
+    return getOrCreateConstant(rewriter,
+                               rewriter.getInsertionBlock()->getParentOp(),
+                               DenseElementsAttr::get(resultType, denseAttr.getSplatValue<Attribute>()),
+                               resultType);
+
+  SmallVector<Attribute> inputValues(denseAttr.getValues<Attribute>());
+  SmallVector<Attribute> resultValues(inputValues.size());
+  SmallVector<int64_t> inputStrides = computeRowMajorStrides(inputType.getShape());
+  SmallVector<int64_t> resultStrides = computeRowMajorStrides(resultType.getShape());
+  SmallVector<int64_t> inputIndices(inputType.getRank(), 0);
+
+  for (auto [linearIndex, value] : llvm::enumerate(inputValues)) {
+    int64_t remaining = static_cast<int64_t>(linearIndex);
+    for (int64_t dim = 0; dim < inputType.getRank(); ++dim) {
+      inputIndices[dim] = inputStrides.empty() ? 0 : remaining / inputStrides[dim];
+      remaining = inputStrides.empty() ? 0 : remaining % inputStrides[dim];
+    }
+
+    int64_t resultLinearIndex = 0;
+    for (int64_t dim = 0; dim < resultType.getRank(); ++dim)
+      resultLinearIndex += inputIndices[permutation[dim]] * resultStrides[dim];
+
+    resultValues[resultLinearIndex] = value;
+  }
+
+  return getOrCreateConstant(rewriter,
+                             rewriter.getInsertionBlock()->getParentOp(),
+                             DenseElementsAttr::get(resultType, resultValues),
+                             resultType);
+}
+
+struct TransposeToLinalgTranspose : OpConversionPattern<ONNXTransposeOp> {
+  using OpConversionPattern::OpConversionPattern;
+
+  LogicalResult matchAndRewrite(ONNXTransposeOp transposeOp,
+                                ONNXTransposeOpAdaptor adaptor,
+                                ConversionPatternRewriter& rewriter) const override {
+    auto inputType = dyn_cast<RankedTensorType>(adaptor.getData().getType());
+    auto resultType = dyn_cast<RankedTensorType>(transposeOp.getResult().getType());
+    if (!inputType || !resultType)
+      return failure();
+
+    auto permutation = getTransposePermutationChecked(transposeOp.getPermAttr(), inputType.getRank());
+    if (failed(permutation))
+      return failure();
+    if (isCompileTimeComputable(adaptor.getData())) {
+      auto constantTranspose =
+        materializeTransposedConstant(adaptor.getData(), resultType, *permutation, rewriter, transposeOp.getLoc());
+      if (succeeded(constantTranspose)) {
+        rewriter.replaceOp(transposeOp, *constantTranspose);
+        return success();
+      }
+    }
+    Value init = createTransposeInit(adaptor.getData(), resultType, *permutation, rewriter, transposeOp.getLoc());
+    Value transposed =
+      linalg::TransposeOp::create(rewriter, transposeOp.getLoc(), adaptor.getData(), init, *permutation).getResult()[0];
+    rewriter.replaceOp(transposeOp, transposed);
+    return success();
+  }
+};
+
+} // namespace
+
+void populateTransposePatterns(RewritePatternSet& patterns, MLIRContext* ctx) {
+  patterns.add<TransposeToLinalgTranspose>(ctx);
+}
+
+} // namespace onnx_mlir

src/PIM/Conversion/ONNXToSpatial/PostPatterns.cpp

@@ -1,286 +0,0 @@
-#include "mlir/Dialect/Arith/IR/Arith.h"
-#include "mlir/Dialect/Func/IR/FuncOps.h"
-#include "mlir/Dialect/Tensor/IR/Tensor.h"
-#include "mlir/IR/IRMapping.h"
-#include "mlir/IR/PatternMatch.h"
-
-#include "llvm/ADT/STLExtras.h"
-#include "llvm/ADT/SmallVector.h"
-
-#include "src/Accelerators/PIM/Common/IR/WeightUtils.hpp"
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/Common/WeightMaterialization.hpp"
-#include "src/Accelerators/PIM/Conversion/ONNXToSpatial/PostPatterns.hpp"
-#include "src/Accelerators/PIM/Dialect/Spatial/SpatialOps.hpp"
-#include "src/Dialect/ONNX/ONNXOps.hpp"
-
-using namespace mlir;
-
-namespace onnx_mlir {
-
-namespace {
-
-static bool isWeightMaterializationHelperUser(Operation* op) {
-  return isa<tensor::ExtractSliceOp, tensor::ExpandShapeOp, tensor::CollapseShapeOp, ONNXTransposeOp>(op);
-}
-
-static bool canPromoteInputBlockArgument(BlockArgument arg) {
-  return !arg.use_empty() && llvm::all_of(arg.getUsers(), isWeightMaterializationHelperUser);
-}
-
-static bool isDirectConstantValue(Value value) {
-  return isa_and_nonnull<arith::ConstantOp, ONNXConstantOp>(value.getDefiningOp());
-}
-
-template <typename ComputeOpTy>
-static bool hasPromotableWeightLikeInputs(ComputeOpTy compute) {
-  Block& block = compute.getBody().front();
-  for (auto [inputIdx, input] : llvm::enumerate(compute.getInputs())) {
-    if (inputIdx >= block.getNumArguments())
-      continue;
-    if (!isWeightLikeComputeOperand(input))
-      continue;
-    if (isDirectConstantValue(input) && !canPromoteInputBlockArgument(block.getArgument(inputIdx)))
-      continue;
-    return true;
-  }
-  return false;
-}
-
-// Collapses one-lane batches so later phases do not carry batch-only structure unnecessarily.
-struct FoldSingleLaneComputeBatchPattern : OpRewritePattern<spatial::SpatComputeBatch> {
-  using OpRewritePattern<spatial::SpatComputeBatch>::OpRewritePattern;
-
-  LogicalResult matchAndRewrite(spatial::SpatComputeBatch batchOp, PatternRewriter& rewriter) const override {
-    if (batchOp.getLaneCount() != 1)
-      return rewriter.notifyMatchFailure(batchOp, "requires a single lane");
-
-    auto loc = batchOp.getLoc();
-    rewriter.setInsertionPoint(batchOp);
-    auto computeOp =
-      spatial::SpatCompute::create(rewriter, loc, batchOp.getResultTypes(), batchOp.getWeights(), batchOp.getInputs());
-    computeOp.getProperties().setOperandSegmentSizes(
-      {static_cast<int>(batchOp.getWeights().size()), static_cast<int>(batchOp.getInputs().size())});
-
-    Block& templateBlock = batchOp.getBody().front();
-    SmallVector<Type> blockArgTypes;
-    SmallVector<Location> blockArgLocs;
-    blockArgTypes.reserve(templateBlock.getNumArguments());
-    blockArgLocs.reserve(templateBlock.getNumArguments());
-    for (BlockArgument arg : templateBlock.getArguments()) {
-      blockArgTypes.push_back(arg.getType());
-      blockArgLocs.push_back(loc);
-    }
-
-    auto* newBlock =
-      rewriter.createBlock(&computeOp.getBody(), computeOp.getBody().end(), TypeRange(blockArgTypes), blockArgLocs);
-    IRMapping mapper;
-    for (auto [oldArg, newArg] : llvm::zip(templateBlock.getArguments(), newBlock->getArguments()))
-      mapper.map(oldArg, newArg);
-
-    rewriter.setInsertionPointToEnd(newBlock);
-    for (Operation& op : templateBlock)
-      rewriter.clone(op, mapper);
-
-    batchOp->replaceAllUsesWith(computeOp->getResults());
-    rewriter.eraseOp(batchOp);
-    return success();
-  }
-};
-
-// Promotes foldable helper chains from runtime inputs to weights to avoid artificial compute inputs.
-struct PromoteWeightLikeComputeInputsPattern : OpRewritePattern<spatial::SpatCompute> {
-  using OpRewritePattern<spatial::SpatCompute>::OpRewritePattern;
-
-  LogicalResult matchAndRewrite(spatial::SpatCompute compute, PatternRewriter& rewriter) const override {
-    SmallVector<bool> promoteInput(compute.getInputs().size(), false);
-    bool needsRewrite = false;
-    Block& oldBlock = compute.getBody().front();
-    for (auto [inputIdx, input] : llvm::enumerate(compute.getInputs())) {
-      if (inputIdx >= oldBlock.getNumArguments())
-        continue;
-      if (!isWeightLikeComputeOperand(input))
-        continue;
-      if (isDirectConstantValue(input) && !canPromoteInputBlockArgument(oldBlock.getArgument(inputIdx)))
-        continue;
-      promoteInput[inputIdx] = true;
-      needsRewrite = true;
-    }
-    if (!needsRewrite)
-      return rewriter.notifyMatchFailure(compute, "no weight-like inputs to promote");
-
-    rewriter.setInsertionPointAfter(compute);
-
-    SmallVector<Value> newWeights(compute.getWeights().begin(), compute.getWeights().end());
-    SmallVector<Value> newInputs;
-    SmallVector<Type> newInputTypes;
-    SmallVector<Location> newInputLocs;
-    newWeights.reserve(compute.getWeights().size() + compute.getInputs().size());
-    newInputs.reserve(compute.getInputs().size());
-    newInputTypes.reserve(compute.getInputs().size());
-    newInputLocs.reserve(compute.getInputs().size());
-
-    for (auto [inputIdx, input] : llvm::enumerate(compute.getInputs())) {
-      if (promoteInput[inputIdx]) {
-        newWeights.push_back(input);
-        continue;
-      }
-      newInputs.push_back(input);
-      newInputTypes.push_back(input.getType());
-      newInputLocs.push_back(input.getLoc());
-    }
-
-    auto newCompute =
-      spatial::SpatCompute::create(rewriter, compute.getLoc(), compute.getResultTypes(), newWeights, newInputs);
-    auto* newBlock =
-      rewriter.createBlock(&newCompute.getBody(), newCompute.getBody().end(), newInputTypes, newInputLocs);
-    newCompute.getProperties().setOperandSegmentSizes(
-      {static_cast<int>(newWeights.size()), static_cast<int>(newInputs.size())});
-    rewriter.setInsertionPointToStart(newBlock);
-
-    IRRewriter bodyRewriter(rewriter.getContext());
-    bodyRewriter.setInsertionPointToStart(newBlock);
-
-    IRMapping mapper;
-    size_t newInputIdx = 0;
-    for (auto [oldInputIdx, oldArg] : llvm::enumerate(oldBlock.getArguments())) {
-      if (!promoteInput[oldInputIdx]) {
-        mapper.map(oldArg, newBlock->getArgument(newInputIdx++));
-        continue;
-      }
-
-      auto clonedValue = materializeWeightLikeValueInBlock(compute.getInputs()[oldInputIdx], bodyRewriter, mapper);
-      if (failed(clonedValue))
-        return rewriter.notifyMatchFailure(compute, "failed to materialize promoted weight-like operand");
-      mapper.map(oldArg, *clonedValue);
-    }
-
-    for (Operation& op : oldBlock.without_terminator())
-      rewriter.clone(op, mapper);
-
-    auto oldYield = cast<spatial::SpatYieldOp>(oldBlock.getTerminator());
-    SmallVector<Value> newYieldOperands;
-    newYieldOperands.reserve(oldYield.getOutputs().size());
-    for (Value operand : oldYield.getOutputs()) {
-      auto mapped = mapper.lookupOrNull(operand);
-      newYieldOperands.push_back(mapped ? cast<Value>(mapped) : operand);
-    }
-    spatial::SpatYieldOp::create(rewriter, oldYield.getLoc(), newYieldOperands);
-
-    rewriter.replaceOp(compute, newCompute.getResults());
-    return success();
-  }
-};
-
-// Promotes foldable batch helper chains to weights while preserving compact compute_batch IR.
-struct PromoteWeightLikeComputeBatchInputsPattern : OpRewritePattern<spatial::SpatComputeBatch> {
-  using OpRewritePattern<spatial::SpatComputeBatch>::OpRewritePattern;
-
-  LogicalResult matchAndRewrite(spatial::SpatComputeBatch compute, PatternRewriter& rewriter) const override {
-    SmallVector<bool> promoteInput(compute.getInputs().size(), false);
-    bool needsRewrite = false;
-    Block& oldBlock = compute.getBody().front();
-    for (auto [inputIdx, input] : llvm::enumerate(compute.getInputs())) {
-      if (inputIdx >= oldBlock.getNumArguments())
-        continue;
-      if (!isWeightLikeComputeOperand(input))
-        continue;
-      if (isDirectConstantValue(input) && !canPromoteInputBlockArgument(oldBlock.getArgument(inputIdx)))
-        continue;
-      promoteInput[inputIdx] = true;
-      needsRewrite = true;
-    }
-    if (!needsRewrite)
-      return rewriter.notifyMatchFailure(compute, "no weight-like batch inputs to promote");
-
-    rewriter.setInsertionPointAfter(compute);
-
-    SmallVector<Value> newWeights(compute.getWeights().begin(), compute.getWeights().end());
-    SmallVector<Value> newInputs;
-    SmallVector<Type> newInputTypes;
-    SmallVector<Location> newInputLocs;
-    newWeights.reserve(compute.getWeights().size() + compute.getInputs().size());
-    newInputs.reserve(compute.getInputs().size());
-    newInputTypes.reserve(compute.getInputs().size());
-    newInputLocs.reserve(compute.getInputs().size());
-
-    for (auto [inputIdx, input] : llvm::enumerate(compute.getInputs())) {
-      if (promoteInput[inputIdx]) {
-        newWeights.push_back(input);
-        continue;
-      }
-      newInputs.push_back(input);
-      newInputTypes.push_back(input.getType());
-      newInputLocs.push_back(input.getLoc());
-    }
-
-    auto newCompute =
-      spatial::SpatComputeBatch::create(rewriter,
-                                        compute.getLoc(),
-                                        compute.getResultTypes(),
-                                        rewriter.getI32IntegerAttr(static_cast<int32_t>(compute.getLaneCount())),
-                                        newWeights,
-                                        newInputs);
-    auto* newBlock =
-      rewriter.createBlock(&newCompute.getBody(), newCompute.getBody().end(), newInputTypes, newInputLocs);
-    newCompute.getProperties().setOperandSegmentSizes(
-      {static_cast<int>(newWeights.size()), static_cast<int>(newInputs.size())});
-    rewriter.setInsertionPointToStart(newBlock);
-
-    IRRewriter bodyRewriter(rewriter.getContext());
-    bodyRewriter.setInsertionPointToStart(newBlock);
-
-    IRMapping mapper;
-    size_t newInputIdx = 0;
-    for (auto [oldInputIdx, oldArg] : llvm::enumerate(oldBlock.getArguments())) {
-      if (!promoteInput[oldInputIdx]) {
-        mapper.map(oldArg, newBlock->getArgument(newInputIdx++));
-        continue;
-      }
-
-      auto clonedValue = materializeWeightLikeValueInBlock(compute.getInputs()[oldInputIdx], bodyRewriter, mapper);
-      if (failed(clonedValue))
-        return rewriter.notifyMatchFailure(compute, "failed to materialize promoted batch weight-like operand");
-      mapper.map(oldArg, *clonedValue);
-    }
-
-    for (Operation& op : oldBlock.without_terminator())
-      rewriter.clone(op, mapper);
-
-    auto oldYield = cast<spatial::SpatYieldOp>(oldBlock.getTerminator());
-    SmallVector<Value> newYieldOperands;
-    newYieldOperands.reserve(oldYield.getOutputs().size());
-    for (Value operand : oldYield.getOutputs()) {
-      auto mapped = mapper.lookupOrNull(operand);
-      newYieldOperands.push_back(mapped ? cast<Value>(mapped) : operand);
-    }
-    spatial::SpatYieldOp::create(rewriter, oldYield.getLoc(), newYieldOperands);
-
-    rewriter.replaceOp(compute, newCompute.getResults());
-    return success();
-  }
-};
-
-} // namespace
-
-void populateEarlyPostPatterns(RewritePatternSet& patterns, MLIRContext* ctx) {
-  patterns.add<FoldSingleLaneComputeBatchPattern>(ctx);
-}
-
-void populatePostPatterns(RewritePatternSet& patterns, MLIRContext* ctx) {
-  patterns.add<PromoteWeightLikeComputeInputsPattern, PromoteWeightLikeComputeBatchInputsPattern>(ctx);
-}
-
-void annotateWeightsConstants(func::FuncOp funcOp) {
-  funcOp.walk([&](arith::ConstantOp constantOp) {
-    if (hasOnlySpatialMvmVmmWeightUses(constantOp.getResult()))
-      markWeightAlways(constantOp);
-  });
-}
-
-bool requiresEarlyPostRewrite(spatial::SpatComputeBatch batchOp) { return batchOp.getLaneCount() == 1; }
-
-bool requiresPostRewrite(spatial::SpatCompute computeOp) { return hasPromotableWeightLikeInputs(computeOp); }
-
-bool requiresPostRewrite(spatial::SpatComputeBatch computeOp) { return hasPromotableWeightLikeInputs(computeOp); }
-
-} // namespace onnx_mlir

src/PIM/Conversion/ONNXToSpatial/PostPatterns.hpp

@@ -1,22 +0,0 @@
-#pragma once
-
-#include "mlir/Dialect/Func/IR/FuncOps.h"
-#include "mlir/IR/MLIRContext.h"
-
-#include "src/Accelerators/PIM/Dialect/Spatial/SpatialOps.hpp"
-
-namespace onnx_mlir {
-
-bool requiresEarlyPostRewrite(spatial::SpatComputeBatch batchOp);
-
-bool requiresPostRewrite(spatial::SpatCompute computeOp);
-
-bool requiresPostRewrite(spatial::SpatComputeBatch computeOp);
-
-void populateEarlyPostPatterns(mlir::RewritePatternSet& patterns, mlir::MLIRContext* ctx);
-
-void populatePostPatterns(mlir::RewritePatternSet& patterns, mlir::MLIRContext* ctx);
-
-void annotateWeightsConstants(mlir::func::FuncOp funcOp);
-
-} // namespace onnx_mlir

src/PIM/Conversion/ONNXToSpatial/PrePatterns.hpp

@@ -1,10 +0,0 @@
-#pragma once
-
-#include "mlir/IR/MLIRContext.h"
-#include "mlir/Transforms/DialectConversion.h"
-
-namespace onnx_mlir {
-
-void populatePrePatterns(mlir::RewritePatternSet& patterns, mlir::MLIRContext* ctx);
-
-} // namespace onnx_mlir

src/PIM/Conversion/SpatialToPim/BatchCoreLoweringPatterns.cpp

@@ -1,11 +1,15 @@
+#include "mlir/Dialect/Arith/IR/Arith.h"
 #include "mlir/Dialect/Bufferization/IR/Bufferization.h"
+#include "mlir/Dialect/Func/IR/FuncOps.h"
 #include "mlir/Dialect/MemRef/IR/MemRef.h"
 #include "mlir/Dialect/Tensor/IR/Tensor.h"
 #include "mlir/IR/IRMapping.h"
+#include "mlir/IR/Matchers.h"
 
 #include "Conversion/ONNXToSpatial/Common/Common.hpp"
+#include "Conversion/SpatialToPim/SpatialToPimPass.hpp"
+#include "src/Accelerators/PIM/Common/IR/BatchCoreUtils.hpp"
 #include "src/Accelerators/PIM/Common/PimCommon.hpp"
-#include "src/Accelerators/PIM/Conversion/SpatialToPim/BatchCoreLoweringPatterns.hpp"
 #include "src/Accelerators/PIM/Conversion/SpatialToPim/Common.hpp"
 #include "src/Accelerators/PIM/Dialect/Pim/PimOps.hpp"
 
@@ -15,7 +19,11 @@ using namespace onnx_mlir::pim;
 namespace onnx_mlir {
 namespace {
 
-static int32_t translateSpatialCoreIdToPimCoreId(size_t spatialCoreId) { return static_cast<int32_t>(spatialCoreId); }
+static bool isUsedOnlyAsExplicitHostOperand(Value value) {
+  return !value.use_empty() && llvm::all_of(value.getUses(), [](OpOperand& use) {
+    return isExplicitDevToHostTargetOperand(use.getOwner(), use.getOperandNumber());
+  });
+}
 
 static SmallVector<int32_t> getPimCoreIdsForBatchOp(spatial::SpatComputeBatch computeBatchOp, size_t& fallbackCoreId) {
   if (auto coreIdsAttr = computeBatchOp->getAttrOfType<DenseI32ArrayAttr>(onnx_mlir::kCoreIdsAttrName))
@@ -28,54 +36,73 @@ static SmallVector<int32_t> getPimCoreIdsForBatchOp(spatial::SpatComputeBatch co
   return coreIds;
 }
 
-static void lowerChannelSendTensorBatch(spatial::SpatChannelSendTensorBatchOp sendTensorBatchOp,
-                                        IRMapping& mapper,
-                                        IRRewriter& rewriter) {
-  SmallVector<int32_t> targetCoreIds;
-  targetCoreIds.reserve(sendTensorBatchOp.getTargetCoreIds().size());
-  for (int32_t targetCoreId : sendTensorBatchOp.getTargetCoreIds())
-    targetCoreIds.push_back(translateSpatialCoreIdToPimCoreId(targetCoreId));
+static FailureOr<unsigned> getDirectReturnOperandIndex(OpResult result) {
+  if (!result.hasOneUse())
+    return failure();
 
-  pim::PimSendTensorBatchOp::create(rewriter,
-                                    sendTensorBatchOp.getLoc(),
-                                    mapper.lookup(sendTensorBatchOp.getInput()),
-                                    rewriter.getDenseI32ArrayAttr(targetCoreIds));
+  auto returnOp = dyn_cast<func::ReturnOp>(*result.getUsers().begin());
+  if (!returnOp)
+    return failure();
+  return result.getUses().begin()->getOperandNumber();
 }
 
-static void lowerChannelReceiveTensorBatch(spatial::SpatChannelReceiveTensorBatchOp receiveTensorBatchOp,
-                                           IRMapping& mapper,
-                                           IRRewriter& rewriter) {
-  SmallVector<int32_t> sourceCoreIds;
-  sourceCoreIds.reserve(receiveTensorBatchOp.getSourceCoreIds().size());
-  for (int32_t sourceCoreId : receiveTensorBatchOp.getSourceCoreIds())
-    sourceCoreIds.push_back(translateSpatialCoreIdToPimCoreId(sourceCoreId));
+static Value createScaledIndexValue(IRRewriter& rewriter, Location loc, Value base, int64_t scale) {
+  if (scale == 1)
+    return base;
 
-  auto outputType = cast<ShapedType>(receiveTensorBatchOp.getOutput().getType());
-  auto outputBuffer = createEmptyTensorFromShaped(rewriter, receiveTensorBatchOp.getLoc(), outputType);
-  Value received = pim::PimReceiveTensorBatchOp::create(rewriter,
-                                                        receiveTensorBatchOp.getLoc(),
-                                                        outputBuffer.getType(),
-                                                        outputBuffer,
-                                                        rewriter.getDenseI32ArrayAttr(sourceCoreIds))
-                     .getOutput();
-  mapper.map(receiveTensorBatchOp.getOutput(), received);
+  auto scaleValue = getOrCreateIndexConstant(rewriter, rewriter.getInsertionBlock()->getParentOp(), scale);
+  return arith::MulIOp::create(rewriter, loc, base, scaleValue).getResult();
+}
+
+static Value createHostTargetOffset(IRRewriter& rewriter,
+                                    tensor::ParallelInsertSliceOp insertSlice,
+                                    ShapedType destinationType,
+                                    IRMapping& mapper) {
+  int64_t elementBytes = static_cast<int64_t>(getElementTypeSizeInBytes(destinationType.getElementType()));
+  SmallVector<int64_t> strides = computeRowMajorStrides(destinationType.getShape());
+
+  Value totalOffset;
+  Location loc = insertSlice.getLoc();
+  for (auto [dim, offset] : llvm::enumerate(insertSlice.getMixedOffsets())) {
+    int64_t scale = strides[dim] * elementBytes;
+    Value scaledOffset;
+    if (auto attr = dyn_cast<Attribute>(offset)) {
+      auto intAttr = dyn_cast<IntegerAttr>(attr);
+      assert(intAttr && "expected integer offset attribute");
+      scaledOffset =
+        getOrCreateIndexConstant(rewriter, rewriter.getInsertionBlock()->getParentOp(), intAttr.getInt() * scale);
+    }
+    else {
+      scaledOffset = createScaledIndexValue(rewriter, loc, mapper.lookupOrDefault(cast<Value>(offset)), scale);
+    }
+
+    totalOffset =
+      totalOffset ? arith::AddIOp::create(rewriter, loc, totalOffset, scaledOffset).getResult() : scaledOffset;
+  }
+
+  if (!totalOffset)
+    totalOffset = getOrCreateIndexConstant(rewriter, rewriter.getInsertionBlock()->getParentOp(), 0);
+  return totalOffset;
 }
 
 } // namespace
 
-LogicalResult
-lowerComputeBatchOp(spatial::SpatComputeBatch computeBatchOp, CoreLoweringState& state, IRRewriter& rewriter) {
-  if (computeBatchOp.getNumResults() != 0)
-    return computeBatchOp.emitOpError(
-      "batched Spatial-to-PIM lowering currently requires channelized compute_batch with no results");
-
+LogicalResult raptor::SpatialToPimPass::lowerComputeBatchOp(spatial::SpatComputeBatch computeBatchOp,
+                                                            IRRewriter& rewriter) {
   Location loc = computeBatchOp.getLoc();
   Block& oldBlock = computeBatchOp.getBody().front();
-  auto oldYield = cast<spatial::SpatYieldOp>(oldBlock.getTerminator());
-  if (oldYield.getNumOperands() != 0)
-    return computeBatchOp.emitOpError("batched Spatial-to-PIM lowering currently requires empty spat.yield");
+  auto oldYield = dyn_cast<spatial::SpatYieldOp>(oldBlock.getTerminator());
+  auto inParallelOp = dyn_cast<spatial::SpatInParallelOp>(oldBlock.getTerminator());
+  if (computeBatchOp.getNumResults() == 0) {
+    if (!oldYield || oldYield.getNumOperands() != 0)
+      return computeBatchOp.emitOpError("resultless compute_batch lowering requires empty spat.yield");
+  }
+  else if (!inParallelOp) {
+    return computeBatchOp.emitOpError(
+      "resultful compute_batch lowering currently requires a spat.in_parallel terminator");
+  }
 
-  SmallVector<int32_t> coreIds = getPimCoreIdsForBatchOp(computeBatchOp, state.nextCoreId);
+  SmallVector<int32_t> coreIds = getPimCoreIdsForBatchOp(computeBatchOp, coreId);
   SmallVector<Value> batchWeights(computeBatchOp.getWeights().begin(), computeBatchOp.getWeights().end());
   SmallVector<Value> batchInputs;
   if (!computeBatchOp.getInputs().empty())
@@ -91,9 +118,22 @@ lowerComputeBatchOp(spatial::SpatComputeBatch computeBatchOp, CoreLoweringState&
     {static_cast<int>(batchWeights.size()), static_cast<int>(batchInputs.size())});
   coreBatchOp->setAttr(onnx_mlir::kCoreIdsAttrName, rewriter.getDenseI32ArrayAttr(coreIds));
 
+  SmallVector<unsigned> returnOperandIndices;
+  if (computeBatchOp.getNumResults() != 0) {
+    returnOperandIndices.resize(computeBatchOp.getNumResults());
+    for (auto [resultIndex, result] : llvm::enumerate(computeBatchOp.getResults())) {
+      FailureOr<unsigned> returnOperandIndex = getDirectReturnOperandIndex(cast<OpResult>(result));
+      if (failed(returnOperandIndex))
+        return computeBatchOp.emitOpError(
+          "resultful compute_batch lowering currently requires each result to be used directly by func.return");
+      returnOperandIndices[resultIndex] = *returnOperandIndex;
+    }
+  }
+
   SmallVector<Type> blockArgTypes;
   SmallVector<Location> blockArgLocs;
-  for (BlockArgument arg : oldBlock.getArguments()) {
+  unsigned inputArgLimit = 1 + computeBatchOp.getWeights().size() + computeBatchOp.getInputs().size();
+  for (BlockArgument arg : oldBlock.getArguments().take_front(inputArgLimit)) {
     blockArgTypes.push_back(arg.getType());
     blockArgLocs.push_back(arg.getLoc());
   }
@@ -102,38 +142,44 @@ lowerComputeBatchOp(spatial::SpatComputeBatch computeBatchOp, CoreLoweringState&
 
   IRMapping mapper;
   rewriter.setInsertionPointToStart(newBlock);
-  for (auto [oldArg, newArg] : llvm::zip(oldBlock.getArguments(), newBlock->getArguments())) {
+  auto oldLaneArg = computeBatchOp.getLaneArgument();
+  if (!oldLaneArg)
+    return computeBatchOp.emitOpError("expected compute_batch lane block argument before lowering");
+  mapper.map(*oldLaneArg, coreBatchOp.getLaneArgument());
+  for (unsigned weightIndex = 0; weightIndex < computeBatchOp.getWeights().size(); ++weightIndex) {
+    auto oldWeightArg = computeBatchOp.getWeightArgument(weightIndex);
+    if (!oldWeightArg)
+      return computeBatchOp.emitOpError("expected compute_batch weight block arguments before lowering");
+    mapper.map(*oldWeightArg, coreBatchOp.getWeightArgument(weightIndex));
+  }
+  for (unsigned inputIndex = 0; inputIndex < computeBatchOp.getInputs().size(); ++inputIndex) {
+    auto oldArg = computeBatchOp.getInputArgument(inputIndex);
+    if (!oldArg)
+      return computeBatchOp.emitOpError("expected compute_batch input block arguments before lowering");
+    BlockArgument newArg = coreBatchOp.getInputArgument(inputIndex);
     auto newArgType = cast<ShapedType>(newArg.getType());
     auto outputBuffer = createEmptyTensorFromShaped(rewriter, loc, newArgType);
-    auto copied = pim::PimMemCopyHostToDevBatchOp::create(rewriter,
+    Value zeroOffset = getOrCreateIndexConstant(rewriter, coreBatchOp.getOperation(), 0);
+    auto copied = pim::PimMemCopyHostToDevOp::create(rewriter,
                                                      loc,
                                                      outputBuffer.getType(),
+                                                     zeroOffset,
+                                                     zeroOffset,
                                                      outputBuffer,
                                                      newArg,
-                                                          rewriter.getI32IntegerAttr(0),
-                                                          rewriter.getI32IntegerAttr(0),
                                                      getTensorSizeInBytesAttr(rewriter, newArg))
                     .getOutput();
-    mapper.map(oldArg, copied);
+    mapper.map(*oldArg, copied);
   }
 
-  auto materializeCapturedTensor = [&](Value capturedTensor) -> Value {
-    if (auto mapped = mapper.lookupOrNull(capturedTensor))
-      return mapped;
+  SmallVector<Value> hostOutputTensors(returnOperandIndices.size());
+  auto getOrCreateHostOutputTensor = [&](unsigned resultIndex, Location resultLoc) -> Value {
+    Value& hostOutputTensor = hostOutputTensors[resultIndex];
+    if (hostOutputTensor)
+      return hostOutputTensor;
 
-    auto capturedType = cast<ShapedType>(capturedTensor.getType());
-    auto outputBuffer = createEmptyTensorFromShaped(rewriter, loc, capturedType);
-    auto copied = pim::PimMemCopyHostToDevBatchOp::create(rewriter,
-                                                          loc,
-                                                          outputBuffer.getType(),
-                                                          outputBuffer,
-                                                          capturedTensor,
-                                                          rewriter.getI32IntegerAttr(0),
-                                                          rewriter.getI32IntegerAttr(0),
-                                                          getTensorSizeInBytesAttr(rewriter, capturedTensor))
-                    .getOutput();
-    mapper.map(capturedTensor, copied);
-    return copied;
+    hostOutputTensor = outputTensors[returnOperandIndices[resultIndex]](rewriter, resultLoc);
+    return hostOutputTensor;
   };
 
   rewriter.setInsertionPointToEnd(newBlock);
@@ -141,36 +187,37 @@ lowerComputeBatchOp(spatial::SpatComputeBatch computeBatchOp, CoreLoweringState&
     if (isa<spatial::SpatYieldOp>(op))
       continue;
 
-    if (auto sendBatchOp = dyn_cast<spatial::SpatChannelSendBatchOp>(op)) {
-      pim::PimSendBatchOp::create(rewriter,
-                                  loc,
-                                  mapper.lookup(sendBatchOp.getInput()),
-                                  getTensorSizeInBytesAttr(rewriter, mapper.lookup(sendBatchOp.getInput())),
-                                  sendBatchOp.getTargetCoreIdsAttr());
-      continue;
-    }
+    if (auto parallelOp = dyn_cast<spatial::SpatInParallelOp>(op)) {
+      auto firstOutputArg = computeBatchOp.getOutputArgument(0);
+      if (!firstOutputArg)
+        return computeBatchOp.emitOpError("expected compute_batch output block arguments before lowering");
+      for (Operation& nestedOp : parallelOp.getRegion().front()) {
+        auto insertSlice = dyn_cast<tensor::ParallelInsertSliceOp>(&nestedOp);
+        if (!insertSlice)
+          return parallelOp.emitOpError("expected only tensor.parallel_insert_slice in spat.in_parallel");
 
-    if (auto sendTensorBatchOp = dyn_cast<spatial::SpatChannelSendTensorBatchOp>(op)) {
-      lowerChannelSendTensorBatch(sendTensorBatchOp, mapper, rewriter);
-      continue;
-    }
+        auto outputArg = dyn_cast<BlockArgument>(insertSlice.getDest());
+        if (!outputArg || outputArg.getOwner() != &oldBlock)
+          return insertSlice.emitOpError("expected compute_batch output block argument destination");
 
-    if (auto receiveBatchOp = dyn_cast<spatial::SpatChannelReceiveBatchOp>(op)) {
-      auto outputType = cast<ShapedType>(receiveBatchOp.getOutput().getType());
-      auto outputBuffer = createEmptyTensorFromShaped(rewriter, loc, outputType);
-      auto received = pim::PimReceiveBatchOp::create(rewriter,
-                                                     loc,
-                                                     outputBuffer.getType(),
-                                                     outputBuffer,
-                                                     getTensorSizeInBytesAttr(rewriter, receiveBatchOp.getOutput()),
-                                                     receiveBatchOp.getSourceCoreIdsAttr())
-                        .getOutput();
-      mapper.map(receiveBatchOp.getOutput(), received);
-      continue;
-    }
+        unsigned resultIndex = outputArg.getArgNumber() - firstOutputArg->getArgNumber();
+        if (resultIndex >= returnOperandIndices.size())
+          return insertSlice.emitOpError("result index out of range while lowering host batch output");
 
-    if (auto receiveTensorBatchOp = dyn_cast<spatial::SpatChannelReceiveTensorBatchOp>(op)) {
-      lowerChannelReceiveTensorBatch(receiveTensorBatchOp, mapper, rewriter);
+        Value mappedSource = mapper.lookup(insertSlice.getSource());
+        Value hostTarget = getOrCreateHostOutputTensor(resultIndex, insertSlice.getLoc());
+        auto hostTargetType = cast<ShapedType>(hostTarget.getType());
+        Value hostTargetOffset = createHostTargetOffset(rewriter, insertSlice, hostTargetType, mapper);
+        Value zeroOffset = getOrCreateIndexConstant(rewriter, coreBatchOp.getOperation(), 0);
+        pim::PimMemCopyDevToHostOp::create(rewriter,
+                                           insertSlice.getLoc(),
+                                           hostTarget.getType(),
+                                           hostTargetOffset,
+                                           zeroOffset,
+                                           hostTarget,
+                                           mappedSource,
+                                           getTensorSizeInBytesAttr(rewriter, mappedSource));
+      }
       continue;
     }
 
@@ -178,15 +225,20 @@ lowerComputeBatchOp(spatial::SpatComputeBatch computeBatchOp, CoreLoweringState&
       if (isa_and_present<memref::GetGlobalOp>(toTensorOp.getBuffer().getDefiningOp())) {
         Operation* cloned = rewriter.clone(op, mapper);
         auto clonedTensor = cloned->getResult(0);
+        if (isUsedOnlyAsExplicitHostOperand(toTensorOp.getResult())) {
+          mapper.map(toTensorOp.getResult(), clonedTensor);
+          continue;
+        }
         auto clonedType = cast<ShapedType>(clonedTensor.getType());
         auto outputBuffer = createEmptyTensorFromShaped(rewriter, loc, clonedType);
-        auto copied = pim::PimMemCopyHostToDevBatchOp::create(rewriter,
+        Value zeroOffset = getOrCreateIndexConstant(rewriter, coreBatchOp.getOperation(), 0);
+        auto copied = pim::PimMemCopyHostToDevOp::create(rewriter,
                                                          loc,
                                                          outputBuffer.getType(),
+                                                         zeroOffset,
+                                                         zeroOffset,
                                                          outputBuffer,
                                                          clonedTensor,
-                                                              rewriter.getI32IntegerAttr(0),
-                                                              rewriter.getI32IntegerAttr(0),
                                                          getTensorSizeInBytesAttr(rewriter, clonedTensor))
                         .getOutput();
         mapper.map(toTensorOp.getResult(), copied);
@@ -194,15 +246,18 @@ lowerComputeBatchOp(spatial::SpatComputeBatch computeBatchOp, CoreLoweringState&
       }
     }
 
-    for (Value operand : op.getOperands()) {
+    for (auto [operandIndex, operand] : llvm::enumerate(op.getOperands())) {
       if (!isa<TensorType>(operand.getType()) || mapper.contains(operand))
         continue;
+      if (isExplicitDevToHostTargetOperand(&op, operandIndex))
+        continue;
 
       Operation* definingOp = operand.getDefiningOp();
       if (definingOp && definingOp->getBlock() == &oldBlock)
         continue;
 
-      materializeCapturedTensor(operand);
+      return computeBatchOp.emitOpError(
+        "expected external tensor communication to be materialized in Spatial before batch lowering");
     }
 
     Operation* cloned = rewriter.clone(op, mapper);

src/PIM/Conversion/SpatialToPim/BatchCoreLoweringPatterns.hpp

@@ -1,10 +0,0 @@
-#pragma once
-
-#include "src/Accelerators/PIM/Conversion/SpatialToPim/CoreLoweringPatterns.hpp"
-
-namespace onnx_mlir {
-
-mlir::LogicalResult
-lowerComputeBatchOp(spatial::SpatComputeBatch computeBatchOp, CoreLoweringState& state, mlir::IRRewriter& rewriter);
-
-} // namespace onnx_mlir

src/PIM/Conversion/SpatialToPim/CMakeLists.txt

@@ -3,17 +3,17 @@ mlir_tablegen(SpatialToPim.hpp.inc -gen-rewriters "-I${ONNX_MLIR_SRC_ROOT}")
 add_public_tablegen_target(SpatialToPimIncGen)
 
 add_pim_library(OMSpatialToPim
+  Patterns.cpp
   SpatialToPimPass.cpp
   BatchCoreLoweringPatterns.cpp
-  ChannelLoweringPatterns.cpp
-  Cleanup.cpp
   Common.cpp
   ComputeLikeRegionUtils.cpp
   CoreLoweringPatterns.cpp
-  GlobalTensorMaterialization.cpp
-  PhaseVerification.cpp
   ReturnPathNormalization.cpp
-  TensorPackingPatterns.cpp
+  Patterns/ChannelLowering.cpp
+  Patterns/GlobalTensorMaterialization.cpp
+  Patterns/TensorPacking.cpp
+  Patterns/Transpose.cpp
 
   EXCLUDE_FROM_OM_LIBS
 
@@ -21,7 +21,10 @@ add_pim_library(OMSpatialToPim
   SpatialToPimIncGen
 
   LINK_LIBS PUBLIC
+  MLIRLinalgDialect
   MLIRSCFDialect
+  MLIRSCFUtils
+  MLIRTransformUtils
   MLIRTosaDialect
   OMCompilerOptions
   OMPimCommon

src/PIM/Conversion/SpatialToPim/ChannelLoweringPatterns.hpp

@@ -1,9 +0,0 @@
-#pragma once
-
-#include "mlir/IR/PatternMatch.h"
-
-namespace onnx_mlir {
-
-void populateChannelLoweringPatterns(mlir::RewritePatternSet& patterns);
-
-} // namespace onnx_mlir

src/PIM/Conversion/SpatialToPim/Cleanup.cpp

@@ -1,42 +0,0 @@
-#include "llvm/ADT/STLExtras.h"
-
-#include "src/Accelerators/PIM/Conversion/SpatialToPim/Cleanup.hpp"
-
-using namespace mlir;
-
-namespace onnx_mlir {
-
-LogicalResult erasePendingOps(SmallVectorImpl<Operation*>& pendingOps, IRRewriter& rewriter) {
-  while (!pendingOps.empty()) {
-    bool erasedAnyOp = false;
-    for (auto it = pendingOps.begin(); it != pendingOps.end();) {
-      Operation* opToRemove = *it;
-      if (!opToRemove->use_empty()) {
-        ++it;
-        continue;
-      }
-
-      rewriter.eraseOp(opToRemove);
-      it = pendingOps.erase(it);
-      erasedAnyOp = true;
-    }
-
-    if (erasedAnyOp)
-      continue;
-
-    for (Operation* opToRemove : pendingOps) {
-      InFlightDiagnostic diag = opToRemove->emitError("pending Spatial-to-PIM cleanup could not erase operation");
-      diag << "; op has " << llvm::range_size(opToRemove->getUsers()) << " remaining user(s)";
-      for (Operation* user : opToRemove->getUsers()) {
-        bool userPendingRemoval = llvm::is_contained(pendingOps, user);
-        opToRemove->emitRemark() << "remaining user `" << user->getName() << "`"
-                                 << (userPendingRemoval ? " is also pending removal" : " is not pending removal");
-      }
-    }
-    return failure();
-  }
-
-  return success();
-}
-
-} // namespace onnx_mlir

src/PIM/Conversion/SpatialToPim/Cleanup.hpp

@@ -1,11 +0,0 @@
-#pragma once
-
-#include "mlir/IR/Operation.h"
-#include "mlir/IR/PatternMatch.h"
-#include "mlir/Support/LLVM.h"
-
-namespace onnx_mlir {
-
-mlir::LogicalResult erasePendingOps(llvm::SmallVectorImpl<mlir::Operation*>& pendingOps, mlir::IRRewriter& rewriter);
-
-} // namespace onnx_mlir

src/PIM/Conversion/SpatialToPim/Common.cpp

@@ -1,10 +1,8 @@
 #include "mlir/IR/ValueRange.h"
 
 #include "llvm/ADT/STLExtras.h"
-#include "llvm/ADT/StringRef.h"
 
 #include <cassert>
-#include <cstddef>
 
 #include "Common.hpp"
 
@@ -13,52 +11,6 @@ using namespace mlir;
 
 namespace onnx_mlir {
 
-size_t getSliceActualOffset(tensor::ExtractSliceOp& sliceOp, ShapedType& inputShape) {
-  /*
-    EXAMPLE RUN:
-    [1, 10, 3, 4] inputShape
-    [0, 2, 1, 3] offsets
-
-    acc = 1
-    ---
-    ret = 3
-    acc = 4
-    ---
-    ret = 3 + 4 * 1 = 7
-    acc = 12
-    ---
-    ret = 7 + 12 * 2 = 31
-    acc = 120
-    ---
-    ret = 31 + 120 * 0 = 31
-    acc = 120
-  */
-
-  size_t returnValue = 0;
-
-  auto sliceOffsets = sliceOp.getStaticOffsets();
-  auto inputDimSizes = inputShape.getShape();
-
-  assert(sliceOffsets.size() == inputDimSizes.size());
-
-  size_t accumulatedDimensionSize = 1;
-
-  // Reverse iterate the two vectors
-  for (auto it : reverse(zip(sliceOffsets, inputDimSizes))) {
-    auto curSliceOffset = std::get<0>(it);
-    auto curInputDimSize = std::get<1>(it);
-
-    returnValue += accumulatedDimensionSize * curSliceOffset;
-    accumulatedDimensionSize *= curInputDimSize;
-  }
-
-  return returnValue;
-}
-
-size_t getShapedTypeSizeInBytes(ShapedType shapedType) {
-  return shapedType.getNumElements() * shapedType.getElementTypeBitWidth() / 8;
-}
-
 IntegerAttr getTensorSizeInBytesAttr(Builder& builder, mlir::Value value) {
   return builder.getI32IntegerAttr(static_cast<int32_t>(getShapedTypeSizeInBytes(cast<ShapedType>(value.getType()))));
 }

src/PIM/Conversion/SpatialToPim/Common.hpp

@@ -6,22 +6,6 @@
 
 namespace onnx_mlir {
 
-/**
- * \brief Get the offset of the ExtractSliceOp based on its static offsets and
- * its static tensor input.
- *
- * The static offsets represent the starting position of the slice in each
- * dimension, while the static tensor input gives its dimension size.
- *
- * \param sliceOp The ExtractSliceOp for which the actual offset needs to be
- * calculated.
- * \param inputShape The ShapedType of the ExtractSliceOp's input tensor
- * \return The actual offset of the ExtractSliceOp.
- */
-size_t getSliceActualOffset(mlir::tensor::ExtractSliceOp& sliceOp, mlir::ShapedType& inputShape);
-
-size_t getShapedTypeSizeInBytes(mlir::ShapedType shapedType);
-
 mlir::IntegerAttr getTensorSizeInBytesAttr(mlir::Builder& builder, mlir::Value value);
 
 template <class T>

src/PIM/Conversion/SpatialToPim/ComputeLikeRegionUtils.cpp

@@ -1,3 +1,5 @@
+#include <cassert>
+
 #include "src/Accelerators/PIM/Conversion/SpatialToPim/ComputeLikeRegionUtils.hpp"
 #include "src/Accelerators/PIM/Dialect/Spatial/SpatialOps.hpp"
 
@@ -29,7 +31,18 @@ void replaceAndEraseDirectComputeLikeInput(PatternRewriter& rewriter,
                                            unsigned inputIndex,
                                            Value replacement) {
   Block& body = owner->getRegion(0).front();
-  BlockArgument bodyArgument = body.getArgument(inputIndex);
+  BlockArgument bodyArgument;
+  if (auto compute = dyn_cast<spatial::SpatCompute>(owner)) {
+    auto computeArg = compute.getInputArgument(inputIndex);
+    assert(computeArg && "expected compute input block argument");
+    bodyArgument = *computeArg;
+  }
+  else {
+    auto batchArg = cast<spatial::SpatComputeBatch>(owner).getInputArgument(inputIndex);
+    assert(batchArg && "expected compute_batch input block argument");
+    bodyArgument = *batchArg;
+  }
+  unsigned bodyArgIndex = bodyArgument.getArgNumber();
 
   rewriter.startOpModification(owner);
   bodyArgument.replaceAllUsesWith(replacement);
@@ -37,7 +50,7 @@ void replaceAndEraseDirectComputeLikeInput(PatternRewriter& rewriter,
     compute.getInputsMutable().erase(inputIndex);
   else
     cast<spatial::SpatComputeBatch>(owner).getInputsMutable().erase(inputIndex);
-  body.eraseArgument(inputIndex);
+  body.eraseArgument(bodyArgIndex);
   rewriter.finalizeOpModification(owner);
 }
 

src/PIM/Conversion/SpatialToPim/CoreLoweringPatterns.cpp

@@ -1,13 +1,15 @@
 #include "mlir/Dialect/Arith/IR/Arith.h"
 #include "mlir/Dialect/Func/IR/FuncOps.h"
+#include "mlir/Dialect/Linalg/IR/Linalg.h"
 #include "mlir/Dialect/Tensor/IR/Tensor.h"
 #include "mlir/Dialect/Tosa/IR/TosaOps.h"
 #include "mlir/IR/IRMapping.h"
+#include "mlir/IR/Matchers.h"
 
 #include "Conversion/ONNXToSpatial/Common/Common.hpp"
+#include "Conversion/SpatialToPim/SpatialToPimPass.hpp"
 #include "src/Accelerators/PIM/Common/PimCommon.hpp"
 #include "src/Accelerators/PIM/Conversion/SpatialToPim/Common.hpp"
-#include "src/Accelerators/PIM/Conversion/SpatialToPim/CoreLoweringPatterns.hpp"
 #include "src/Accelerators/PIM/Dialect/Pim/PimOps.hpp"
 #include "src/Dialect/ONNX/ONNXOps.hpp"
 
@@ -23,11 +25,12 @@ static bool isChannelUseChainOp(Operation* op) {
              tensor::ExpandShapeOp,
              tensor::CastOp,
              tosa::ReshapeOp,
-             ONNXTransposeOp,
+             linalg::TransposeOp,
              pim::PimTransposeOp>(op);
 }
 
-static void cloneMappedHelperOperands(Operation* op, IRMapping& mapping, IRRewriter& rewriter) {
+static void
+cloneMappedHelperOperands(Operation* op, IRMapping& mapping, IRRewriter& rewriter, OperationFolder& constantFolder) {
   for (Value operand : op->getOperands()) {
     if (mapping.lookupOrNull(operand))
       continue;
@@ -36,7 +39,12 @@ static void cloneMappedHelperOperands(Operation* op, IRMapping& mapping, IRRewri
     if (!definingOp)
       continue;
 
-    if (!isa<tensor::EmptyOp, arith::ConstantOp>(definingOp))
+    if (auto constantOp = dyn_cast<arith::ConstantOp>(definingOp)) {
+      mapping.map(operand, getOrCreateConstantLike(constantFolder, constantOp));
+      continue;
+    }
+
+    if (!isa<tensor::EmptyOp>(definingOp))
       continue;
 
     Operation* clonedOp = rewriter.clone(*definingOp, mapping);
@@ -46,8 +54,6 @@ static void cloneMappedHelperOperands(Operation* op, IRMapping& mapping, IRRewri
   }
 }
 
-static int32_t translateSpatialCoreIdToPimCoreId(size_t spatialCoreId) { return static_cast<int32_t>(spatialCoreId); }
-
 static int32_t getPimCoreIdForComputeOp(spatial::SpatCompute computeOp, size_t& fallbackCoreId) {
   if (auto spatialCoreIdAttr = computeOp->getAttrOfType<IntegerAttr>(onnx_mlir::kCoreIdAttrName))
     return static_cast<int32_t>(spatialCoreIdAttr.getInt());
@@ -92,7 +98,9 @@ static LogicalResult collectHelperComputeChain(spatial::SpatCompute computeOp,
   return success();
 }
 
-static bool inlineInputlessHelperComputeForWeightLikeUsers(spatial::SpatCompute computeOp, IRRewriter& rewriter) {
+static bool inlineInputlessHelperComputeForWeightLikeUsers(spatial::SpatCompute computeOp,
+                                                           IRRewriter& rewriter,
+                                                           OperationFolder& constantFolder) {
   if (!computeOp.getInputs().empty() || computeOp.getNumResults() != 1)
     return false;
   if (!llvm::all_of(computeOp.getResult(0).getUsers(), [](Operation* user) {
@@ -101,7 +109,7 @@ static bool inlineInputlessHelperComputeForWeightLikeUsers(spatial::SpatCompute
     return false;
 
   Block& block = computeOp.getBody().front();
-  if (block.getNumArguments() != 0)
+  if (block.getNumArguments() != computeOp.getWeights().size())
     return false;
 
   auto yieldOp = dyn_cast<spatial::SpatYieldOp>(block.getTerminator());
@@ -110,8 +118,14 @@ static bool inlineInputlessHelperComputeForWeightLikeUsers(spatial::SpatCompute
 
   rewriter.setInsertionPoint(computeOp);
   IRMapping mapping;
+  for (auto [weightIndex, weight] : llvm::enumerate(computeOp.getWeights())) {
+    auto weightArg = computeOp.getWeightArgument(weightIndex);
+    if (!weightArg)
+      return false;
+    mapping.map(*weightArg, weight);
+  }
   for (Operation& op : block.without_terminator()) {
-    cloneMappedHelperOperands(&op, mapping, rewriter);
+    cloneMappedHelperOperands(&op, mapping, rewriter, constantFolder);
     Operation* clonedOp = rewriter.clone(op, mapping);
     for (auto [originalResult, newResult] : llvm::zip(op.getResults(), clonedOp->getResults()))
       mapping.map(originalResult, newResult);
@@ -125,15 +139,12 @@ static bool inlineInputlessHelperComputeForWeightLikeUsers(spatial::SpatCompute
 
 } // namespace
 
-void markOpToRemove(CoreLoweringState& state, Operation* op) {
-  if (!llvm::is_contained(state.operationsToRemove, op))
-    state.operationsToRemove.push_back(op);
-}
-
-LogicalResult lowerComputeOp(spatial::SpatCompute computeOp, CoreLoweringState& state, IRRewriter& rewriter) {
+LogicalResult raptor::SpatialToPimPass::lowerComputeOp(spatial::SpatCompute computeOp,
+                                                       IRRewriter& rewriter,
+                                                       OperationFolder& constantFolder) {
   Location loc = computeOp->getLoc();
 
-  if (inlineInputlessHelperComputeForWeightLikeUsers(computeOp, rewriter))
+  if (inlineInputlessHelperComputeForWeightLikeUsers(computeOp, rewriter, constantFolder))
     return success();
 
   SmallVector<Operation*> helperChain;
@@ -143,21 +154,24 @@ LogicalResult lowerComputeOp(spatial::SpatCompute computeOp, CoreLoweringState&
   auto& block = computeOp.getRegion().front();
   auto yieldOp = cast<spatial::SpatYieldOp>(block.getTerminator());
 
-  for (auto [argIndex, blockArg] : llvm::enumerate(block.getArguments())) {
-    auto receiveOp = dyn_cast_or_null<spatial::SpatChannelReceiveOp>(computeOp.getInputs()[argIndex].getDefiningOp());
-    if (!receiveOp || blockArg.use_empty())
-      continue;
-
-    rewriter.setInsertionPoint(getEarliestUserWithinBlock(blockArg));
-    auto outputType = cast<ShapedType>(blockArg.getType());
+  for (auto [inputIndex, input] : llvm::enumerate(computeOp.getInputs())) {
+    auto blockArg = computeOp.getInputArgument(inputIndex);
+    if (!blockArg)
+      return computeOp.emitOpError("expected compute input block arguments during lowering");
+    auto receiveOp = dyn_cast_or_null<spatial::SpatChannelReceiveOp>(input.getDefiningOp());
+    if (receiveOp && !blockArg->use_empty()) {
+      rewriter.setInsertionPoint(getEarliestUserWithinBlock(*blockArg));
+      auto outputType = cast<ShapedType>(blockArg->getType());
       auto outputBuffer = createEmptyTensorFromShaped(rewriter, receiveOp.getLoc(), outputType);
-    auto sizeAttr = getTensorSizeInBytesAttr(rewriter, blockArg);
-    auto sourceCoreIdAttr = rewriter.getI32IntegerAttr(translateSpatialCoreIdToPimCoreId(receiveOp.getSourceCoreId()));
-    Value received = PimReceiveOp::create(
-                       rewriter, receiveOp.getLoc(), outputBuffer.getType(), outputBuffer, sizeAttr, sourceCoreIdAttr)
+      auto sizeAttr = getTensorSizeInBytesAttr(rewriter, *blockArg);
+      Value received =
+        PimReceiveOp::create(
+          rewriter, receiveOp.getLoc(), outputBuffer.getType(), outputBuffer, sizeAttr, receiveOp.getSourceCoreId())
           .getOutput();
-    blockArg.replaceAllUsesWith(received);
-    markOpToRemove(state, receiveOp);
+      blockArg->replaceAllUsesWith(received);
+      markOpToRemove(receiveOp);
+      continue;
+    }
   }
 
   if (computeOp.getNumResults() != yieldOp.getNumOperands())
@@ -167,9 +181,8 @@ LogicalResult lowerComputeOp(spatial::SpatCompute computeOp, CoreLoweringState&
     if (result.use_empty())
       continue;
 
-    ReturnPathState returnPathState {state.outputTensors, state.operationsToRemove};
     ReturnPathLoweringResult returnPathResult =
-      lowerComputeResultReturnPath(computeOp, cast<OpResult>(result), yieldValue, returnPathState, rewriter);
+      lowerComputeResultReturnPath(computeOp, cast<OpResult>(result), yieldValue, rewriter);
     if (returnPathResult == ReturnPathLoweringResult::Failure)
       return failure();
     if (returnPathResult == ReturnPathLoweringResult::Handled)
@@ -193,15 +206,40 @@ LogicalResult lowerComputeOp(spatial::SpatCompute computeOp, CoreLoweringState&
   if (!computeOp.getWeights().empty())
     computeWeights.append(computeOp.getWeights().begin(), computeOp.getWeights().end());
   rewriter.setInsertionPointAfter(computeOp);
-  auto coreOp = PimCoreOp::create(rewriter,
-                                  loc,
-                                  ValueRange(computeWeights),
-                                  rewriter.getI32IntegerAttr(getPimCoreIdForComputeOp(computeOp, state.nextCoreId)));
+  auto coreOp = PimCoreOp::create(
+    rewriter, loc, ValueRange(computeWeights), rewriter.getI32IntegerAttr(getPimCoreIdForComputeOp(computeOp, coreId)));
+  rewriter.setInsertionPointToStart(&block);
   auto& coreOpBlocks = coreOp.getBody().getBlocks();
-  for (auto [argIndex, blockArg] : llvm::enumerate(block.getArguments()))
-    if (!blockArg.use_empty())
-      blockArg.replaceAllUsesWith(computeOp.getInputs()[argIndex]);
-  block.eraseArguments(0, block.getNumArguments());
+  for (auto [inputIndex, input] : llvm::enumerate(computeOp.getInputs())) {
+    auto blockArg = computeOp.getInputArgument(inputIndex);
+    if (!blockArg)
+      return computeOp.emitOpError("expected compute input block arguments during input materialization");
+    if (blockArg->use_empty())
+      continue;
+
+    if (auto constantOp = input.getDefiningOp<arith::ConstantOp>()) {
+      blockArg->replaceAllUsesWith(getOrCreateConstantLike(constantFolder, constantOp));
+      continue;
+    }
+
+    auto inputType = dyn_cast<ShapedType>(input.getType());
+    if (!inputType)
+      return computeOp.emitOpError("expected shaped compute input during pim.core lowering");
+    auto outputBuffer = createEmptyTensorFromShaped(rewriter, loc, inputType);
+    auto copied =
+      PimMemCopyHostToDevOp::create(rewriter,
+                                    loc,
+                                    outputBuffer.getType(),
+                                    getOrCreateIndexConstant(constantFolder, outputBuffer.getOperation(), 0),
+                                    getOrCreateIndexConstant(constantFolder, outputBuffer.getOperation(), 0),
+                                    outputBuffer,
+                                    input,
+                                    getTensorSizeInBytesAttr(rewriter, input))
+        .getOutput();
+    blockArg->replaceAllUsesWith(copied);
+  }
+  if (!computeOp.getInputs().empty())
+    block.eraseArguments(computeOp.getWeights().size(), computeOp.getInputs().size());
   coreOpBlocks.splice(coreOpBlocks.begin(), computeOp.getBody().getBlocks());
   Block* tempComputeBlock = new Block();
   computeOp.getBody().push_back(tempComputeBlock);

src/PIM/Conversion/SpatialToPim/CoreLoweringPatterns.hpp

@@ -1,21 +0,0 @@
-#pragma once
-
-#include "mlir/IR/PatternMatch.h"
-
-#include "src/Accelerators/PIM/Conversion/SpatialToPim/ReturnPathNormalization.hpp"
-#include "src/Accelerators/PIM/Dialect/Spatial/SpatialOps.hpp"
-
-namespace onnx_mlir {
-
-struct CoreLoweringState {
-  size_t& nextCoreId;
-  llvm::SmallVectorImpl<OutputTensorFactory>& outputTensors;
-  llvm::SmallVectorImpl<mlir::Operation*>& operationsToRemove;
-};
-
-void markOpToRemove(CoreLoweringState& state, mlir::Operation* op);
-
-mlir::LogicalResult
-lowerComputeOp(spatial::SpatCompute computeOp, CoreLoweringState& state, mlir::IRRewriter& rewriter);
-
-} // namespace onnx_mlir

src/PIM/Conversion/SpatialToPim/GlobalTensorMaterialization.hpp

@@ -1,9 +0,0 @@
-#pragma once
-
-#include "mlir/IR/PatternMatch.h"
-
-namespace onnx_mlir {
-
-void populateGlobalTensorMaterializationPatterns(mlir::RewritePatternSet& patterns);
-
-}

src/PIM/Conversion/SpatialToPim/Patterns.cpp

@@ -0,0 +1,24 @@
+#include "src/Accelerators/PIM/Conversion/SpatialToPim/Common.hpp"
+#include "src/Accelerators/PIM/Conversion/SpatialToPim/Patterns.hpp"
+#include "src/Accelerators/PIM/Dialect/Pim/PimOps.hpp"
+
+using namespace mlir;
+
+namespace onnx_mlir {
+namespace raptor {
+
+#include "src/Accelerators/PIM/Conversion/SpatialToPim/SpatialToPim.hpp.inc"
+
+} // namespace raptor
+
+void populateInitialPatterns(RewritePatternSet& patterns) {
+  raptor::populateWithGenerated(patterns);
+  populateTransposeLoweringPatterns(patterns);
+}
+
+void populateCoreBodyPatterns(RewritePatternSet& patterns) {
+  raptor::populateWithGenerated(patterns);
+  populateTransposeLoweringPatterns(patterns);
+}
+
+} // namespace onnx_mlir

src/PIM/Conversion/SpatialToPim/Patterns.hpp

@@ -8,6 +8,14 @@
 
 namespace onnx_mlir {
 
+void populateInitialPatterns(mlir::RewritePatternSet& patterns);
+void populateCoreBodyPatterns(mlir::RewritePatternSet& patterns);
+
+void populateTransposeLoweringPatterns(mlir::RewritePatternSet& patterns);
+void populateChannelLoweringPatterns(mlir::RewritePatternSet& patterns);
+void populateGlobalTensorMaterializationPatterns(mlir::RewritePatternSet& patterns);
+void populateTensorPackingPatterns(mlir::RewritePatternSet& patterns);
+
 mlir::RankedTensorType getPackedTensorType(mlir::RankedTensorType elementType, int64_t count);
 mlir::Value extractPackedChunk(mlir::Value packedValue,
                                mlir::RankedTensorType chunkType,
@@ -20,7 +28,6 @@ mlir::Value createPackedExtractRowsSlice(spatial::SpatExtractRowsOp extractRowsO
                                          mlir::OpBuilder& builder,
                                          mlir::Location loc);
 mlir::Value createPackedExtractSliceTensor(mlir::ValueRange values, mlir::OpBuilder& builder, mlir::Location loc);
-void populateTensorPackingPatterns(mlir::RewritePatternSet& patterns);
 void eraseUnusedTensorPackingOps(mlir::func::FuncOp funcOp, mlir::IRRewriter& rewriter);
 
 } // namespace onnx_mlir

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

@@ -1,7 +1,7 @@
 #include "mlir/Dialect/Tensor/IR/Tensor.h"
 
-#include "src/Accelerators/PIM/Conversion/SpatialToPim/ChannelLoweringPatterns.hpp"
 #include "src/Accelerators/PIM/Conversion/SpatialToPim/Common.hpp"
+#include "src/Accelerators/PIM/Conversion/SpatialToPim/Patterns.hpp"
 #include "src/Accelerators/PIM/Dialect/Pim/PimOps.hpp"
 #include "src/Accelerators/PIM/Dialect/Spatial/SpatialOps.hpp"
 
@@ -10,17 +10,12 @@ using namespace mlir;
 namespace onnx_mlir {
 namespace {
 
-static int32_t toPimCoreId(int32_t spatialCoreId) { return spatialCoreId; }
-
 struct ChannelSendLowering : OpRewritePattern<spatial::SpatChannelSendOp> {
   using OpRewritePattern::OpRewritePattern;
 
   LogicalResult matchAndRewrite(spatial::SpatChannelSendOp op, PatternRewriter& rewriter) const override {
-    pim::PimSendOp::create(rewriter,
-                           op.getLoc(),
-                           op.getInput(),
-                           getTensorSizeInBytesAttr(rewriter, op.getInput()),
-                           rewriter.getI32IntegerAttr(toPimCoreId(op.getTargetCoreId())));
+    pim::PimSendOp::create(
+      rewriter, op.getLoc(), op.getInput(), getTensorSizeInBytesAttr(rewriter, op.getInput()), op.getTargetCoreId());
     rewriter.eraseOp(op);
     return success();
   }
@@ -42,41 +37,7 @@ struct ChannelReceiveLowering : OpRewritePattern<spatial::SpatChannelReceiveOp>
                                                op.getResult().getType(),
                                                outputBuffer,
                                                getTensorSizeInBytesAttr(rewriter, op.getResult()),
-                                               rewriter.getI32IntegerAttr(toPimCoreId(op.getSourceCoreId())))
-                       .getOutput();
-    rewriter.replaceOp(op, received);
-    return success();
-  }
-};
-
-struct ChannelSendTensorLowering : OpRewritePattern<spatial::SpatChannelSendTensorOp> {
-  using OpRewritePattern::OpRewritePattern;
-
-  LogicalResult matchAndRewrite(spatial::SpatChannelSendTensorOp op, PatternRewriter& rewriter) const override {
-    SmallVector<int32_t> targetCoreIds;
-    targetCoreIds.reserve(op.getTargetCoreIds().size());
-    for (int32_t targetCoreId : op.getTargetCoreIds())
-      targetCoreIds.push_back(toPimCoreId(targetCoreId));
-    pim::PimSendTensorOp::create(rewriter, op.getLoc(), op.getInput(), rewriter.getDenseI32ArrayAttr(targetCoreIds));
-    rewriter.eraseOp(op);
-    return success();
-  }
-};
-
-struct ChannelReceiveTensorLowering : OpRewritePattern<spatial::SpatChannelReceiveTensorOp> {
-  using OpRewritePattern::OpRewritePattern;
-
-  LogicalResult matchAndRewrite(spatial::SpatChannelReceiveTensorOp op, PatternRewriter& rewriter) const override {
-    SmallVector<int32_t> sourceCoreIds;
-    sourceCoreIds.reserve(op.getSourceCoreIds().size());
-    for (int32_t sourceCoreId : op.getSourceCoreIds())
-      sourceCoreIds.push_back(toPimCoreId(sourceCoreId));
-    auto outputType = cast<ShapedType>(op.getOutput().getType());
-    Value outputBuffer =
-      tensor::EmptyOp::create(rewriter, op.getLoc(), outputType.getShape(), outputType.getElementType()).getResult();
-    Value received =
-      pim::PimReceiveTensorOp::create(
-        rewriter, op.getLoc(), op.getOutput().getType(), outputBuffer, rewriter.getDenseI32ArrayAttr(sourceCoreIds))
+                                               op.getSourceCoreId())
                        .getOutput();
     rewriter.replaceOp(op, received);
     return success();
@@ -125,12 +86,7 @@ struct ConcatLowering : OpRewritePattern<spatial::SpatConcatOp> {
 } // namespace
 
 void populateChannelLoweringPatterns(RewritePatternSet& patterns) {
-  patterns.add<ChannelSendLowering,
-               ChannelReceiveLowering,
-               ChannelSendTensorLowering,
-               ChannelReceiveTensorLowering,
-               ExtractRowsLowering,
-               ConcatLowering>(patterns.getContext());
+  patterns.add<ChannelSendLowering, ChannelReceiveLowering, ExtractRowsLowering, ConcatLowering>(patterns.getContext());
 }
 
 } // namespace onnx_mlir

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

@@ -16,7 +16,7 @@
 
 #include "Common/PimCommon.hpp"
 #include "src/Accelerators/PIM/Conversion/SpatialToPim/ComputeLikeRegionUtils.hpp"
-#include "src/Accelerators/PIM/Conversion/SpatialToPim/GlobalTensorMaterialization.hpp"
+#include "src/Accelerators/PIM/Conversion/SpatialToPim/Patterns.hpp"
 #include "src/Accelerators/PIM/Dialect/Spatial/SpatialOps.hpp"
 
 using namespace mlir;
@@ -76,10 +76,11 @@ struct MoveExtractSliceIntoCompute final : OpRewritePattern<mlir::tensor::Extrac
         auto inputIndex = getDirectComputeLikeInputIndex(spatCompute, uses.getOperandNumber());
         if (!inputIndex)
           return failure();
-        auto BBArgIndex = *inputIndex;
-        auto BBArgValue = spatCompute.getBody().front().getArgument(BBArgIndex);
+        auto BBArgValue = spatCompute.getInputArgument(*inputIndex);
+        if (!BBArgValue)
+          return failure();
 
-        if (BBArgValue.use_empty())
+        if (BBArgValue->use_empty())
           continue;
 
         rewriter.setInsertionPoint(&spatCompute.getBody().front().front());
@@ -89,16 +90,17 @@ struct MoveExtractSliceIntoCompute final : OpRewritePattern<mlir::tensor::Extrac
         }
 
         replaceAndEraseDirectComputeLikeInput(
-          rewriter, spatCompute.getOperation(), BBArgIndex, mapSpatToExtract[spatCompute.getOperation()]);
+          rewriter, spatCompute.getOperation(), *inputIndex, mapSpatToExtract[spatCompute.getOperation()]);
       }
       else if (auto spatComputeBatch = dyn_cast<spatial::SpatComputeBatch>(uses.getOwner())) {
         auto inputIndex = getDirectComputeLikeInputIndex(spatComputeBatch, uses.getOperandNumber());
         if (!inputIndex)
           return failure();
-        auto BBArgIndex = *inputIndex;
-        auto BBArgValue = spatComputeBatch.getBody().front().getArgument(BBArgIndex);
+        auto BBArgValue = spatComputeBatch.getInputArgument(*inputIndex);
+        if (!BBArgValue)
+          return failure();
 
-        if (BBArgValue.use_empty())
+        if (BBArgValue->use_empty())
           continue;
 
         rewriter.setInsertionPoint(&spatComputeBatch.getBody().front().front());
@@ -108,7 +110,7 @@ struct MoveExtractSliceIntoCompute final : OpRewritePattern<mlir::tensor::Extrac
         }
 
         replaceAndEraseDirectComputeLikeInput(
-          rewriter, spatComputeBatch.getOperation(), BBArgIndex, mapSpatToExtract[spatComputeBatch.getOperation()]);
+          rewriter, spatComputeBatch.getOperation(), *inputIndex, mapSpatToExtract[spatComputeBatch.getOperation()]);
       }
       else {
         {
@@ -143,170 +145,6 @@ struct MoveExtractSliceIntoCompute final : OpRewritePattern<mlir::tensor::Extrac
   }
 };
 
-// Turns runtime constants consumed by compute regions into private globals and local loads.
-struct ArithConstToGlobalMemoryPattern final : OpRewritePattern<mlir::arith::ConstantOp> {
-  using OpRewritePattern::OpRewritePattern;
-
-  LogicalResult matchAndRewrite(mlir::arith::ConstantOp constantOp, PatternRewriter& rewriter) const override {
-    Location loc = constantOp.getLoc();
-
-    if (hasWeightAlways(constantOp))
-      return failure();
-
-    if (!isa<func::FuncOp>(constantOp->getParentOp()))
-      return failure();
-
-    if (llvm::all_of(constantOp->getUsers(), [](Operation* op) {
-          if (isa<spatial::SpatCompute>(op))
-            return false;
-          if (isa<func::FuncOp>(op->getParentOp()))
-            return true;
-          return false;
-        }))
-      return failure();
-
-    rewriter.setInsertionPoint(constantOp->getParentOfType<func::FuncOp>());
-
-    auto constRankedTensorType = llvm::dyn_cast<mlir::RankedTensorType>(constantOp.getType());
-
-    if (constRankedTensorType) {
-      mlir::MemRefType memRefType =
-        mlir::MemRefType::get(constRankedTensorType.getShape(), constRankedTensorType.getElementType());
-      auto globalOp = createPrivateMemrefGlobalWithUniqueName(rewriter,
-                                                              loc,
-                                                              constantOp->getParentOfType<ModuleOp>(),
-                                                              "const",
-                                                              memRefType,
-                                                              constantOp.getValueAttr(),
-                                                              rewriter.getUnitAttr());
-      std::string argName = globalOp.getSymName().str();
-
-      llvm::DenseMap<Operation*, Value> mapSpatComputeToConst;
-
-      for (auto& constUses : llvm::make_early_inc_range(constantOp->getUses())) {
-        auto constUsers = constUses.getOwner();
-
-        if (auto spatCompute = llvm::dyn_cast<spatial::SpatCompute>(constUsers)) {
-          auto inputIndex = getDirectComputeLikeInputIndex(spatCompute, constUses.getOperandNumber());
-          if (!inputIndex)
-            return failure();
-          auto BBArgIndex = *inputIndex;
-          rewriter.setInsertionPoint(&spatCompute.getBody().front().front());
-          if (!mapSpatComputeToConst.contains(spatCompute.getOperation())) {
-            auto getGlobalOp = memref::GetGlobalOp::create(rewriter, loc, memRefType, argName);
-            auto toTensor = bufferization::ToTensorOp::create(
-              rewriter, loc, constRankedTensorType, getGlobalOp, rewriter.getUnitAttr(), rewriter.getUnitAttr());
-            mapSpatComputeToConst.insert({spatCompute.getOperation(), toTensor.getResult()});
-          }
-
-          replaceAndEraseDirectComputeLikeInput(
-            rewriter, spatCompute.getOperation(), BBArgIndex, mapSpatComputeToConst[spatCompute.getOperation()]);
-        }
-        else if (auto spatComputeBatch = llvm::dyn_cast<spatial::SpatComputeBatch>(constUsers)) {
-          auto inputIndex = getDirectComputeLikeInputIndex(spatComputeBatch, constUses.getOperandNumber());
-          if (!inputIndex)
-            return failure();
-          auto BBArgIndex = *inputIndex;
-          rewriter.setInsertionPoint(&spatComputeBatch.getBody().front().front());
-          if (!mapSpatComputeToConst.contains(spatComputeBatch.getOperation())) {
-            auto getGlobalOp = memref::GetGlobalOp::create(rewriter, loc, memRefType, argName);
-            auto toTensor = bufferization::ToTensorOp::create(
-              rewriter, loc, constRankedTensorType, getGlobalOp, rewriter.getUnitAttr(), rewriter.getUnitAttr());
-            mapSpatComputeToConst.insert({spatComputeBatch.getOperation(), toTensor.getResult()});
-          }
-
-          replaceAndEraseDirectComputeLikeInput(rewriter,
-                                                spatComputeBatch.getOperation(),
-                                                BBArgIndex,
-                                                mapSpatComputeToConst[spatComputeBatch.getOperation()]);
-        }
-        else {
-          {
-
-            if (auto spatCompute = constUses.getOwner()->getParentOfType<spatial::SpatCompute>()) {
-              rewriter.setInsertionPoint(&spatCompute.getBody().front().front());
-              if (!mapSpatComputeToConst.contains(spatCompute.getOperation())) {
-                auto getGlobalOp = memref::GetGlobalOp::create(rewriter, loc, memRefType, argName);
-                auto toTensor = bufferization::ToTensorOp::create(
-                  rewriter, loc, constRankedTensorType, getGlobalOp, rewriter.getUnitAttr(), rewriter.getUnitAttr());
-                mapSpatComputeToConst.insert({spatCompute.getOperation(), toTensor.getResult()});
-              }
-
-              rewriter.startOpModification(spatCompute.getOperation());
-              constUses.set(mapSpatComputeToConst[spatCompute.getOperation()]);
-              rewriter.finalizeOpModification(spatCompute.getOperation());
-            }
-            else if (auto spatComputeBatch = constUses.getOwner()->getParentOfType<spatial::SpatComputeBatch>()) {
-              rewriter.setInsertionPoint(&spatComputeBatch.getBody().front().front());
-              if (!mapSpatComputeToConst.contains(spatComputeBatch.getOperation())) {
-                auto getGlobalOp = memref::GetGlobalOp::create(rewriter, loc, memRefType, argName);
-                auto toTensor = bufferization::ToTensorOp::create(
-                  rewriter, loc, constRankedTensorType, getGlobalOp, rewriter.getUnitAttr(), rewriter.getUnitAttr());
-                mapSpatComputeToConst.insert({spatComputeBatch.getOperation(), toTensor.getResult()});
-              }
-
-              rewriter.startOpModification(spatComputeBatch.getOperation());
-              constUses.set(mapSpatComputeToConst[spatComputeBatch.getOperation()]);
-              rewriter.finalizeOpModification(spatComputeBatch.getOperation());
-            }
-          }
-        }
-      }
-    }
-    else if (constantOp.getType().isIntOrIndexOrFloat()) {
-      llvm::DenseMap<Operation*, Value> mapSpatComputeToConst;
-
-      for (auto& constUses : llvm::make_early_inc_range(constantOp->getUses())) {
-        auto constUsers = constUses.getOwner();
-
-        if (auto spatCompute = llvm::dyn_cast<spatial::SpatCompute>(constUsers)) {
-          auto inputIndex = getDirectComputeLikeInputIndex(spatCompute, constUses.getOperandNumber());
-          if (!inputIndex)
-            return failure();
-          auto BBArgIndex = *inputIndex;
-          rewriter.setInsertionPoint(&spatCompute.getBody().front().front());
-          auto newConst = rewriter.clone(*constantOp);
-
-          replaceAndEraseDirectComputeLikeInput(
-            rewriter, spatCompute.getOperation(), BBArgIndex, newConst->getResult(0));
-        }
-        else if (auto spatComputeBatch = llvm::dyn_cast<spatial::SpatComputeBatch>(constUsers)) {
-          auto inputIndex = getDirectComputeLikeInputIndex(spatComputeBatch, constUses.getOperandNumber());
-          if (!inputIndex)
-            return failure();
-          auto BBArgIndex = *inputIndex;
-          rewriter.setInsertionPoint(&spatComputeBatch.getBody().front().front());
-          auto newConst = rewriter.clone(*constantOp);
-
-          replaceAndEraseDirectComputeLikeInput(
-            rewriter, spatComputeBatch.getOperation(), BBArgIndex, newConst->getResult(0));
-        }
-        else if (auto parent = constUsers->getParentOfType<spatial::SpatCompute>()) {
-          if (!mapSpatComputeToConst.contains(parent)) {
-            rewriter.setInsertionPoint(&parent.getBody().front().front());
-            auto newConst = rewriter.clone(*constantOp);
-            mapSpatComputeToConst.insert({parent.getOperation(), newConst->getResult(0)});
-          }
-          constUses.set(mapSpatComputeToConst[parent.getOperation()]);
-        }
-        else {
-          auto batchParent = constUsers->getParentOfType<spatial::SpatComputeBatch>();
-          assert(batchParent && "Global Constant used direcly not within a compute");
-          if (!mapSpatComputeToConst.contains(batchParent.getOperation())) {
-            rewriter.setInsertionPoint(&batchParent.getBody().front().front());
-            auto newConst = rewriter.clone(*constantOp);
-            mapSpatComputeToConst.insert({batchParent.getOperation(), newConst->getResult(0)});
-          }
-          constUses.set(mapSpatComputeToConst[batchParent.getOperation()]);
-        }
-      }
-    }
-    if (constantOp->use_empty())
-      rewriter.eraseOp(constantOp);
-    return success();
-  }
-};
-
 // Materializes public function tensor inputs as globals so compute bodies can load them uniformly.
 struct FuncOpArgToGlobalMemoryPattern final : OpRewritePattern<mlir::func::FuncOp> {
   using OpRewritePattern::OpRewritePattern;
@@ -383,8 +221,7 @@ struct FuncOpArgToGlobalMemoryPattern final : OpRewritePattern<mlir::func::FuncO
 
 } // namespace
 void populateGlobalTensorMaterializationPatterns(RewritePatternSet& patterns) {
-  patterns.add<MoveExtractSliceIntoCompute, FuncOpArgToGlobalMemoryPattern, ArithConstToGlobalMemoryPattern>(
-    patterns.getContext());
+  patterns.add<MoveExtractSliceIntoCompute, FuncOpArgToGlobalMemoryPattern>(patterns.getContext());
 }
 
 } // namespace onnx_mlir

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

@@ -1,4 +1,4 @@
-#include "src/Accelerators/PIM/Conversion/SpatialToPim/TensorPackingPatterns.hpp"
+#include "src/Accelerators/PIM/Conversion/SpatialToPim/Patterns.hpp"
 #include "src/Accelerators/PIM/Dialect/Pim/PimOps.hpp"
 
 using namespace mlir;

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

@@ -0,0 +1,38 @@
+#include "mlir/Dialect/Linalg/IR/Linalg.h"
+
+#include "src/Accelerators/PIM/Conversion/SpatialToPim/Patterns.hpp"
+#include "src/Accelerators/PIM/Dialect/Pim/PimOps.hpp"
+
+using namespace mlir;
+
+namespace onnx_mlir {
+namespace {
+
+struct LinalgTransposeToPim final : OpRewritePattern<linalg::TransposeOp> {
+  using OpRewritePattern::OpRewritePattern;
+
+  LogicalResult matchAndRewrite(linalg::TransposeOp transposeOp, PatternRewriter& rewriter) const override {
+    SmallVector<Attribute> permutationAttrs;
+    permutationAttrs.reserve(transposeOp.getPermutation().size());
+    for (int64_t dim : transposeOp.getPermutation())
+      permutationAttrs.push_back(rewriter.getI64IntegerAttr(dim));
+
+    auto permutation = rewriter.getArrayAttr(permutationAttrs);
+    auto pimTranspose = pim::PimTransposeOp::create(rewriter,
+                                                    transposeOp.getLoc(),
+                                                    TypeRange {transposeOp->getResult(0).getType()},
+                                                    transposeOp.getInput(),
+                                                    permutation,
+                                                    transposeOp.getInit());
+    rewriter.replaceOp(transposeOp, pimTranspose.getOutput());
+    return success();
+  }
+};
+
+} // namespace
+
+void populateTransposeLoweringPatterns(RewritePatternSet& patterns) {
+  patterns.add<LinalgTransposeToPim>(patterns.getContext());
+}
+
+} // namespace onnx_mlir

src/PIM/Conversion/SpatialToPim/PhaseVerification.cpp

@@ -1,20 +0,0 @@
-#include "src/Accelerators/PIM/Conversion/SpatialToPim/PhaseVerification.hpp"
-#include "src/Accelerators/PIM/Dialect/Spatial/SpatialOps.hpp"
-
-using namespace mlir;
-
-namespace onnx_mlir {
-
-LogicalResult verifySpatialToPimBoundary(ModuleOp moduleOp) {
-  bool hasFailure = false;
-  moduleOp.walk([&](Operation* op) {
-    if (op->getDialect()->getNamespace() != "spat")
-      return;
-
-    op->emitError("illegal Spatial operation remains after Spatial-to-PIM lowering");
-    hasFailure = true;
-  });
-  return success(!hasFailure);
-}
-
-} // namespace onnx_mlir

src/PIM/Conversion/SpatialToPim/PhaseVerification.hpp

@@ -1,9 +0,0 @@
-#pragma once
-
-#include "mlir/IR/BuiltinOps.h"
-
-namespace onnx_mlir {
-
-mlir::LogicalResult verifySpatialToPimBoundary(mlir::ModuleOp moduleOp);
-
-} // namespace onnx_mlir

src/PIM/Conversion/SpatialToPim/ReturnPathNormalization.cpp

@@ -1,15 +1,18 @@
 #include "mlir/Dialect/Arith/IR/Arith.h"
 #include "mlir/Dialect/Bufferization/IR/Bufferization.h"
+#include "mlir/Dialect/Linalg/IR/Linalg.h"
 #include "mlir/Dialect/MemRef/IR/MemRef.h"
 #include "mlir/Dialect/Tensor/IR/Tensor.h"
 #include "mlir/Dialect/Tosa/IR/TosaOps.h"
 #include "mlir/IR/BuiltinOps.h"
 #include "mlir/IR/IRMapping.h"
 #include "mlir/IR/SymbolTable.h"
+#include "mlir/Transforms/FoldUtils.h"
 
 #include "Conversion/ONNXToSpatial/Common/Common.hpp"
+#include "Conversion/SpatialToPim/SpatialToPimPass.hpp"
+#include "src/Accelerators/PIM/Common/PimCommon.hpp"
 #include "src/Accelerators/PIM/Conversion/SpatialToPim/Common.hpp"
-#include "src/Accelerators/PIM/Conversion/SpatialToPim/ReturnPathNormalization.hpp"
 #include "src/Accelerators/PIM/Dialect/Pim/PimOps.hpp"
 #include "src/Dialect/ONNX/ONNXOps.hpp"
 
@@ -38,15 +41,10 @@ static bool isReturnHelperChainOp(Operation* op) {
              tensor::ExpandShapeOp,
              tensor::CastOp,
              tosa::ReshapeOp,
-             ONNXTransposeOp,
+             linalg::TransposeOp,
              pim::PimTransposeOp>(op);
 }
 
-static void markOpToRemove(ReturnPathState& state, Operation* op) {
-  if (!llvm::is_contained(state.operationsToRemove, op))
-    state.operationsToRemove.push_back(op);
-}
-
 static std::string makeUniqueSymbolName(Operation* symbolTableOp, StringRef baseName) {
   std::string name = baseName.str();
   unsigned suffix = 0;
@@ -279,11 +277,10 @@ static LogicalResult mapIndicesThroughHelperChain(ArrayRef<int64_t> sourceIndice
       continue;
     }
 
-    if (auto transposeOp = dyn_cast<ONNXTransposeOp>(op)) {
+    if (auto transposeOp = dyn_cast<linalg::TransposeOp>(op)) {
       SmallVector<int64_t> nextIndices(currentIndices.size());
       SmallVector<int64_t> nextShape(currentShape.size());
-      for (auto [destIndex, attr] : llvm::enumerate(transposeOp.getPermAttr().getAsRange<IntegerAttr>())) {
-        int64_t sourceIndex = attr.getInt();
+      for (auto [destIndex, sourceIndex] : llvm::enumerate(transposeOp.getPermutation())) {
         nextIndices[destIndex] = currentIndices[sourceIndex];
         nextShape[destIndex] = currentShape[sourceIndex];
       }
@@ -318,7 +315,8 @@ static LogicalResult mapIndicesThroughHelperChain(ArrayRef<int64_t> sourceIndice
   return success();
 }
 
-static void cloneMappedHelperOperands(Operation* op, IRMapping& mapping, IRRewriter& rewriter) {
+static void
+cloneMappedHelperOperands(Operation* op, IRMapping& mapping, IRRewriter& rewriter, OperationFolder& constantFolder) {
   for (Value operand : op->getOperands()) {
     if (mapping.lookupOrNull(operand))
       continue;
@@ -327,7 +325,12 @@ static void cloneMappedHelperOperands(Operation* op, IRMapping& mapping, IRRewri
     if (!definingOp)
       continue;
 
-    if (!isa<tensor::EmptyOp, arith::ConstantOp>(definingOp))
+    if (auto constantOp = dyn_cast<arith::ConstantOp>(definingOp)) {
+      mapping.map(operand, getOrCreateConstantLike(constantFolder, constantOp));
+      continue;
+    }
+
+    if (!isa<tensor::EmptyOp>(definingOp))
       continue;
 
     Operation* clonedOp = rewriter.clone(*definingOp, mapping);
@@ -337,15 +340,18 @@ static void cloneMappedHelperOperands(Operation* op, IRMapping& mapping, IRRewri
   }
 }
 
-static void
-cloneHelperChain(Value sourceValue, ArrayRef<Operation*> helperChain, IRRewriter& rewriter, Value& clonedValue) {
+static void cloneHelperChain(Value sourceValue,
+                             ArrayRef<Operation*> helperChain,
+                             IRRewriter& rewriter,
+                             OperationFolder& constantFolder,
+                             Value& clonedValue) {
   IRMapping mapping;
   mapping.map(sourceValue, sourceValue);
   clonedValue = sourceValue;
 
   rewriter.setInsertionPointAfterValue(sourceValue);
   for (Operation* op : helperChain) {
-    cloneMappedHelperOperands(op, mapping, rewriter);
+    cloneMappedHelperOperands(op, mapping, rewriter, constantFolder);
     Operation* clonedOp = rewriter.clone(*op, mapping);
     for (auto [originalResult, newResult] : llvm::zip(op->getResults(), clonedOp->getResults()))
       mapping.map(originalResult, newResult);
@@ -360,23 +366,26 @@ static Value emitHostCopy(IRRewriter& rewriter,
                           Value sourceValue,
                           int32_t hostTargetOffset,
                           int32_t deviceSourceOffset,
-                          int32_t sizeInBytes) {
+                          int32_t sizeInBytes,
+                          OperationFolder& constantFolder) {
+  Operation* anchorOp = sourceValue.getDefiningOp() ? sourceValue.getDefiningOp() : outputTensor.getDefiningOp();
+  assert(anchorOp && "expected a concrete op anchor for return-path host copy constants");
+  Value hostTargetOffsetValue = getOrCreateIndexConstant(constantFolder, anchorOp, hostTargetOffset);
+  Value deviceSourceOffsetValue = getOrCreateIndexConstant(constantFolder, anchorOp, deviceSourceOffset);
   return PimMemCopyDevToHostOp::create(rewriter,
                                        loc,
                                        outputTensor.getType(),
+                                       hostTargetOffsetValue,
+                                       deviceSourceOffsetValue,
                                        outputTensor,
                                        sourceValue,
-                                       rewriter.getI32IntegerAttr(hostTargetOffset),
-                                       rewriter.getI32IntegerAttr(deviceSourceOffset),
                                        rewriter.getI32IntegerAttr(sizeInBytes))
     .getOutput();
 }
 
 } // namespace
 
-void addReturnOutputBuffers(func::ReturnOp returnOp,
-                            IRRewriter& rewriter,
-                            SmallVectorImpl<OutputTensorFactory>& outputTensors) {
+void raptor::SpatialToPimPass::addReturnOutputBuffers(func::ReturnOp returnOp, IRRewriter& rewriter) {
   outputTensors.reserve(returnOp->getNumOperands());
   for (auto [index, returnValue] : llvm::enumerate(returnOp->getOperands())) {
     Value currentReturnValue = returnValue;
@@ -411,70 +420,85 @@ void addReturnOutputBuffers(func::ReturnOp returnOp,
   }
 }
 
-ReturnPathLoweringResult lowerComputeResultReturnPath(
-  spatial::SpatCompute computeOp, OpResult result, Value yieldValue, ReturnPathState& state, IRRewriter& rewriter) {
-  Location loc = computeOp->getLoc();
-  auto yieldType = cast<TensorType>(yieldValue.getType());
+raptor::SpatialToPimPass::ReturnPathLoweringResult raptor::SpatialToPimPass::lowerProducedValueReturnPath(
+  Operation* producerOp, Value producedValue, Value storedValue, IRRewriter& rewriter) {
+  Location loc = producerOp->getLoc();
+  OperationFolder constantFolder(producerOp->getContext());
+  auto storedTensorType = cast<TensorType>(storedValue.getType());
 
-  if (auto returnUse = analyzeReturnUse(result)) {
-    Value storedValue = yieldValue;
-    cloneHelperChain(yieldValue, returnUse->helperChain, rewriter, storedValue);
+  if (auto returnUse = analyzeReturnUse(producedValue)) {
+    Value currentStoredValue = storedValue;
+    cloneHelperChain(storedValue, returnUse->helperChain, rewriter, constantFolder, currentStoredValue);
     for (Operation* op : returnUse->helperChain)
-      markOpToRemove(state, op);
+      markOpToRemove(op);
 
-    auto storedType = cast<ShapedType>(storedValue.getType());
-    size_t elementSize = storedType.getElementTypeBitWidth() / 8;
-    if (auto storedOp = storedValue.getDefiningOp())
+    auto storedType = cast<ShapedType>(currentStoredValue.getType());
+    size_t elementSize = getElementTypeSizeInBytes(storedType.getElementType());
+    if (auto storedOp = currentStoredValue.getDefiningOp())
       rewriter.setInsertionPointAfter(storedOp);
-    Value outputTensor = state.outputTensors[returnUse->returnIndex](rewriter, loc);
-    emitHostCopy(
-      rewriter, loc, outputTensor, storedValue, 0, 0, static_cast<int32_t>(storedType.getNumElements() * elementSize));
+    Value outputTensor = outputTensors[returnUse->returnIndex](rewriter, loc);
+    emitHostCopy(rewriter,
+                 loc,
+                 outputTensor,
+                 currentStoredValue,
+                 0,
+                 0,
+                 static_cast<int32_t>(storedType.getNumElements() * elementSize),
+                 constantFolder);
     return ReturnPathLoweringResult::Handled;
   }
 
-  auto resultUses = result.getUses();
+  auto resultUses = producedValue.getUses();
   if (rangeLength(resultUses) == 1) {
     OpOperand& resultUse = *resultUses.begin();
     Operation* resultUser = resultUse.getOwner();
 
     if (isa<func::ReturnOp>(resultUser)) {
       size_t resultIndexInReturn = resultUse.getOperandNumber();
-      size_t elementSize = yieldType.getElementType().getIntOrFloatBitWidth() / 8;
-      rewriter.setInsertionPointAfterValue(yieldValue);
-      Value outputTensor = state.outputTensors[resultIndexInReturn](rewriter, loc);
-      emitHostCopy(
-        rewriter, loc, outputTensor, yieldValue, 0, 0, static_cast<int32_t>(yieldType.getNumElements() * elementSize));
+      size_t elementSize = getElementTypeSizeInBytes(storedTensorType.getElementType());
+      rewriter.setInsertionPointAfterValue(storedValue);
+      Value outputTensor = outputTensors[resultIndexInReturn](rewriter, loc);
+      emitHostCopy(rewriter,
+                   loc,
+                   outputTensor,
+                   storedValue,
+                   0,
+                   0,
+                   static_cast<int32_t>(storedTensorType.getNumElements() * elementSize),
+                   constantFolder);
       return ReturnPathLoweringResult::Handled;
     }
   }
 
-  if (auto concatReturnUse = analyzeConcatReturnUse(result)) {
-    size_t elementSize = yieldType.getElementTypeBitWidth() / 8;
+  if (auto concatReturnUse = analyzeConcatReturnUse(producedValue)) {
+    size_t elementSize = getElementTypeSizeInBytes(storedTensorType.getElementType());
     for (Operation* concatOp : concatReturnUse->concatChain)
-      markOpToRemove(state, concatOp);
+      markOpToRemove(concatOp);
 
     if (concatReturnUse->helperChain.empty()) {
-      rewriter.setInsertionPointAfterValue(yieldValue);
-      Value outputTensor = state.outputTensors[concatReturnUse->returnIndex](rewriter, loc);
+      rewriter.setInsertionPointAfterValue(storedValue);
+      Value outputTensor = outputTensors[concatReturnUse->returnIndex](rewriter, loc);
       auto outputType = cast<ShapedType>(outputTensor.getType());
       int64_t flatOffset = computeFlatElementIndex(concatReturnUse->sliceOffsets, outputType.getShape());
       emitHostCopy(rewriter,
                    loc,
                    outputTensor,
-                   yieldValue,
+                   storedValue,
                    static_cast<int32_t>(flatOffset * elementSize),
                    0,
-                   static_cast<int32_t>(yieldType.getNumElements() * elementSize));
+                   static_cast<int32_t>(storedTensorType.getNumElements() * elementSize),
+                   constantFolder);
       return ReturnPathLoweringResult::Handled;
     }
 
-    auto storedType = dyn_cast<RankedTensorType>(yieldValue.getType());
+    auto storedType = dyn_cast<RankedTensorType>(storedValue.getType());
     if (!storedType) {
-      computeOp.emitOpError("has an unsupported non-ranked concat-return helper yield during Spatial-to-PIM lowering");
+      producerOp->emitOpError(
+        "has an unsupported non-ranked concat-return helper yield during Spatial-to-PIM lowering");
       return ReturnPathLoweringResult::Failure;
     }
-    rewriter.setInsertionPointAfterValue(yieldValue);
-    Value outputTensor = state.outputTensors[concatReturnUse->returnIndex](rewriter, loc);
+    rewriter.setInsertionPointAfterValue(storedValue);
+    Value outputTensor = outputTensors[concatReturnUse->returnIndex](rewriter, loc);
     auto outputType = cast<ShapedType>(outputTensor.getType());
     for (int64_t linearIndex = 0; linearIndex < storedType.getNumElements(); ++linearIndex) {
       SmallVector<int64_t> sourceIndices = expandFlatElementIndex(linearIndex, storedType.getShape());
@@ -484,7 +508,7 @@ ReturnPathLoweringResult lowerComputeResultReturnPath(
       SmallVector<int64_t> destinationIndices;
       if (failed(mapIndicesThroughHelperChain(
             sourceIndices, concatReturnUse->concatShape, concatReturnUse->helperChain, destinationIndices))) {
-        computeOp.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;
       }
 
@@ -503,7 +527,7 @@ ReturnPathLoweringResult lowerComputeResultReturnPath(
       auto scalarTensorType =
         RankedTensorType::get(SmallVector<int64_t>(storedType.getRank(), 1), storedType.getElementType());
       auto elementSlice = tensor::ExtractSliceOp::create(
-        rewriter, loc, scalarTensorType, yieldValue, extractOffsets, extractSizes, extractStrides);
+        rewriter, loc, scalarTensorType, storedValue, extractOffsets, extractSizes, extractStrides);
       rewriter.setInsertionPointAfter(elementSlice);
 
       int64_t destinationFlatOffset = computeFlatElementIndex(destinationIndices, outputType.getShape());
@@ -513,7 +537,8 @@ ReturnPathLoweringResult lowerComputeResultReturnPath(
                                   elementSlice.getResult(),
                                   static_cast<int32_t>(destinationFlatOffset * elementSize),
                                   0,
-                                  static_cast<int32_t>(elementSize));
+                                  static_cast<int32_t>(elementSize),
+                                  constantFolder);
     }
     return ReturnPathLoweringResult::Handled;
   }
@@ -521,7 +546,12 @@ ReturnPathLoweringResult lowerComputeResultReturnPath(
   return ReturnPathLoweringResult::NotReturnPath;
 }
 
-void replaceReturnWithOutputBuffers(func::ReturnOp returnOp, IRRewriter& rewriter, ReturnPathState& state) {
+raptor::SpatialToPimPass::ReturnPathLoweringResult raptor::SpatialToPimPass::lowerComputeResultReturnPath(
+  spatial::SpatCompute computeOp, OpResult result, Value yieldValue, IRRewriter& rewriter) {
+  return lowerProducedValueReturnPath(computeOp.getOperation(), result, yieldValue, rewriter);
+}
+
+void raptor::SpatialToPimPass::replaceReturnWithOutputBuffers(func::ReturnOp returnOp, IRRewriter& rewriter) {
   auto markOwnedReturnChain = [&](Operation* op, auto&& markOwnedReturnChain) -> void {
     if (!op)
       return;
@@ -538,13 +568,13 @@ void replaceReturnWithOutputBuffers(func::ReturnOp returnOp, IRRewriter& rewrite
 
     if (isReturnHelperChainOp(op)) {
       Value source = op->getOperand(0);
-      markOpToRemove(state, op);
+      markOpToRemove(op);
       markOwnedReturnChain(source.getDefiningOp(), markOwnedReturnChain);
       return;
     }
 
     if (auto computeOp = dyn_cast<spatial::SpatCompute>(op)) {
-      markOpToRemove(state, computeOp);
+      markOpToRemove(computeOp);
       if (!computeOp.getInputs().empty())
         for (Value input : computeOp.getInputs())
           markOwnedReturnChain(input.getDefiningOp(), markOwnedReturnChain);
@@ -552,23 +582,29 @@ void replaceReturnWithOutputBuffers(func::ReturnOp returnOp, IRRewriter& rewrite
     }
 
     if (auto concatOp = dyn_cast<tensor::ConcatOp>(op)) {
-      markOpToRemove(state, concatOp);
+      markOpToRemove(concatOp);
       for (Value operand : concatOp.getOperands())
         markOwnedReturnChain(operand.getDefiningOp(), markOwnedReturnChain);
       return;
     }
 
     if (auto concatOp = dyn_cast<spatial::SpatConcatOp>(op)) {
-      markOpToRemove(state, concatOp);
+      markOpToRemove(concatOp);
       for (Value operand : concatOp.getInputs())
         markOwnedReturnChain(operand.getDefiningOp(), markOwnedReturnChain);
       return;
     }
 
     if (auto concatOp = dyn_cast<pim::PimConcatOp>(op)) {
-      markOpToRemove(state, concatOp);
+      markOpToRemove(concatOp);
       for (Value operand : concatOp.getInputs())
         markOwnedReturnChain(operand.getDefiningOp(), markOwnedReturnChain);
+      return;
+    }
+
+    if (auto receiveOp = dyn_cast<spatial::SpatChannelReceiveOp>(op)) {
+      markOpToRemove(receiveOp);
+      return;
     }
   };
 
@@ -578,7 +614,7 @@ void replaceReturnWithOutputBuffers(func::ReturnOp returnOp, IRRewriter& rewrite
     size_t orderWithinReturn = it.index();
     Operation* returnOperand = it.value().getDefiningOp();
     rewriter.setInsertionPoint(returnOp);
-    Value outputTensor = state.outputTensors[orderWithinReturn](rewriter, loc);
+    Value outputTensor = outputTensors[orderWithinReturn](rewriter, loc);
     rewriter.modifyOpInPlace(returnOp, [&] { returnOp.setOperand(orderWithinReturn, outputTensor); });
     markOwnedReturnChain(returnOperand, markOwnedReturnChain);
   }