compact syntax for spatial tensor ops
Validate Operations / validate-operations (push) Has been cancelled
Validate Operations / validate-operations (push) Has been cancelled
better IR compaction after dcp merge remove pim.mvm op better memory report
This commit is contained in:
@@ -21,12 +21,6 @@ def spatToPimVMM : Pat<
|
||||
(NativeCodeCall<"onnx_mlir::getBestOutputTensorFromOperandsOrAllocate($_builder, $0.getDefiningOp())"> $srcOpRes))
|
||||
>;
|
||||
|
||||
def spatToPimMVM : Pat<
|
||||
(SpatMVMOp:$srcOpRes $weightIndex, $vector),
|
||||
(PimMVMOp $weightIndex, $vector,
|
||||
(NativeCodeCall<"onnx_mlir::getBestOutputTensorFromOperandsOrAllocate($_builder, $0.getDefiningOp())"> $srcOpRes))
|
||||
>;
|
||||
|
||||
def spatToPimVVAdd : Pat<
|
||||
(SpatVAddOp:$srcOpRes $a, $b),
|
||||
(PimVVAddOp $a, $b,
|
||||
|
||||
Reference in New Issue
Block a user