projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 394db22) | patch
[MLIR] Vector store to load forwarding
authorAnand Kodnani <anand.kodnani@intel.com>
Tue, 28 Jul 2020 17:37:16 +0000 (10:37 -0700)
committerDiego Caballero <diego.caballero@intel.com>
Tue, 28 Jul 2020 18:30:54 +0000 (11:30 -0700)
commit834133c950fce120d0378d09718d32a320cbcd72
treef86ba85763029cb5d1f53a1462cd680c41d88ed2tree | snapshot
parent394db2259575ef3cac8d3d37836b11eb2373c435commit | diff
[MLIR] Vector store to load forwarding

The MemRefDataFlow pass does store to load forwarding
only for affine store/loads. This patch updates the pass
to use affine read/write interface which enables vector
forwarding.

Reviewed By: dcaballe, bondhugula, ftynse

Differential Revision: https://reviews.llvm.org/D84302
mlir/include/mlir/Dialect/Affine/IR/AffineMemoryOpInterfaces.td diff | blob | history
mlir/include/mlir/Dialect/Affine/IR/AffineOps.td diff | blob | history
mlir/lib/Transforms/MemRefDataFlowOpt.cpp diff | blob | history
mlir/test/Transforms/memref-dataflow-opt.mlir diff | blob | history
Domain: System / Toolchain;