projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 9ca97cd) | patch
[mlir] Rename ShapeTypeConversion to ShapeBufferize
authorSean Silva <silvasean@google.com>
Mon, 12 Oct 2020 19:23:45 +0000 (12:23 -0700)
committerSean Silva <silvasean@google.com>
Wed, 14 Oct 2020 19:39:16 +0000 (12:39 -0700)
commit6b30fb7653948fec80ca0cea19d8691495c96c28
treebc19adbe849afc773ce7ac2f9c20d2d318e355a7tree | snapshot
parent9ca97cde8508b92856d22e2164c8b6fb6756696ecommit | diff
[mlir] Rename ShapeTypeConversion to ShapeBufferize

Once we have tensor_to_memref ops suitable for type materializations,
this pass can be split into a generic type conversion pattern.

Part of the refactor discussed in:
https://llvm.discourse.group/t/what-is-the-strategy-for-tensor-memref-conversion-bufferization/1938/17

Differential Revision: https://reviews.llvm.org/D89258
mlir/include/mlir/Dialect/Shape/Transforms/Passes.h diff | blob | history
mlir/include/mlir/Dialect/Shape/Transforms/Passes.td diff | blob | history
mlir/lib/Dialect/Shape/Transforms/Bufferize.cpp [moved from mlir/lib/Dialect/Shape/Transforms/ShapeTypeConversion.cpp with 84% similarity] diff | blob | history
mlir/lib/Dialect/Shape/Transforms/CMakeLists.txt diff | blob | history
mlir/test/Dialect/Shape/bufferize.mlir [moved from mlir/test/Dialect/Shape/shape-type-conversion.mlir with 84% similarity] diff | blob | history
Domain: System / Toolchain;