projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: a14945c) | patch
[mlir][sparse] adjust output shape inference to new tensor abstraction
authorAart Bik <ajcbik@google.com>
Tue, 5 Jan 2021 21:29:28 +0000 (13:29 -0800)
committerAart Bik <ajcbik@google.com>
Tue, 5 Jan 2021 23:31:39 +0000 (15:31 -0800)
commit8b124c19f52cb8ed0236b602df56787553e1e1b6
treeed18295a7a6a27afdde8027f8e8fb8f35cd16966tree | snapshot
parenta14945c1db614261a6f8d5d199e246d78f51e977commit | diff
[mlir][sparse] adjust output shape inference to new tensor abstraction

Nicolas changed the tensor abstraction so that every output has
its own shape definition. This simplifies the "inference" that
was used in the sparse compiler.

Reviewed By: penpornk

Differential Revision: https://reviews.llvm.org/D94119
mlir/lib/Dialect/Linalg/Transforms/Sparsification.cpp diff | blob | history
mlir/test/Dialect/Linalg/sparse_2d.mlir diff | blob | history
Domain: System / Toolchain;