diff options
| author | Nicolas Vasilache <ntv@google.com> | 2020-05-12 22:21:36 -0400 |
|---|---|---|
| committer | Nicolas Vasilache <ntv@google.com> | 2020-05-13 10:11:30 -0400 |
| commit | e0b99a5de4cbf00bfa46d06caf1ebf64a6456537 (patch) | |
| tree | b8e9d530999c2988bb15ba005ca5f8d0d74f7d92 /llvm/utils/FileCheck/FileCheck.cpp | |
| parent | 195de442daba5ae1754528eb2869ca50c3af2669 (diff) | |
| download | llvm-e0b99a5de4cbf00bfa46d06caf1ebf64a6456537.zip llvm-e0b99a5de4cbf00bfa46d06caf1ebf64a6456537.tar.gz llvm-e0b99a5de4cbf00bfa46d06caf1ebf64a6456537.tar.bz2 | |
[mlir] Add SubViewOp::getOrCreateRanges and fix folding pattern
The existing implementation of SubViewOp::getRanges relies on all
offsets/sizes/strides to be dynamic values and does not work in
combination with canonicalization. This revision adds a
SubViewOp::getOrCreateRanges to create the missing constants in the
canonicalized case.
This allows reactivating the fused pass with staged pattern
applications.
However another issue surfaces that the SubViewOp verifier is now too
strict to allow folding. The existing folding pattern is turned into a
canonicalization pattern which rewrites memref_cast + subview into
subview + memref_cast.
The transform-patterns-matmul-to-vector can then be reactivated.
Differential Revision: https://reviews.llvm.org/D79759
Diffstat (limited to 'llvm/utils/FileCheck/FileCheck.cpp')
0 files changed, 0 insertions, 0 deletions