diff options
| author | Adam Paszke <apaszke@google.com> | 2023-12-06 11:15:47 +0100 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2023-12-06 11:15:47 +0100 |
| commit | 34df53739af2ce0ffb2625075ee2e613b278969c (patch) | |
| tree | 810985ceda5ecd5a1a91140c7a010139f477373d /flang/lib/Frontend/CompilerInvocation.cpp | |
| parent | 6b1aa319754e76366edd88e10034e0539710d946 (diff) | |
| download | llvm-34df53739af2ce0ffb2625075ee2e613b278969c.zip llvm-34df53739af2ce0ffb2625075ee2e613b278969c.tar.gz llvm-34df53739af2ce0ffb2625075ee2e613b278969c.tar.bz2 | |
Revert "[mlir][Vector] Add fold transpose(shape_cast) -> shape_cast (#73951)" (#74579)
This reverts commit f42b7615b862bb5f77981f619f92877eb20adf54.
The fold pattern is incorrect, because it does not even look at the
permutation of non-unit dims and is happy to replace a pattern such as
```
%22 = vector.shape_cast %21 : vector<1x256x256xf32> to vector<256x256xf32>
%23 = vector.transpose %22, [1, 0] : vector<256x256xf32> to vector<256x256xf32>
```
with
```
%22 = vector.shape_cast %21 : vector<1x256x256xf32> to vector<256x256xf32>
```
which is obviously incorrect.
Diffstat (limited to 'flang/lib/Frontend/CompilerInvocation.cpp')
0 files changed, 0 insertions, 0 deletions