diff options
author | Hanhan Wang <hanchung@google.com> | 2022-12-16 13:03:10 -0800 |
---|---|---|
committer | Hanhan Wang <hanchung@google.com> | 2022-12-19 17:52:10 -0800 |
commit | 3ebc6bee6b23f163f4977836bd50bc952449836b (patch) | |
tree | 6fe20c4eabbaa4f59033dffe9ebdc389dd44a050 /llvm/lib/CodeGen/CodeGen.cpp | |
parent | aee1adedc8a73653e1bc743a39d3ff2966da7a77 (diff) | |
download | llvm-3ebc6bee6b23f163f4977836bd50bc952449836b.zip llvm-3ebc6bee6b23f163f4977836bd50bc952449836b.tar.gz llvm-3ebc6bee6b23f163f4977836bd50bc952449836b.tar.bz2 |
[mlir][tensor][linalg] Add a pattern that generalizes tensor.unpack op.
The pattern generalizes a tensor::UnPackOp into a sequence of tensor +
Linalg ops, when the outer dims are all 1s. It uses the trick of
rank-reduced tensor.extract_slice to get the tile; transpose the tile;
extract sub tile for incomplete cases if needed; use tensor.insert_slice
to insert it to the destination tensor.
Reviewed By: tyb0807, chelini
Differential Revision: https://reviews.llvm.org/D140254
Diffstat (limited to 'llvm/lib/CodeGen/CodeGen.cpp')
0 files changed, 0 insertions, 0 deletions