| commit | 6e4fb35e80829197e53f108a56ae21745d2b392e | [log] [tgz] |
|---|---|---|
| author | Simon Pilgrim <llvm-dev@redking.me.uk> | Wed Jul 18 19:55:19 2018 +0000 |
| committer | Simon Pilgrim <llvm-dev@redking.me.uk> | Wed Jul 18 19:55:19 2018 +0000 |
| tree | 2ad939bfe29628468508646ff844730a6effddf8 | |
| parent | 5aa954ccaf2163b91cbf9ac531e9c1ec4ea8d9fa [diff] |
[X86][SSE] Canonicalize scalar fp arithmetic shuffle patterns As discussed on PR38197, this canonicalizes MOVS*(N0, OP(N0, N1)) --> MOVS*(N0, SCALAR_TO_VECTOR(OP(N0[0], N1[0]))) This returns the scalar-fp codegen lost by rL336971. Additionally it handles the OP(N1, N0)) case for commutable (FADD/FMUL) ops. Differential Revision: https://reviews.llvm.org/D49474 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@337419 91177308-0d34-0410-b5e6-96231b3b80d8