projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: a6be243) | patch
[X86][SSE] Blend any v8i16/v4i32 shift with 2 shift unique values
authorSimon Pilgrim <llvm-dev@redking.me.uk>
Mon, 2 Jul 2018 15:14:07 +0000 (15:14 +0000)
committerSimon Pilgrim <llvm-dev@redking.me.uk>
Mon, 2 Jul 2018 15:14:07 +0000 (15:14 +0000)
commit2bc8e079f2202ce6f223d034ed2dde141f7eec60
treec47aee9429b1eaccee84f2648a86522c8f9c57eatree | snapshot
parenta6be2437e776e0c2f1d39a42beeb2ffe3166ed2bcommit | diff
[X86][SSE] Blend any v8i16/v4i32 shift with 2 shift unique values

We were only doing this for basic blends, despite shuffle lowering now being good enough to handle more complex blends. This means that the two v8i16 splat shifts are performed in parallel instead of serially as the general shift case.

llvm-svn: 336113
llvm/lib/Target/X86/X86ISelLowering.cpp diff | blob | history
llvm/test/CodeGen/X86/lower-vec-shift.ll diff | blob | history
Domain: System / Toolchain;