| commit | ac77529540b03ac974fc203e668747b2d27374ec | [log] [tgz] |
|---|---|---|
| author | Simon Pilgrim <llvm-dev@redking.me.uk> | Mon Aug 10 20:21:15 2015 +0000 |
| committer | Simon Pilgrim <llvm-dev@redking.me.uk> | Mon Aug 10 20:21:15 2015 +0000 |
| tree | 5235faeed1b503f0863818173c22dd8a4b33876d | |
| parent | 90622a1bc989e5bcb05a5d1fcf9c36fb14a2bbfb [diff] |
[InstCombine] Move SSE2/AVX2 arithmetic vector shift folding to instcombiner As discussed in D11760, this patch moves the (V)PSRA(WD) arithmetic shift-by-constant folding to InstCombine to match the logical shift implementations. Differential Revision: http://reviews.llvm.org/D11886 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244495 91177308-0d34-0410-b5e6-96231b3b80d8