projects / oota-llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 9526911) | patch
[X86] Generate VPSHUFB for in-place v16i16 shuffles
authorAdam Nemet <anemet@apple.com>
Tue, 25 Mar 2014 17:47:06 +0000 (17:47 +0000)
committerAdam Nemet <anemet@apple.com>
Tue, 25 Mar 2014 17:47:06 +0000 (17:47 +0000)
commit6f4f46cf112567e273504b41b47301f922347577
treec591c72b3595d8f550180a850f76660ca6a97790tree | snapshot
parent95269118099c8cf56f7f24df56687a73152c5c85commit | diff
[X86] Generate VPSHUFB for in-place v16i16 shuffles

This used to resort to splitting the 256-bit operation into two 128-bit
shuffles and then recombining the results.

Fixes <rdar://problem/16167303>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204735 91177308-0d34-0410-b5e6-96231b3b80d8
lib/Target/X86/X86ISelLowering.cpp diff | blob | history
test/CodeGen/X86/vec_shuffle-40.ll [new file with mode: 0644] blob
C/C++ LLVM-based compilers that forbids OOTA behaviors