[PATCH] D21144: [X86]: Quit promoting 8 and 16 bit compares to 32 bit.

Kevin B. Smith via llvm-commits llvm-commits at lists.llvm.org
Wed Jun 8 10:48:18 PDT 2016


kbsmith1 created this revision.
kbsmith1 added reviewers: echristo, DavidKreitzer, sanjoy.
kbsmith1 added a subscriber: llvm-commits.
Herald added a subscriber: mehdi_amini.

This change effectively just reverts r195496, and updates the tests as needed.
8 and 16 bit compares no longer promoted up into 32 bit compares. This has some
nice performance improvements, especially in eembc/rgbcmykv2.  In order for this
not to cause performance regressions in 401.bzip2, changes http://reviews.llvm.org/D21085
are also necessary to get all the necessary movb and movw instructions promoted to
movzbl/movzwl.

http://reviews.llvm.org/D21144

Files:
  lib/Target/X86/X86ISelLowering.cpp
  test/CodeGen/X86/3addr-16bit.ll
  test/CodeGen/X86/atomic16.ll
  test/CodeGen/X86/atomic8.ll
  test/CodeGen/X86/ctpop-combine.ll
  test/CodeGen/X86/machine-sink-and-implicit-null-checks.ll
  test/CodeGen/X86/memcmp.ll
  test/CodeGen/X86/pr5145.ll
  test/CodeGen/X86/x86-shrink-wrapping.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D21144.60068.patch
Type: text/x-patch
Size: 10579 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20160608/6b6f50af/attachment.bin>


More information about the llvm-commits mailing list