[PATCH] D21358: Cannonicalize applicable vector shit immediates as swaps

[Nemanja Ivanovic via llvm-commits]

nemanjai created this revision.
nemanjai added reviewers: wschmidt, hfinkel, kbarton, amehsan, seurer, cycheng, tjablin.
nemanjai added a subscriber: llvm-commits.
nemanjai set the repository for this revision to rL LLVM.

The VSX swap optimization knows what to do with swaps and can often eliminate them if they're redundant. This patch hopefully provides more opportunities to swap optimization by emitting the actual swap operation rather than an equivalent shift operation (VSLDOI, XXSLDWI).

Repository:
  rL LLVM

http://reviews.llvm.org/D21358

Files:
  lib/Target/PowerPC/PPCISelLowering.cpp
  lib/Target/PowerPC/PPCISelLowering.h
  lib/Target/PowerPC/PPCInstrVSX.td
  test/CodeGen/PowerPC/cannonicalize-vector-shifts.ll
  test/CodeGen/PowerPC/p8-scalar_vector_conversions.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D21358.60776.patch
Type: text/x-patch
Size: 7157 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20160614/41efa545/attachment.bin>