[PATCH] D27381: AMDGPU: Make f16 ConstantFP legal

[Matt Arsenault via Phabricator via llvm-commits]

arsenm created this revision.
arsenm added a subscriber: llvm-commits.
Herald added a reviewer: tstellarAMD.
Herald added subscribers: tony-tye, yaxunl, nhaehnle, wdng, kzhuravl.

Not having this legal led to combine failures, resulting
in dumb things like bitcasts of constants not being folded
away.

      

The only reason I'm leaving the v_mov_b32 hack that f32
 already uses is to avoid madak formation test regressions.
PeepholeOptimizer has an ordering issue where the immediate
 fold attempt is into the sgpr->vgpr copy instead of the actual
 use. Running it twice avoids that problem.


https://reviews.llvm.org/D27381

Files:
  lib/Target/AMDGPU/SIISelLowering.cpp
  lib/Target/AMDGPU/SIISelLowering.h
  lib/Target/AMDGPU/SIInstructions.td
  test/CodeGen/AMDGPU/br_cc.f16.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D27381.80169.patch
Type: text/x-patch
Size: 3529 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20161203/4f438bd6/attachment.bin>