[PATCH] D26855: New unsafe-fp-math implementation for X86 target

[Andrew V. Tischenko via Phabricator via llvm-commits]

avt77 updated this revision to Diff 88003.
avt77 added a comment.
Herald added a subscriber: igorb.

I fixed all known issues:

- AVX512 is now again supported by DAGCombiner
- FMA instructions are being used when FMA is enabled

This version clearly shows the advantage of sched model usage: it selects reciprocal code when it's profit only (e.g. compare v8f32_one_step and v8f32_one_step_2_divs, etc.)

In addition I removed the speedup related code from this patch because it was included into the special review: D29627 <https://reviews.llvm.org/D29627>


https://reviews.llvm.org/D26855

Files:
  include/llvm/CodeGen/MachineCombinerPattern.h
  include/llvm/Target/TargetLowering.h
  lib/CodeGen/MachineCombiner.cpp
  lib/CodeGen/MachineTraceMetrics.cpp
  lib/CodeGen/TargetLoweringBase.cpp
  lib/Target/X86/X86ISelLowering.cpp
  lib/Target/X86/X86ISelLowering.h
  lib/Target/X86/X86InstrInfo.cpp
  lib/Target/X86/X86InstrInfo.h
  test/CodeGen/X86/recip-fastmath.ll
  test/CodeGen/X86/recip-fastmath2.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D26855.88003.patch
Type: text/x-patch
Size: 101388 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20170210/eb77867d/attachment.bin>