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

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

avt77 updated this revision to Diff 81735.
avt77 added a comment.

I've updated the reciprocal related tests to see difference between old and new code gen more clearly. In fact there is no real difference but the new approach allows to take into account the schedule cost model when we deal with different machine code patterns. This patch should become the first step in the future similar optimizations like rsqrt, etc.


https://reviews.llvm.org/D26855

Files:
  include/llvm/CodeGen/MachineCombinerPattern.h
  include/llvm/CodeGen/MachineInstr.h
  lib/CodeGen/MachineCombiner.cpp
  lib/CodeGen/MachineInstr.cpp
  lib/CodeGen/SelectionDAG/DAGCombiner.cpp
  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: D26855.81735.patch
Type: text/x-patch
Size: 39570 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20161216/1b7346f5/attachment-0001.bin>