[PATCH] D83981: [x86] split FMA with fast-math-flags to avoid libcall

[Sanjay Patel via Phabricator via llvm-commits]

spatel marked 2 inline comments as done.
spatel added inline comments.


================
Comment at: llvm/test/CodeGen/X86/fma.ll:1612
-; FMACALL32_BDVER2-NEXT:    addl $108, %esp ## encoding: [0x83,0xc4,0x6c]
-; FMACALL32_BDVER2-NEXT:    retl ## encoding: [0xc3]
   %call = call reassoc <2 x double> @llvm.fma.v2f64(<2 x double> %a, <2 x double> %b, <2 x double> %c)
----------------
cameron.mcinally wrote:
> These Piledriver checks disappeared. Was that deliberate?
They were subsumed into the common prefix of "FMACALL32". It's a bit strange that we don't have a unique prefix for this run line:
; RUN: llc < %s -mtriple=i386-apple-darwin10  -mattr=+avx,-fma,-fma4 -show-mc-encoding | FileCheck %s --check-prefix=FMACALL32

I can try to make that clearer with a cleanup patch.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D83981/new/

https://reviews.llvm.org/D83981