[PATCH] D95101: AMDGPU: Use more accurate fast f64 fdiv

[Jay Foad via Phabricator via llvm-commits]

foad added inline comments.


================
Comment at: llvm/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp:3115
+
+  auto Tmp0 = B.buildFMA(ResTy, NegY, R, One);
+  R = B.buildFMA(ResTy, Tmp0, R, R);
----------------
Can you add a comment here and/or in the sdag equivalent showing what the code you're building here will look like, and preferably where it came from and what kind of accuracy you expect from it?


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D95101/new/

https://reviews.llvm.org/D95101