[PATCH] D88683: [SVE] Lower fixed length vector fneg and fsqrt operations.

[Cameron McInally via Phabricator via llvm-commits]

cameron.mcinally added inline comments.


================
Comment at: llvm/test/CodeGen/AArch64/sve-fp.ll:333
+  %res = fneg <vscale x 8 x half> %a
+  ret <vscale x 8 x half> %res
 }
----------------
cameron.mcinally wrote:
> Nit: These changes should probably go in with a standalone patch, but seem harmless enough to be included with this one.
> 
> @arsenm Also, there's an ongoing project to stop FMUL(-1.0, X)->FNEG(X) transformations when denormals are present, so this is a good change.
Just noticed a typo here:

>Also, there's an ongoing project to stop FMUL(-1.0, X)->FNEG(X) transformations when denormals are present, so this is a good change.

Should say "when denormals are flushed".


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D88683/new/

https://reviews.llvm.org/D88683