[PATCH] D152809: [AMDGPU][GFX11] Add test coverage for 16-bit conversions, part 8.

[Jay Foad via Phabricator via llvm-commits]

foad added inline comments.


================
Comment at: llvm/test/CodeGen/AMDGPU/v_madak_f16.ll:74
+; GFX11-NEXT:    s_waitcnt vmcnt(0)
+; GFX11-NEXT:    v_mul_f16_e32 v0, v0, v1
+; GFX11-NEXT:    s_delay_alu instid0(VALU_DEP_1)
----------------
Joe_Nash wrote:
> Nit: Why aren't we using v_fmaak or v_fmamk here?
It's not allowed to fuse mul+add->fma by default, without "contract" or some other fast math flags.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D152809/new/

https://reviews.llvm.org/D152809