[PATCH] D78893: [X86][CostModel] Improve costs for fp_to_uint/fp_to_sint for vXi8/vXi16/v2i32 results.

[Simon Pilgrim via Phabricator via llvm-commits]

RKSimon accepted this revision.
RKSimon added a comment.
This revision is now accepted and ready to land.

LGTM - cheers



================
Comment at: llvm/lib/Target/X86/X86TargetTransformInfo.cpp:1506
+    { ISD::FP_TO_SINT,  MVT::v16i8,  MVT::v16f32, 3 },
+    { ISD::FP_TO_SINT,  MVT::v16i16, MVT::v16f32, 3 },
+
----------------
This probably means the AVX512 i32->i16/i8 truncation costs above need bumping as well


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D78893/new/

https://reviews.llvm.org/D78893