[all-commits] [llvm/llvm-project] fca41f: [X86] Replace BF16 to F32 conversions with generic...

[Simon Pilgrim via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: fca41f4aa105f30af75d88d993539d043ac66460
      https://github.com/llvm/llvm-project/commit/fca41f4aa105f30af75d88d993539d043ac66460
  Author: Simon Pilgrim <llvm-dev at redking.me.uk>
  Date:   2025-11-27 (Thu, 27 Nov 2025)

  Changed paths:
    M clang/include/clang/Basic/BuiltinsX86.td
    M clang/lib/CodeGen/TargetBuiltins/X86.cpp
    M clang/lib/Headers/avx512bf16intrin.h
    M clang/lib/Headers/avx512vlbf16intrin.h
    M clang/test/CodeGen/X86/avx512bf16-builtins.c
    M clang/test/CodeGen/X86/avx512vlbf16-builtins.c

  Log Message:
  -----------
  [X86] Replace BF16 to F32 conversions with generic conversions (#169781)

Let standard casting / builtin_convertvector handle the conversions from BF16 to F32

My only query is how to best implement _mm_cvtpbh_ps - I went for the
v8bf16 -> v8f32 conversion followed by subvector extraction in the end,
but could just as easily extract a v4bf16 first - makes no difference to
final optimized codegen.

First part of #154911



To unsubscribe from these emails, change your notification settings at https://github.com/llvm/llvm-project/settings/notifications