[all-commits] [llvm/llvm-project] 5b156a: [AMDGPU] Add half vector support for table-driven ...

[Steffen Larsen via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 5b156a4372acb1c8dcc837211ede7216d1454de3
      https://github.com/llvm/llvm-project/commit/5b156a4372acb1c8dcc837211ede7216d1454de3
  Author: Steffen Larsen <sholstla at amd.com>
  Date:   2026-03-04 (Wed, 04 Mar 2026)

  Changed paths:
    M llvm/lib/Target/AMDGPU/AMDGPULibCalls.cpp
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-acos.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-acosh.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-acospi.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-asin.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-asinh.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-asinpi.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-atan.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-atanh.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-atanpi.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-cbrt.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-cos.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-cosh.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-cospi.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-erf.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-erfc.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-exp.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-exp10.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-exp2.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-expm1.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-log.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-log10.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-log2.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-rsqrt.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-sin.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-sinh.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-sinpi.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-sqrt.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-tan.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-tanh.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-tanpi.ll
    A llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-tdo-tgamma.ll

  Log Message:
  -----------
  [AMDGPU] Add half vector support for table-driven libcall optimization (#178638)

When replacing certain AMDGPU library calls with constant data vectors,
the existing implementation only handled single and double-precision
floats. This change extends the functionality to also support
half-precision floats.

Additionally, it refactors the function responsible for generating
constant float data vectors to improve readability and reduces code
duplication. In tandem with this refactoring, the patch relaxes the
check for constant data vectors to include any constant of vector type.
This allows other constant vectors to be processed, such as those
created from constant aggregate zeros (e.g. `<2 x float>
zeroinitializer`).

---------

Signed-off-by: Steffen Holst Larsen <sholstla at amd.com>



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