[all-commits] [llvm/llvm-project] b6f32a: [NVPTX] Switch to untyped float registers (#137011)

[Alex MacLean via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: b6f32ad8b03dccaba0db7ded9d561ee83e4530ab
      https://github.com/llvm/llvm-project/commit/b6f32ad8b03dccaba0db7ded9d561ee83e4530ab
  Author: Alex MacLean <amaclean at nvidia.com>
  Date:   2025-04-23 (Wed, 23 Apr 2025)

  Changed paths:
    M llvm/lib/Target/NVPTX/NVPTXRegisterInfo.cpp
    M llvm/test/CodeGen/NVPTX/LoadStoreVectorizer.ll
    M llvm/test/CodeGen/NVPTX/and-or-setcc.ll
    M llvm/test/CodeGen/NVPTX/atomics.ll
    M llvm/test/CodeGen/NVPTX/bf16-instructions.ll
    M llvm/test/CodeGen/NVPTX/bf16x2-instructions-approx.ll
    M llvm/test/CodeGen/NVPTX/bf16x2-instructions.ll
    M llvm/test/CodeGen/NVPTX/convert-fp-i8.ll
    M llvm/test/CodeGen/NVPTX/convert-sm100.ll
    M llvm/test/CodeGen/NVPTX/convert-sm100a.ll
    M llvm/test/CodeGen/NVPTX/convert-sm80.ll
    M llvm/test/CodeGen/NVPTX/convert-sm90.ll
    M llvm/test/CodeGen/NVPTX/copysign.ll
    M llvm/test/CodeGen/NVPTX/distributed-shared-cluster.ll
    M llvm/test/CodeGen/NVPTX/div.ll
    M llvm/test/CodeGen/NVPTX/f16-abs.ll
    M llvm/test/CodeGen/NVPTX/f16x2-instructions.ll
    M llvm/test/CodeGen/NVPTX/f32-ex2.ll
    M llvm/test/CodeGen/NVPTX/f32-lg2.ll
    M llvm/test/CodeGen/NVPTX/fabs-intrinsics.ll
    M llvm/test/CodeGen/NVPTX/fexp2.ll
    M llvm/test/CodeGen/NVPTX/flog2.ll
    M llvm/test/CodeGen/NVPTX/fma-relu-contract.ll
    M llvm/test/CodeGen/NVPTX/fma-relu-fma-intrinsic.ll
    M llvm/test/CodeGen/NVPTX/fma-relu-instruction-flag.ll
    M llvm/test/CodeGen/NVPTX/fp-contract.ll
    M llvm/test/CodeGen/NVPTX/frem.ll
    M llvm/test/CodeGen/NVPTX/i8x4-instructions.ll
    M llvm/test/CodeGen/NVPTX/intrinsics.ll
    M llvm/test/CodeGen/NVPTX/ldg-invariant.ll
    M llvm/test/CodeGen/NVPTX/ldu-ldg.ll
    M llvm/test/CodeGen/NVPTX/load-store-scalars.ll
    M llvm/test/CodeGen/NVPTX/load-store-vectors.ll
    M llvm/test/CodeGen/NVPTX/math-intrins.ll
    M llvm/test/CodeGen/NVPTX/misched_func_call.ll
    M llvm/test/CodeGen/NVPTX/param-add.ll
    M llvm/test/CodeGen/NVPTX/rcp-opt.ll
    M llvm/test/CodeGen/NVPTX/reduction-intrinsics.ll
    M llvm/test/CodeGen/NVPTX/redux-sync-f32.ll
    M llvm/test/CodeGen/NVPTX/reg-types.ll
    M llvm/test/CodeGen/NVPTX/st-param-imm.ll
    M llvm/test/CodeGen/NVPTX/surf-read-cuda.ll
    M llvm/test/CodeGen/NVPTX/tex-read-cuda.ll
    M llvm/test/CodeGen/NVPTX/variadics-backend.ll
    M llvm/test/DebugInfo/NVPTX/debug-info.ll

  Log Message:
  -----------
  [NVPTX] Switch to untyped float registers (#137011)

Register types in PTX are simply syntactic sugar and emitting them has
added lots of unnecessary complexity to the NVPTX backend. This change
takes the first step to their removal by using ".b" registers instead of
".f" in all cases. This should shake out any potential issues or bugs in
ptxas preventing full removal and pre-fetches many of the required test
updates.



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