[all-commits] [llvm/llvm-project] 28b7e4: AMDGPU/GFX12: Add new dot4 fp8/bf8 instructions (#...

[Mariusz Sikora via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 28b7e498b6a1dbfa1ac249acec45f948665ba58e
      https://github.com/llvm/llvm-project/commit/28b7e498b6a1dbfa1ac249acec45f948665ba58e
  Author: Mariusz Sikora <mariusz.sikora at amd.com>
  Date:   2024-01-18 (Thu, 18 Jan 2024)

  Changed paths:
    M clang/include/clang/Basic/BuiltinsAMDGPU.def
    M clang/test/CodeGenOpenCL/builtins-amdgcn-dl-insts-err.cl
    A clang/test/CodeGenOpenCL/builtins-amdgcn-dl-insts-gfx12.cl
    M llvm/include/llvm/IR/IntrinsicsAMDGPU.td
    M llvm/lib/Target/AMDGPU/AMDGPURegisterBankInfo.cpp
    M llvm/lib/Target/AMDGPU/AsmParser/AMDGPUAsmParser.cpp
    M llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPUInstPrinter.cpp
    M llvm/lib/Target/AMDGPU/VOP3PInstructions.td
    M llvm/lib/Target/AMDGPU/VOPInstructions.td
    A llvm/test/CodeGen/AMDGPU/llvm.amdgcn.dot4.f32.ll
    M llvm/test/MC/AMDGPU/gfx12_asm_vop3p.s
    M llvm/test/MC/AMDGPU/gfx12_asm_vop3p_dpp16.s
    A llvm/test/MC/AMDGPU/gfx12_asm_vop3p_dpp16_err.s
    M llvm/test/MC/AMDGPU/gfx12_asm_vop3p_dpp8.s
    A llvm/test/MC/AMDGPU/gfx12_asm_vop3p_dpp8_err.s
    A llvm/test/MC/AMDGPU/gfx12_asm_vop3p_err.s
    M llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop3p.txt
    M llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop3p_dpp16.txt
    M llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop3p_dpp8.txt

  Log Message:
  -----------
  AMDGPU/GFX12: Add new dot4 fp8/bf8 instructions (#77892)

Endoding is VOP3P. Tagged as deep/machine learning instructions. i32
type (v4fp8 or v4bf8 packed in i32) is used for src0 and src1. src0 and
src1 have no src_modifiers. src2 is f32 and has src_modifiers: f32
fneg(neg_lo[2]) and f32 fabs(neg_hi[2]).

---------

Co-authored-by: Petar Avramovic <Petar.Avramovic at amd.com>