File: llvm.r600.dot4.ll

package info (click to toggle)
llvm-toolchain-16 1%3A16.0.6-15~deb12u1
  • links: PTS, VCS
  • area: main
  • in suites: bookworm
  • size: 1,634,792 kB
  • sloc: cpp: 6,179,261; ansic: 1,216,205; asm: 741,319; python: 196,614; objc: 75,325; f90: 49,640; lisp: 32,396; pascal: 12,286; sh: 9,394; perl: 7,442; ml: 5,494; awk: 3,523; makefile: 2,723; javascript: 1,206; xml: 886; fortran: 581; cs: 573
file content (11 lines) | stat: -rw-r--r-- 524 bytes parent folder | download | duplicates (4)
1
2
3
4
5
6
7
8
9
10
11
; RUN: llc -march=r600 -mcpu=redwood -verify-machineinstrs < %s

declare float @llvm.r600.dot4(<4 x float>, <4 x float>) nounwind readnone

define amdgpu_kernel void @test_dp4(ptr addrspace(1) %out, ptr addrspace(1) %a, ptr addrspace(1) %b) nounwind {
  %src0 = load <4 x float>, ptr addrspace(1) %a, align 16
  %src1 = load <4 x float>, ptr addrspace(1) %b, align 16
  %dp4 = call float @llvm.r600.dot4(<4 x float> %src0, <4 x float> %src1) nounwind readnone
  store float %dp4, ptr addrspace(1) %out, align 4
  ret void
}