File: llvm.r600.dot4.ll

package info (click to toggle)
llvm-toolchain-20 1%3A20.1.6-1~exp1
  • links: PTS, VCS
  • area: main
  • in suites: experimental
  • size: 2,111,304 kB
  • sloc: cpp: 7,438,677; ansic: 1,393,822; asm: 1,012,926; python: 241,650; f90: 86,635; objc: 75,479; lisp: 42,144; pascal: 17,286; sh: 10,027; ml: 5,082; perl: 4,730; awk: 3,523; makefile: 3,349; javascript: 2,251; xml: 892; fortran: 672
file content (11 lines) | stat: -rw-r--r-- 526 bytes parent folder | download | duplicates (9) 
1
2
3
4
5
6
7
8
9
10
11
 
; RUN: llc -mtriple=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
}