Home | History | Annotate | Download | only in AMDGPU
      1 ; RUN: llc -march=r600 -mcpu=redwood -mtriple=r600-- < %s | FileCheck %s
      2 
      3 ; We want all MULLO_INT inst to be last in their instruction group
      4 ;CHECK: {{^}}fill3d:
      5 ;CHECK-NOT: MULLO_INT T[0-9]+
      6 
      7 define void @fill3d(i32 addrspace(1)* nocapture %out) #0 {
      8 entry:
      9   %x.i = tail call i32 @llvm.r600.read.global.size.x() #1
     10   %y.i18 = tail call i32 @llvm.r600.read.global.size.y() #1
     11   %mul = mul i32 %y.i18, %x.i
     12   %z.i17 = tail call i32 @llvm.r600.read.global.size.z() #1
     13   %mul3 = mul i32 %mul, %z.i17
     14   %x.i.i = tail call i32 @llvm.r600.read.tgid.x() #1
     15   %x.i12.i = tail call i32 @llvm.r600.read.local.size.x() #1
     16   %mul26.i = mul i32 %x.i12.i, %x.i.i
     17   %x.i4.i = tail call i32 @llvm.r600.read.tidig.x() #1
     18   %add.i16 = add i32 %x.i4.i, %mul26.i
     19   %mul7 = mul i32 %add.i16, %y.i18
     20   %y.i.i = tail call i32 @llvm.r600.read.tgid.y() #1
     21   %y.i14.i = tail call i32 @llvm.r600.read.local.size.y() #1
     22   %mul30.i = mul i32 %y.i14.i, %y.i.i
     23   %y.i6.i = tail call i32 @llvm.r600.read.tidig.y() #1
     24   %add.i14 = add i32 %mul30.i, %mul7
     25   %mul819 = add i32 %add.i14, %y.i6.i
     26   %add = mul i32 %mul819, %z.i17
     27   %z.i.i = tail call i32 @llvm.r600.read.tgid.z() #1
     28   %z.i16.i = tail call i32 @llvm.r600.read.local.size.z() #1
     29   %mul33.i = mul i32 %z.i16.i, %z.i.i
     30   %z.i8.i = tail call i32 @llvm.r600.read.tidig.z() #1
     31   %add.i = add i32 %z.i8.i, %mul33.i
     32   %add13 = add i32 %add.i, %add
     33   %arrayidx = getelementptr inbounds i32, i32 addrspace(1)* %out, i32 %add13
     34   store i32 %mul3, i32 addrspace(1)* %arrayidx, align 4
     35   ret void
     36 }
     37 
     38 ; Function Attrs: nounwind readnone
     39 declare i32 @llvm.r600.read.tgid.x() #1
     40 
     41 ; Function Attrs: nounwind readnone
     42 declare i32 @llvm.r600.read.tgid.y() #1
     43 
     44 ; Function Attrs: nounwind readnone
     45 declare i32 @llvm.r600.read.tgid.z() #1
     46 
     47 ; Function Attrs: nounwind readnone
     48 declare i32 @llvm.r600.read.local.size.x() #1
     49 
     50 ; Function Attrs: nounwind readnone
     51 declare i32 @llvm.r600.read.local.size.y() #1
     52 
     53 ; Function Attrs: nounwind readnone
     54 declare i32 @llvm.r600.read.local.size.z() #1
     55 
     56 ; Function Attrs: nounwind readnone
     57 declare i32 @llvm.r600.read.tidig.x() #1
     58 
     59 ; Function Attrs: nounwind readnone
     60 declare i32 @llvm.r600.read.tidig.y() #1
     61 
     62 ; Function Attrs: nounwind readnone
     63 declare i32 @llvm.r600.read.tidig.z() #1
     64 
     65 ; Function Attrs: nounwind readnone
     66 declare i32 @llvm.r600.read.global.size.x() #1
     67 
     68 ; Function Attrs: nounwind readnone
     69 declare i32 @llvm.r600.read.global.size.y() #1
     70 
     71 ; Function Attrs: nounwind readnone
     72 declare i32 @llvm.r600.read.global.size.z() #1
     73 
     74 attributes #0 = { nounwind "less-precise-fpmad"="false" "no-frame-pointer-elim"="false" "no-frame-pointer-elim-non-leaf"="false" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "unsafe-fp-math"="false" "use-soft-float"="false" }
     75 attributes #1 = { nounwind readnone }
     76 
     77 !opencl.kernels = !{!0, !1, !2}
     78 
     79 !0 = !{null}
     80 !1 = !{null}
     81 !2 = !{void (i32 addrspace(1)*)* @fill3d}
     82