Home | History | Annotate | Download | only in R600
      1 ; RUN: llc < %s -march=r600 -mcpu=verde -verify-machineinstrs | FileCheck %s
      2 
      3 ; CHECK: @test_8_min_char
      4 ; CHECK: BUFFER_STORE_BYTE
      5 ; CHECK: BUFFER_STORE_BYTE
      6 ; CHECK: BUFFER_STORE_BYTE
      7 ; CHECK: BUFFER_STORE_BYTE
      8 ; CHECK: BUFFER_STORE_BYTE
      9 ; CHECK: BUFFER_STORE_BYTE
     10 ; CHECK: BUFFER_STORE_BYTE
     11 ; CHECK: BUFFER_STORE_BYTE
     12 ; ModuleID = 'radeon'
     13 target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v16:16:16-v24:32:32-v32:32:32-v48:64:64-v64:64:64-v96:128:128-v128:128:128-v192:256:256-v256:256:256-v512:512:512-v1024:1024:1024-v2048:2048:2048-n32:64"
     14 target triple = "r600--"
     15 
     16 ; Function Attrs: nounwind
     17 define void @test_8_min_char(i8 addrspace(1)* nocapture %out, i8 addrspace(1)* nocapture readonly %in0, i8 addrspace(1)* nocapture readonly %in1) #0 {
     18 entry:
     19   %0 = load i8 addrspace(1)* %in0, align 1
     20   %1 = insertelement <8 x i8> undef, i8 %0, i32 0
     21   %arrayidx2.i.i = getelementptr inbounds i8 addrspace(1)* %in0, i64 1
     22   %2 = load i8 addrspace(1)* %arrayidx2.i.i, align 1
     23   %3 = insertelement <8 x i8> %1, i8 %2, i32 1
     24   %arrayidx6.i.i = getelementptr inbounds i8 addrspace(1)* %in0, i64 2
     25   %4 = load i8 addrspace(1)* %arrayidx6.i.i, align 1
     26   %5 = insertelement <8 x i8> %3, i8 %4, i32 2
     27   %arrayidx10.i.i = getelementptr inbounds i8 addrspace(1)* %in0, i64 3
     28   %6 = load i8 addrspace(1)* %arrayidx10.i.i, align 1
     29   %7 = insertelement <8 x i8> %5, i8 %6, i32 3
     30   %arrayidx.i.i = getelementptr inbounds i8 addrspace(1)* %in0, i64 4
     31   %8 = load i8 addrspace(1)* %arrayidx.i.i, align 1
     32   %9 = insertelement <8 x i8> undef, i8 %8, i32 0
     33   %arrayidx2.i9.i = getelementptr inbounds i8 addrspace(1)* %in0, i64 5
     34   %10 = load i8 addrspace(1)* %arrayidx2.i9.i, align 1
     35   %11 = insertelement <8 x i8> %9, i8 %10, i32 1
     36   %arrayidx6.i11.i = getelementptr inbounds i8 addrspace(1)* %in0, i64 6
     37   %12 = load i8 addrspace(1)* %arrayidx6.i11.i, align 1
     38   %13 = insertelement <8 x i8> %11, i8 %12, i32 2
     39   %arrayidx10.i13.i = getelementptr inbounds i8 addrspace(1)* %in0, i64 7
     40   %14 = load i8 addrspace(1)* %arrayidx10.i13.i, align 1
     41   %15 = insertelement <8 x i8> %13, i8 %14, i32 3
     42   %vecinit5.i = shufflevector <8 x i8> %7, <8 x i8> %15, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
     43   %16 = load i8 addrspace(1)* %in1, align 1
     44   %17 = insertelement <8 x i8> undef, i8 %16, i32 0
     45   %arrayidx2.i.i4 = getelementptr inbounds i8 addrspace(1)* %in1, i64 1
     46   %18 = load i8 addrspace(1)* %arrayidx2.i.i4, align 1
     47   %19 = insertelement <8 x i8> %17, i8 %18, i32 1
     48   %arrayidx6.i.i5 = getelementptr inbounds i8 addrspace(1)* %in1, i64 2
     49   %20 = load i8 addrspace(1)* %arrayidx6.i.i5, align 1
     50   %21 = insertelement <8 x i8> %19, i8 %20, i32 2
     51   %arrayidx10.i.i6 = getelementptr inbounds i8 addrspace(1)* %in1, i64 3
     52   %22 = load i8 addrspace(1)* %arrayidx10.i.i6, align 1
     53   %23 = insertelement <8 x i8> %21, i8 %22, i32 3
     54   %arrayidx.i.i7 = getelementptr inbounds i8 addrspace(1)* %in1, i64 4
     55   %24 = load i8 addrspace(1)* %arrayidx.i.i7, align 1
     56   %25 = insertelement <8 x i8> undef, i8 %24, i32 0
     57   %arrayidx2.i9.i8 = getelementptr inbounds i8 addrspace(1)* %in1, i64 5
     58   %26 = load i8 addrspace(1)* %arrayidx2.i9.i8, align 1
     59   %27 = insertelement <8 x i8> %25, i8 %26, i32 1
     60   %arrayidx6.i11.i9 = getelementptr inbounds i8 addrspace(1)* %in1, i64 6
     61   %28 = load i8 addrspace(1)* %arrayidx6.i11.i9, align 1
     62   %29 = insertelement <8 x i8> %27, i8 %28, i32 2
     63   %arrayidx10.i13.i10 = getelementptr inbounds i8 addrspace(1)* %in1, i64 7
     64   %30 = load i8 addrspace(1)* %arrayidx10.i13.i10, align 1
     65   %31 = insertelement <8 x i8> %29, i8 %30, i32 3
     66   %vecinit5.i11 = shufflevector <8 x i8> %23, <8 x i8> %31, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
     67   %cmp.i = icmp slt <8 x i8> %vecinit5.i, %vecinit5.i11
     68   %cond.i = select <8 x i1> %cmp.i, <8 x i8> %vecinit5.i, <8 x i8> %vecinit5.i11
     69   %32 = extractelement <8 x i8> %cond.i, i32 0
     70   store i8 %32, i8 addrspace(1)* %out, align 1
     71   %33 = extractelement <8 x i8> %cond.i, i32 1
     72   %arrayidx2.i.i.i = getelementptr inbounds i8 addrspace(1)* %out, i64 1
     73   store i8 %33, i8 addrspace(1)* %arrayidx2.i.i.i, align 1
     74   %34 = extractelement <8 x i8> %cond.i, i32 2
     75   %arrayidx.i.i.i = getelementptr inbounds i8 addrspace(1)* %out, i64 2
     76   store i8 %34, i8 addrspace(1)* %arrayidx.i.i.i, align 1
     77   %35 = extractelement <8 x i8> %cond.i, i32 3
     78   %arrayidx2.i6.i.i = getelementptr inbounds i8 addrspace(1)* %out, i64 3
     79   store i8 %35, i8 addrspace(1)* %arrayidx2.i6.i.i, align 1
     80   %arrayidx.i.i3 = getelementptr inbounds i8 addrspace(1)* %out, i64 4
     81   %36 = extractelement <8 x i8> %cond.i, i32 4
     82   store i8 %36, i8 addrspace(1)* %arrayidx.i.i3, align 1
     83   %37 = extractelement <8 x i8> %cond.i, i32 5
     84   %arrayidx2.i.i6.i = getelementptr inbounds i8 addrspace(1)* %out, i64 5
     85   store i8 %37, i8 addrspace(1)* %arrayidx2.i.i6.i, align 1
     86   %38 = extractelement <8 x i8> %cond.i, i32 6
     87   %arrayidx.i.i7.i = getelementptr inbounds i8 addrspace(1)* %out, i64 6
     88   store i8 %38, i8 addrspace(1)* %arrayidx.i.i7.i, align 1
     89   %39 = extractelement <8 x i8> %cond.i, i32 7
     90   %arrayidx2.i6.i8.i = getelementptr inbounds i8 addrspace(1)* %out, i64 7
     91   store i8 %39, i8 addrspace(1)* %arrayidx2.i6.i8.i, align 1
     92   ret void
     93 }
     94 
     95 attributes #0 = { nounwind "less-precise-fpmad"="false" "no-frame-pointer-elim"="false" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "no-realign-stack" "stack-protector-buffer-size"="8" "unsafe-fp-math"="false" "use-soft-float"="false" }
     96 
     97 !opencl.kernels = !{!0, !1, !2, !3, !4, !5, !6, !7, !8}
     98 
     99 !0 = metadata !{null}
    100 !1 = metadata !{null}
    101 !2 = metadata !{null}
    102 !3 = metadata !{void (i8 addrspace(1)*, i8 addrspace(1)*, i8 addrspace(1)*)* @test_8_min_char}
    103 !4 = metadata !{null}
    104 !5 = metadata !{null}
    105 !6 = metadata !{null}
    106 !7 = metadata !{null}
    107 !8 = metadata !{null}
    108