1 ; RUN: opt -S -march=r600 -mcpu=cayman -loop-vectorize -force-vector-interleave=1 -force-vector-width=4 -dce -instcombine < %s | FileCheck %s 2 3 ; Artificial datalayout 4 target datalayout = "e-p:32:32:32-p1:16:16:16-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-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" 5 6 7 define void @add_ints_1_1_1(i32 addrspace(1)* %a, i32 addrspace(1)* %b, i32 addrspace(1)* %c) #0 { 8 ; CHECK-LABEL: @add_ints_1_1_1( 9 ; CHECK: <4 x i32> 10 ; CHECK: ret 11 12 entry: 13 br label %for.body 14 15 for.body: ; preds = %entry, %for.body 16 %i.01 = phi i64 [ 0, %entry ], [ %inc, %for.body ] 17 %arrayidx = getelementptr inbounds i32, i32 addrspace(1)* %b, i64 %i.01 18 %0 = load i32, i32 addrspace(1)* %arrayidx, align 4 19 %arrayidx1 = getelementptr inbounds i32, i32 addrspace(1)* %c, i64 %i.01 20 %1 = load i32, i32 addrspace(1)* %arrayidx1, align 4 21 %add = add nsw i32 %0, %1 22 %arrayidx2 = getelementptr inbounds i32, i32 addrspace(1)* %a, i64 %i.01 23 store i32 %add, i32 addrspace(1)* %arrayidx2, align 4 24 %inc = add i64 %i.01, 1 25 %cmp = icmp ult i64 %inc, 200 26 br i1 %cmp, label %for.body, label %for.end 27 28 for.end: ; preds = %for.body 29 ret void 30 } 31 32 define void @add_ints_as_1_0_0(i32 addrspace(1)* %a, i32* %b, i32* %c) #0 { 33 ; CHECK-LABEL: @add_ints_as_1_0_0( 34 ; CHECK-NOT: <4 x i32> 35 ; CHECK: ret 36 37 entry: 38 br label %for.body 39 40 for.body: ; preds = %entry, %for.body 41 %i.01 = phi i64 [ 0, %entry ], [ %inc, %for.body ] 42 %arrayidx = getelementptr inbounds i32, i32* %b, i64 %i.01 43 %0 = load i32, i32* %arrayidx, align 4 44 %arrayidx1 = getelementptr inbounds i32, i32* %c, i64 %i.01 45 %1 = load i32, i32* %arrayidx1, align 4 46 %add = add nsw i32 %0, %1 47 %arrayidx2 = getelementptr inbounds i32, i32 addrspace(1)* %a, i64 %i.01 48 store i32 %add, i32 addrspace(1)* %arrayidx2, align 4 49 %inc = add i64 %i.01, 1 50 %cmp = icmp ult i64 %inc, 200 51 br i1 %cmp, label %for.body, label %for.end 52 53 for.end: ; preds = %for.body 54 ret void 55 } 56 57 define void @add_ints_as_0_1_0(i32* %a, i32 addrspace(1)* %b, i32* %c) #0 { 58 ; CHECK-LABEL: @add_ints_as_0_1_0( 59 ; CHECK-NOT: <4 x i32> 60 ; CHECK: ret 61 62 entry: 63 br label %for.body 64 65 for.body: ; preds = %entry, %for.body 66 %i.01 = phi i64 [ 0, %entry ], [ %inc, %for.body ] 67 %arrayidx = getelementptr inbounds i32, i32 addrspace(1)* %b, i64 %i.01 68 %0 = load i32, i32 addrspace(1)* %arrayidx, align 4 69 %arrayidx1 = getelementptr inbounds i32, i32* %c, i64 %i.01 70 %1 = load i32, i32* %arrayidx1, align 4 71 %add = add nsw i32 %0, %1 72 %arrayidx2 = getelementptr inbounds i32, i32* %a, i64 %i.01 73 store i32 %add, i32* %arrayidx2, align 4 74 %inc = add i64 %i.01, 1 75 %cmp = icmp ult i64 %inc, 200 76 br i1 %cmp, label %for.body, label %for.end 77 78 for.end: ; preds = %for.body 79 ret void 80 } 81 82 define void @add_ints_as_0_1_1(i32* %a, i32 addrspace(1)* %b, i32 addrspace(1)* %c) #0 { 83 ; CHECK-LABEL: @add_ints_as_0_1_1( 84 ; CHECK-NOT: <4 x i32> 85 ; CHECK: ret 86 87 entry: 88 br label %for.body 89 90 for.body: ; preds = %entry, %for.body 91 %i.01 = phi i64 [ 0, %entry ], [ %inc, %for.body ] 92 %arrayidx = getelementptr inbounds i32, i32 addrspace(1)* %b, i64 %i.01 93 %0 = load i32, i32 addrspace(1)* %arrayidx, align 4 94 %arrayidx1 = getelementptr inbounds i32, i32 addrspace(1)* %c, i64 %i.01 95 %1 = load i32, i32 addrspace(1)* %arrayidx1, align 4 96 %add = add nsw i32 %0, %1 97 %arrayidx2 = getelementptr inbounds i32, i32* %a, i64 %i.01 98 store i32 %add, i32* %arrayidx2, align 4 99 %inc = add i64 %i.01, 1 100 %cmp = icmp ult i64 %inc, 200 101 br i1 %cmp, label %for.body, label %for.end 102 103 for.end: ; preds = %for.body 104 ret void 105 } 106 107 define void @add_ints_as_0_1_2(i32* %a, i32 addrspace(1)* %b, i32 addrspace(2)* %c) #0 { 108 ; CHECK-LABEL: @add_ints_as_0_1_2( 109 ; CHECK-NOT: <4 x i32> 110 ; CHECK: ret 111 112 entry: 113 br label %for.body 114 115 for.body: ; preds = %entry, %for.body 116 %i.01 = phi i64 [ 0, %entry ], [ %inc, %for.body ] 117 %arrayidx = getelementptr inbounds i32, i32 addrspace(1)* %b, i64 %i.01 118 %0 = load i32, i32 addrspace(1)* %arrayidx, align 4 119 %arrayidx1 = getelementptr inbounds i32, i32 addrspace(2)* %c, i64 %i.01 120 %1 = load i32, i32 addrspace(2)* %arrayidx1, align 4 121 %add = add nsw i32 %0, %1 122 %arrayidx2 = getelementptr inbounds i32, i32* %a, i64 %i.01 123 store i32 %add, i32* %arrayidx2, align 4 124 %inc = add i64 %i.01, 1 125 %cmp = icmp ult i64 %inc, 200 126 br i1 %cmp, label %for.body, label %for.end 127 128 for.end: ; preds = %for.body 129 ret void 130 } 131 132 attributes #0 = { nounwind uwtable "less-precise-fpmad"="false" "no-frame-pointer-elim"="true" "no-frame-pointer-elim-non-leaf" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "unsafe-fp-math"="false" "use-soft-float"="false" } 133
