1 ; RUN: opt < %s -basicaa -slp-vectorizer -dce -S -mtriple=x86_64-apple-macosx10.8.0 -mcpu=corei7-avx | FileCheck %s 2 3 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-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128" 4 target triple = "x86_64-apple-macosx10.9.0" 5 6 ; int foo(int *A) { 7 ; int r = A[0], g = A[1], b = A[2], a = A[3]; 8 ; for (int i=0; i < A[13]; i++) { 9 ; r*=18; g*=19; b*=12; a *=9; 10 ; } 11 ; A[0] = r; A[1] = g; A[2] = b; A[3] = a; 12 ; } 13 14 ;CHECK-LABEL: @foo 15 ;CHECK: bitcast i32* %A to <4 x i32>* 16 ;CHECK-NEXT: load <4 x i32> 17 ;CHECK: phi <4 x i32> 18 ;CHECK-NEXT: mul <4 x i32> 19 ;CHECK-NOT: mul 20 ;CHECK: phi <4 x i32> 21 ;CHECK: bitcast i32* %A to <4 x i32>* 22 ;CHECK-NEXT: store <4 x i32> 23 ;CHECK-NEXT:ret i32 undef 24 define i32 @foo(i32* nocapture %A) #0 { 25 entry: 26 %0 = load i32* %A, align 4 27 %arrayidx1 = getelementptr inbounds i32* %A, i64 1 28 %1 = load i32* %arrayidx1, align 4 29 %arrayidx2 = getelementptr inbounds i32* %A, i64 2 30 %2 = load i32* %arrayidx2, align 4 31 %arrayidx3 = getelementptr inbounds i32* %A, i64 3 32 %3 = load i32* %arrayidx3, align 4 33 %arrayidx4 = getelementptr inbounds i32* %A, i64 13 34 %4 = load i32* %arrayidx4, align 4 35 %cmp24 = icmp sgt i32 %4, 0 36 br i1 %cmp24, label %for.body, label %for.end 37 38 for.body: ; preds = %entry, %for.body 39 %i.029 = phi i32 [ %inc, %for.body ], [ 0, %entry ] 40 %a.028 = phi i32 [ %mul7, %for.body ], [ %3, %entry ] 41 %b.027 = phi i32 [ %mul6, %for.body ], [ %2, %entry ] 42 %g.026 = phi i32 [ %mul5, %for.body ], [ %1, %entry ] 43 %r.025 = phi i32 [ %mul, %for.body ], [ %0, %entry ] 44 %mul = mul nsw i32 %r.025, 18 45 %mul5 = mul nsw i32 %g.026, 19 46 %mul6 = mul nsw i32 %b.027, 12 47 %mul7 = mul nsw i32 %a.028, 9 48 %inc = add nsw i32 %i.029, 1 49 %cmp = icmp slt i32 %inc, %4 50 br i1 %cmp, label %for.body, label %for.end 51 52 for.end: ; preds = %for.body, %entry 53 %a.0.lcssa = phi i32 [ %3, %entry ], [ %mul7, %for.body ] 54 %b.0.lcssa = phi i32 [ %2, %entry ], [ %mul6, %for.body ] 55 %g.0.lcssa = phi i32 [ %1, %entry ], [ %mul5, %for.body ] 56 %r.0.lcssa = phi i32 [ %0, %entry ], [ %mul, %for.body ] 57 store i32 %r.0.lcssa, i32* %A, align 4 58 store i32 %g.0.lcssa, i32* %arrayidx1, align 4 59 store i32 %b.0.lcssa, i32* %arrayidx2, align 4 60 store i32 %a.0.lcssa, i32* %arrayidx3, align 4 61 ret i32 undef 62 } 63 64 65
