1 ; RUN: opt -S -basicaa -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine < %s | FileCheck %s
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-macosx10.8.0"
6 @uf = common global [100 x i32] zeroinitializer, align 16
7 @xi = common global [100 x i32] zeroinitializer, align 16
8 @q = common global [100 x i32] zeroinitializer, align 16
13 ; Due to a bug in the way we handled reverse induction stores we would generate
22 ; CHECK: [[VAR1:%[a-zA-Z0-9.]+]] = load <4 x i32>
23 ; CHECK: [[VAR2:%[a-zA-Z0-9.]+]] = load <4 x i32>
24 ; CHECK: [[VAR3:%[a-zA-Z0-9]+]] = add nsw <4 x i32> [[VAR2]], [[VAR1]]
25 ; CHECK: store <4 x i32> [[VAR3]]
26 ; CHECK: [[VAR4:%[a-zA-Z0-9.]+]] = load <4 x i32>
27 ; CHECK: add nsw <4 x i32> [[VAR3]], [[VAR4]]
28 ; CHECK-NOT: shufflevector
31 %indvars.iv = phi i64 [ 93, %entry ], [ %indvars.iv.next, %for.body ]
32 %0 = add i64 %indvars.iv, 1
33 %arrayidx = getelementptr inbounds [100 x i32]* @uf, i64 0, i64 %0
34 %arrayidx3 = getelementptr inbounds [100 x i32]* @xi, i64 0, i64 %0
35 %1 = load i32* %arrayidx3, align 4
36 %2 = load i32* %arrayidx, align 4
37 %add4 = add nsw i32 %2, %1
38 store i32 %add4, i32* %arrayidx, align 4
39 %arrayidx7 = getelementptr inbounds [100 x i32]* @q, i64 0, i64 %0
40 %3 = load i32* %arrayidx7, align 4
41 %add8 = add nsw i32 %add4, %3
42 store i32 %add8, i32* %arrayidx, align 4
43 %indvars.iv.next = add i64 %indvars.iv, -1
44 %4 = trunc i64 %indvars.iv.next to i32
45 %cmp = icmp ugt i32 %4, 2
46 br i1 %cmp, label %for.body, label %for.end