1; RUN: opt < %s  -loop-vectorize -force-vector-width=1 -force-target-num-scalar-regs=16 -force-target-max-scalar-interleave=8 -force-target-instruction-cost=1 -small-loop-cost=40 -dce -instcombine -S | FileCheck %s
2
3target 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
5@a = common global [2048 x i32] zeroinitializer, align 16
6
7; This is the loop.
8;  for (i=0; i<n; i++){
9;    a[i] += i;
10;  }
11;CHECK-LABEL: @inc(
12;CHECK: load i32, i32*
13;CHECK: load i32, i32*
14;CHECK: load i32, i32*
15;CHECK: load i32, i32*
16;CHECK-NOT: load i32, i32*
17;CHECK: add nsw i32
18;CHECK: add nsw i32
19;CHECK: add nsw i32
20;CHECK: add nsw i32
21;CHECK-NOT: add nsw i32
22;CHECK: store i32
23;CHECK: store i32
24;CHECK: store i32
25;CHECK: store i32
26;CHECK-NOT: store i32
27;CHECK: add i64 %{{.*}}, 4
28;CHECK: ret void
29define void @inc(i32 %n) nounwind uwtable noinline ssp {
30  %1 = icmp sgt i32 %n, 0
31  br i1 %1, label %.lr.ph, label %._crit_edge
32
33.lr.ph:                                           ; preds = %0, %.lr.ph
34  %indvars.iv = phi i64 [ %indvars.iv.next, %.lr.ph ], [ 0, %0 ]
35  %2 = getelementptr inbounds [2048 x i32], [2048 x i32]* @a, i64 0, i64 %indvars.iv
36  %3 = load i32, i32* %2, align 4
37  %4 = trunc i64 %indvars.iv to i32
38  %5 = add nsw i32 %3, %4
39  store i32 %5, i32* %2, align 4
40  %indvars.iv.next = add i64 %indvars.iv, 1
41  %lftr.wideiv = trunc i64 %indvars.iv.next to i32
42  %exitcond = icmp eq i32 %lftr.wideiv, %n
43  br i1 %exitcond, label %._crit_edge, label %.lr.ph
44
45._crit_edge:                                      ; preds = %.lr.ph, %0
46  ret void
47}
48
49