1 ; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mcpu=corei7-avx | 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 define i32 @foo(i32* noalias nocapture %A, i32* noalias nocapture %B, i32 %start, i32 %end) nounwind uwtable ssp {
8 ;CHECK: cost of 1 {{.*}} icmp
9 %cmp7 = icmp slt i32 %start, %end
10 br i1 %cmp7, label %for.body.lr.ph, label %for.end
12 for.body.lr.ph: ; preds = %entry
13 ;CHECK: cost of 1 {{.*}} sext
14 %0 = sext i32 %start to i64
15 %1 = sub i32 %end, %start
16 %2 = zext i32 %1 to i64
17 %end.idx = add i64 %2, %0
18 ;CHECK: cost of 1 {{.*}} add
19 %n.vec = and i64 %2, 4294967288
20 %end.idx.rnd.down = add i64 %n.vec, %0
21 ;CHECK: cost of 1 {{.*}} icmp
22 %cmp.zero = icmp eq i64 %n.vec, 0
23 br i1 %cmp.zero, label %middle.block, label %vector.body
25 vector.body: ; preds = %for.body.lr.ph, %vector.body
26 %index = phi i64 [ %index.next, %vector.body ], [ %0, %for.body.lr.ph ]
27 %3 = add i64 %index, 2
28 %4 = getelementptr inbounds i32, i32* %B, i64 %3
29 ;CHECK: cost of 0 {{.*}} bitcast
30 %5 = bitcast i32* %4 to <8 x i32>*
31 ;CHECK: cost of 2 {{.*}} load
32 %6 = load <8 x i32>, <8 x i32>* %5, align 4
33 ;CHECK: cost of 4 {{.*}} mul
34 %7 = mul nsw <8 x i32> %6, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>
35 %8 = getelementptr inbounds i32, i32* %A, i64 %index
36 %9 = bitcast i32* %8 to <8 x i32>*
37 ;CHECK: cost of 2 {{.*}} load
38 %10 = load <8 x i32>, <8 x i32>* %9, align 4
39 ;CHECK: cost of 4 {{.*}} add
40 %11 = add nsw <8 x i32> %10, %7
41 ;CHECK: cost of 2 {{.*}} store
42 store <8 x i32> %11, <8 x i32>* %9, align 4
43 %index.next = add i64 %index, 8
44 %12 = icmp eq i64 %index.next, %end.idx.rnd.down
45 ;CHECK: cost of 0 {{.*}} br
46 br i1 %12, label %middle.block, label %vector.body
48 middle.block: ; preds = %vector.body, %for.body.lr.ph
49 %cmp.n = icmp eq i64 %end.idx, %end.idx.rnd.down
50 br i1 %cmp.n, label %for.end, label %for.body
52 for.body: ; preds = %middle.block, %for.body
53 %indvars.iv = phi i64 [ %indvars.iv.next, %for.body ], [ %end.idx.rnd.down, %middle.block ]
54 %13 = add nsw i64 %indvars.iv, 2
55 %arrayidx = getelementptr inbounds i32, i32* %B, i64 %13
56 ;CHECK: cost of 1 {{.*}} load
57 %14 = load i32, i32* %arrayidx, align 4
58 ;CHECK: cost of 1 {{.*}} mul
59 %mul = mul nsw i32 %14, 5
60 %arrayidx2 = getelementptr inbounds i32, i32* %A, i64 %indvars.iv
61 ;CHECK: cost of 1 {{.*}} load
62 %15 = load i32, i32* %arrayidx2, align 4
63 %add3 = add nsw i32 %15, %mul
64 store i32 %add3, i32* %arrayidx2, align 4
65 %indvars.iv.next = add i64 %indvars.iv, 1
66 ;CHECK: cost of 0 {{.*}} trunc
67 %16 = trunc i64 %indvars.iv.next to i32
68 %cmp = icmp slt i32 %16, %end
69 ;CHECK: cost of 0 {{.*}} br
70 br i1 %cmp, label %for.body, label %for.end
72 for.end: ; preds = %middle.block, %for.body, %entry
73 ;CHECK: cost of 0 {{.*}} ret