1; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
2; RUN: opt < %s -basic-aa -slp-vectorizer -dce -S -mtriple=x86_64-apple-macosx10.8.0 -mcpu=corei7-avx | FileCheck %s
3
4target 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"
5target triple = "x86_64-apple-macosx10.8.0"
6
7; At this point we can't vectorize only parts of the tree.
8
9define i32 @test(double* nocapture %A, i8* nocapture %B) {
10; CHECK-LABEL: @test(
11; CHECK-NEXT:  entry:
12; CHECK-NEXT:    [[TMP0:%.*]] = bitcast i8* [[B:%.*]] to <2 x i8>*
13; CHECK-NEXT:    [[TMP1:%.*]] = load <2 x i8>, <2 x i8>* [[TMP0]], align 1
14; CHECK-NEXT:    [[TMP2:%.*]] = add <2 x i8> [[TMP1]], <i8 3, i8 3>
15; CHECK-NEXT:    [[TMP3:%.*]] = sitofp <2 x i8> [[TMP2]] to <2 x double>
16; CHECK-NEXT:    [[TMP4:%.*]] = fmul <2 x double> [[TMP3]], [[TMP3]]
17; CHECK-NEXT:    [[TMP5:%.*]] = fadd <2 x double> [[TMP4]], <double 1.000000e+00, double 1.000000e+00>
18; CHECK-NEXT:    [[TMP6:%.*]] = fmul <2 x double> [[TMP5]], [[TMP5]]
19; CHECK-NEXT:    [[TMP7:%.*]] = fadd <2 x double> [[TMP6]], <double 1.000000e+00, double 1.000000e+00>
20; CHECK-NEXT:    [[TMP8:%.*]] = fmul <2 x double> [[TMP7]], [[TMP7]]
21; CHECK-NEXT:    [[TMP9:%.*]] = fadd <2 x double> [[TMP8]], <double 1.000000e+00, double 1.000000e+00>
22; CHECK-NEXT:    [[TMP10:%.*]] = fmul <2 x double> [[TMP9]], [[TMP9]]
23; CHECK-NEXT:    [[TMP11:%.*]] = fadd <2 x double> [[TMP10]], <double 1.000000e+00, double 1.000000e+00>
24; CHECK-NEXT:    [[TMP12:%.*]] = fmul <2 x double> [[TMP11]], [[TMP11]]
25; CHECK-NEXT:    [[TMP13:%.*]] = fadd <2 x double> [[TMP12]], <double 1.000000e+00, double 1.000000e+00>
26; CHECK-NEXT:    [[TMP14:%.*]] = bitcast double* [[A:%.*]] to <2 x double>*
27; CHECK-NEXT:    store <2 x double> [[TMP13]], <2 x double>* [[TMP14]], align 8
28; CHECK-NEXT:    ret i32 undef
29;
30entry:
31  %0 = load i8, i8* %B, align 1
32  %arrayidx1 = getelementptr inbounds i8, i8* %B, i64 1
33  %1 = load i8, i8* %arrayidx1, align 1
34  %add = add i8 %0, 3
35  %add4 = add i8 %1, 3
36  %conv6 = sitofp i8 %add to double
37  %conv7 = sitofp i8 %add4 to double
38  %mul = fmul double %conv6, %conv6
39  %add8 = fadd double %mul, 1.000000e+00
40  %mul9 = fmul double %conv7, %conv7
41  %add10 = fadd double %mul9, 1.000000e+00
42  %mul11 = fmul double %add8, %add8
43  %add12 = fadd double %mul11, 1.000000e+00
44  %mul13 = fmul double %add10, %add10
45  %add14 = fadd double %mul13, 1.000000e+00
46  %mul15 = fmul double %add12, %add12
47  %add16 = fadd double %mul15, 1.000000e+00
48  %mul17 = fmul double %add14, %add14
49  %add18 = fadd double %mul17, 1.000000e+00
50  %mul19 = fmul double %add16, %add16
51  %add20 = fadd double %mul19, 1.000000e+00
52  %mul21 = fmul double %add18, %add18
53  %add22 = fadd double %mul21, 1.000000e+00
54  %mul23 = fmul double %add20, %add20
55  %add24 = fadd double %mul23, 1.000000e+00
56  %mul25 = fmul double %add22, %add22
57  %add26 = fadd double %mul25, 1.000000e+00
58  store double %add24, double* %A, align 8
59  %arrayidx28 = getelementptr inbounds double, double* %A, i64 1
60  store double %add26, double* %arrayidx28, align 8
61  ret i32 undef
62}
63