1 ; RUN: opt -mtriple=x86_64-unknown-linux-gnu -load-store-vectorizer -mcpu haswell -S -o - %s | FileCheck --check-prefix=CHECK-HSW %s
2 ; RUN: opt -mtriple=x86_64-unknown-linux-gnu -load-store-vectorizer -mcpu knl -S -o - %s | FileCheck --check-prefix=CHECK-KNL %s
3 ; RUN: opt -mtriple=x86_64-unknown-linux-gnu -aa-pipeline=basic-aa -passes='function(load-store-vectorizer)' -mcpu haswell -S -o - %s | FileCheck --check-prefix=CHECK-HSW %s
4 ; RUN: opt -mtriple=x86_64-unknown-linux-gnu -aa-pipeline=basic-aa -passes='function(load-store-vectorizer)' -mcpu knl -S -o - %s | FileCheck --check-prefix=CHECK-KNL %s
6 define <8 x double> @loadwidth_insert_extract(double* %ptr) {
7 %a = bitcast double* %ptr to <2 x double> *
8 %b = getelementptr <2 x double>, <2 x double>* %a, i32 1
9 %c = getelementptr <2 x double>, <2 x double>* %a, i32 2
10 %d = getelementptr <2 x double>, <2 x double>* %a, i32 3
11 ; CHECK-HSW: load <4 x double>
12 ; CHECK-HSW: load <4 x double>
14 ; CHECK-KNL: load <8 x double>
16 %la = load <2 x double>, <2 x double> *%a
17 %lb = load <2 x double>, <2 x double> *%b
18 %lc = load <2 x double>, <2 x double> *%c
19 %ld = load <2 x double>, <2 x double> *%d
20 ; Scalarize everything - Explicitly not a shufflevector to test this code
22 %v1 = extractelement <2 x double> %la, i32 0
23 %v2 = extractelement <2 x double> %la, i32 1
24 %v3 = extractelement <2 x double> %lb, i32 0
25 %v4 = extractelement <2 x double> %lb, i32 1
26 %v5 = extractelement <2 x double> %lc, i32 0
27 %v6 = extractelement <2 x double> %lc, i32 1
28 %v7 = extractelement <2 x double> %ld, i32 0
29 %v8 = extractelement <2 x double> %ld, i32 1
31 %i1 = insertelement <8 x double> undef, double %v1, i32 0
32 %i2 = insertelement <8 x double> %i1, double %v2, i32 1
33 %i3 = insertelement <8 x double> %i2, double %v3, i32 2
34 %i4 = insertelement <8 x double> %i3, double %v4, i32 3
35 %i5 = insertelement <8 x double> %i4, double %v5, i32 4
36 %i6 = insertelement <8 x double> %i5, double %v6, i32 5
37 %i7 = insertelement <8 x double> %i6, double %v7, i32 6
38 %i8 = insertelement <8 x double> %i7, double %v8, i32 7