1 ; RUN: llc < %s -mcpu=corei7 -march=x86 -mattr=+sse4.1 | FileCheck %s
3 define void @t1(float* %R, <4 x float>* %P1) nounwind {
6 ; CHECK-NEXT: movl {{[0-9]+}}(%esp), %eax
7 ; CHECK-NEXT: movl {{[0-9]+}}(%esp), %ecx
8 ; CHECK-NEXT: movss 12(%ecx), %xmm0
9 ; CHECK-NEXT: movss %xmm0, (%eax)
12 %X = load <4 x float>, <4 x float>* %P1
13 %tmp = extractelement <4 x float> %X, i32 3
14 store float %tmp, float* %R
18 define float @t2(<4 x float>* %P1) nounwind {
21 ; CHECK-NEXT: pushl %eax
22 ; CHECK-NEXT: movl {{[0-9]+}}(%esp), %eax
23 ; CHECK-NEXT: movapd (%eax), %xmm0
24 ; CHECK-NEXT: shufpd {{.*#+}} xmm0 = xmm0[1,0]
25 ; CHECK-NEXT: movss %xmm0, (%esp)
26 ; CHECK-NEXT: flds (%esp)
27 ; CHECK-NEXT: popl %eax
30 %X = load <4 x float>, <4 x float>* %P1
31 %tmp = extractelement <4 x float> %X, i32 2
35 define void @t3(i32* %R, <4 x i32>* %P1) nounwind {
38 ; CHECK-NEXT: movl {{[0-9]+}}(%esp), %eax
39 ; CHECK-NEXT: movl {{[0-9]+}}(%esp), %ecx
40 ; CHECK-NEXT: movl 12(%ecx), %ecx
41 ; CHECK-NEXT: movl %ecx, (%eax)
44 %X = load <4 x i32>, <4 x i32>* %P1
45 %tmp = extractelement <4 x i32> %X, i32 3
46 store i32 %tmp, i32* %R
50 define i32 @t4(<4 x i32>* %P1) nounwind {
53 ; CHECK-NEXT: movl {{[0-9]+}}(%esp), %eax
54 ; CHECK-NEXT: movl 12(%eax), %eax
57 %X = load <4 x i32>, <4 x i32>* %P1
58 %tmp = extractelement <4 x i32> %X, i32 3