1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown | FileCheck %s
4 define void @f(<4 x half>* %a, <4 x half>* %b, <8 x half>* %c) {
7 ; CHECK-NEXT: movzwl (%rdi), %eax
8 ; CHECK-NEXT: movzwl 2(%rdi), %ecx
9 ; CHECK-NEXT: movw %cx, -{{[0-9]+}}(%rsp)
10 ; CHECK-NEXT: movw %ax, -{{[0-9]+}}(%rsp)
11 ; CHECK-NEXT: movzwl 6(%rdi), %r8d
12 ; CHECK-NEXT: movzwl 4(%rdi), %r11d
13 ; CHECK-NEXT: movq (%rsi), %rsi
14 ; CHECK-NEXT: movq %rsi, -{{[0-9]+}}(%rsp)
15 ; CHECK-NEXT: movdqa -{{[0-9]+}}(%rsp), %xmm0
16 ; CHECK-NEXT: pextrw $1, %xmm0, %r9d
17 ; CHECK-NEXT: movd %xmm0, %r10d
18 ; CHECK-NEXT: movl -{{[0-9]+}}(%rsp), %esi
19 ; CHECK-NEXT: pextrw $3, %xmm0, %eax
20 ; CHECK-NEXT: pextrw $2, %xmm0, %edi
21 ; CHECK-NEXT: movw %r11w, 8(%rdx)
22 ; CHECK-NEXT: movw %cx, 4(%rdx)
23 ; CHECK-NEXT: movw %r8w, 12(%rdx)
24 ; CHECK-NEXT: movw %si, (%rdx)
25 ; CHECK-NEXT: movw %di, 10(%rdx)
26 ; CHECK-NEXT: movw %ax, 14(%rdx)
27 ; CHECK-NEXT: movw %r10w, 2(%rdx)
28 ; CHECK-NEXT: movw %r9w, 6(%rdx)
30 %tmp4 = load <4 x half>, <4 x half>* %a
31 %tmp5 = load <4 x half>, <4 x half>* %b
32 %tmp7 = shufflevector <4 x half> %tmp4, <4 x half> %tmp5, <2 x i32> <i32 0, i32 4>
33 %tmp8 = shufflevector <4 x half> %tmp4, <4 x half> %tmp5, <2 x i32> <i32 1, i32 5>
34 %tmp9 = shufflevector <4 x half> %tmp4, <4 x half> %tmp5, <2 x i32> <i32 2, i32 6>
35 %tmp10 = shufflevector <4 x half> %tmp4, <4 x half> %tmp5, <2 x i32> <i32 3, i32 7>
36 %tmp11 = extractelement <2 x half> %tmp7, i32 0
37 %tmp12 = insertelement <8 x half> undef, half %tmp11, i32 0
38 %tmp13 = extractelement <2 x half> %tmp7, i32 1
39 %tmp14 = insertelement <8 x half> %tmp12, half %tmp13, i32 1
40 %tmp15 = extractelement <2 x half> %tmp8, i32 0
41 %tmp16 = insertelement <8 x half> %tmp14, half %tmp15, i32 2
42 %tmp17 = extractelement <2 x half> %tmp8, i32 1
43 %tmp18 = insertelement <8 x half> %tmp16, half %tmp17, i32 3
44 %tmp19 = extractelement <2 x half> %tmp9, i32 0
45 %tmp20 = insertelement <8 x half> %tmp18, half %tmp19, i32 4
46 %tmp21 = extractelement <2 x half> %tmp9, i32 1
47 %tmp22 = insertelement <8 x half> %tmp20, half %tmp21, i32 5
48 %tmp23 = extractelement <2 x half> %tmp10, i32 0
49 %tmp24 = insertelement <8 x half> %tmp22, half %tmp23, i32 6
50 %tmp25 = extractelement <2 x half> %tmp10, i32 1
51 %tmp26 = insertelement <8 x half> %tmp24, half %tmp25, i32 7
52 store <8 x half> %tmp26, <8 x half>* %c