1; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --no_x86_scrub_mem_shuffle
2; RUN: llc < %s -mtriple=x86_64-unknown-unknown | FileCheck %s --check-prefix=SSE
3; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse3 | FileCheck %s --check-prefix=SSE
4; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+ssse3 | FileCheck %s --check-prefix=SSE
5; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse4.1 | FileCheck %s --check-prefix=SSE
6; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s --check-prefix=AVX1
7; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=AVX2
8
9; PR21281
10define <64 x i16> @interleave8x8(<8 x i16> %a, <8 x i16> %b, <8 x i16> %c, <8 x i16> %d, <8 x i16> %e, <8 x i16> %f, <8 x i16> %h, <8 x i16> %g) {
11; SSE-LABEL: interleave8x8:
12; SSE:       # %bb.0:
13; SSE-NEXT:    movq %rdi, %rax
14; SSE-NEXT:    movdqa %xmm0, %xmm8
15; SSE-NEXT:    punpcklwd {{.*#+}} xmm8 = xmm8[0],xmm1[0],xmm8[1],xmm1[1],xmm8[2],xmm1[2],xmm8[3],xmm1[3]
16; SSE-NEXT:    punpckhwd {{.*#+}} xmm0 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
17; SSE-NEXT:    movdqa %xmm2, %xmm1
18; SSE-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3]
19; SSE-NEXT:    punpckhwd {{.*#+}} xmm2 = xmm2[4],xmm3[4],xmm2[5],xmm3[5],xmm2[6],xmm3[6],xmm2[7],xmm3[7]
20; SSE-NEXT:    movdqa %xmm0, %xmm3
21; SSE-NEXT:    punpckhdq {{.*#+}} xmm3 = xmm3[2],xmm2[2],xmm3[3],xmm2[3]
22; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
23; SSE-NEXT:    movdqa %xmm8, %xmm2
24; SSE-NEXT:    punpckhdq {{.*#+}} xmm2 = xmm2[2],xmm1[2],xmm2[3],xmm1[3]
25; SSE-NEXT:    punpckldq {{.*#+}} xmm8 = xmm8[0],xmm1[0],xmm8[1],xmm1[1]
26; SSE-NEXT:    movdqa %xmm4, %xmm1
27; SSE-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm5[0],xmm1[1],xmm5[1],xmm1[2],xmm5[2],xmm1[3],xmm5[3]
28; SSE-NEXT:    punpckhwd {{.*#+}} xmm4 = xmm4[4],xmm5[4],xmm4[5],xmm5[5],xmm4[6],xmm5[6],xmm4[7],xmm5[7]
29; SSE-NEXT:    movdqa %xmm7, %xmm5
30; SSE-NEXT:    punpcklwd {{.*#+}} xmm5 = xmm5[0],xmm6[0],xmm5[1],xmm6[1],xmm5[2],xmm6[2],xmm5[3],xmm6[3]
31; SSE-NEXT:    punpckhwd {{.*#+}} xmm7 = xmm7[4],xmm6[4],xmm7[5],xmm6[5],xmm7[6],xmm6[6],xmm7[7],xmm6[7]
32; SSE-NEXT:    movdqa %xmm4, %xmm6
33; SSE-NEXT:    punpckhdq {{.*#+}} xmm6 = xmm6[2],xmm7[2],xmm6[3],xmm7[3]
34; SSE-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm7[0],xmm4[1],xmm7[1]
35; SSE-NEXT:    movdqa %xmm1, %xmm7
36; SSE-NEXT:    punpckhdq {{.*#+}} xmm7 = xmm7[2],xmm5[2],xmm7[3],xmm5[3]
37; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm5[0],xmm1[1],xmm5[1]
38; SSE-NEXT:    movdqa %xmm8, %xmm5
39; SSE-NEXT:    punpcklwd {{.*#+}} xmm5 = xmm5[0],xmm1[0],xmm5[1],xmm1[1],xmm5[2],xmm1[2],xmm5[3],xmm1[3]
40; SSE-NEXT:    punpckhwd {{.*#+}} xmm8 = xmm8[4],xmm1[4],xmm8[5],xmm1[5],xmm8[6],xmm1[6],xmm8[7],xmm1[7]
41; SSE-NEXT:    movdqa %xmm2, %xmm1
42; SSE-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm7[0],xmm1[1],xmm7[1],xmm1[2],xmm7[2],xmm1[3],xmm7[3]
43; SSE-NEXT:    punpckhwd {{.*#+}} xmm2 = xmm2[4],xmm7[4],xmm2[5],xmm7[5],xmm2[6],xmm7[6],xmm2[7],xmm7[7]
44; SSE-NEXT:    movdqa %xmm0, %xmm7
45; SSE-NEXT:    punpcklwd {{.*#+}} xmm7 = xmm7[0],xmm4[0],xmm7[1],xmm4[1],xmm7[2],xmm4[2],xmm7[3],xmm4[3]
46; SSE-NEXT:    punpckhwd {{.*#+}} xmm0 = xmm0[4],xmm4[4],xmm0[5],xmm4[5],xmm0[6],xmm4[6],xmm0[7],xmm4[7]
47; SSE-NEXT:    movdqa %xmm3, %xmm4
48; SSE-NEXT:    punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm6[0],xmm4[1],xmm6[1],xmm4[2],xmm6[2],xmm4[3],xmm6[3]
49; SSE-NEXT:    punpckhwd {{.*#+}} xmm3 = xmm3[4],xmm6[4],xmm3[5],xmm6[5],xmm3[6],xmm6[6],xmm3[7],xmm6[7]
50; SSE-NEXT:    movdqa %xmm3, 112(%rdi)
51; SSE-NEXT:    movdqa %xmm4, 96(%rdi)
52; SSE-NEXT:    movdqa %xmm0, 80(%rdi)
53; SSE-NEXT:    movdqa %xmm7, 64(%rdi)
54; SSE-NEXT:    movdqa %xmm2, 48(%rdi)
55; SSE-NEXT:    movdqa %xmm1, 32(%rdi)
56; SSE-NEXT:    movdqa %xmm8, 16(%rdi)
57; SSE-NEXT:    movdqa %xmm5, (%rdi)
58; SSE-NEXT:    retq
59;
60; AVX1-LABEL: interleave8x8:
61; AVX1:       # %bb.0:
62; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm8 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
63; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm0 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
64; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm1 = xmm2[0],xmm3[0],xmm2[1],xmm3[1],xmm2[2],xmm3[2],xmm2[3],xmm3[3]
65; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm2 = xmm2[4],xmm3[4],xmm2[5],xmm3[5],xmm2[6],xmm3[6],xmm2[7],xmm3[7]
66; AVX1-NEXT:    vpunpckhdq {{.*#+}} xmm9 = xmm0[2],xmm2[2],xmm0[3],xmm2[3]
67; AVX1-NEXT:    vpunpckldq {{.*#+}} xmm2 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
68; AVX1-NEXT:    vpunpckhdq {{.*#+}} xmm3 = xmm8[2],xmm1[2],xmm8[3],xmm1[3]
69; AVX1-NEXT:    vpunpckldq {{.*#+}} xmm0 = xmm8[0],xmm1[0],xmm8[1],xmm1[1]
70; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm1 = xmm4[0],xmm5[0],xmm4[1],xmm5[1],xmm4[2],xmm5[2],xmm4[3],xmm5[3]
71; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm4 = xmm4[4],xmm5[4],xmm4[5],xmm5[5],xmm4[6],xmm5[6],xmm4[7],xmm5[7]
72; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm5 = xmm7[0],xmm6[0],xmm7[1],xmm6[1],xmm7[2],xmm6[2],xmm7[3],xmm6[3]
73; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm6 = xmm7[4],xmm6[4],xmm7[5],xmm6[5],xmm7[6],xmm6[6],xmm7[7],xmm6[7]
74; AVX1-NEXT:    vpunpckhdq {{.*#+}} xmm7 = xmm4[2],xmm6[2],xmm4[3],xmm6[3]
75; AVX1-NEXT:    vpunpckldq {{.*#+}} xmm4 = xmm4[0],xmm6[0],xmm4[1],xmm6[1]
76; AVX1-NEXT:    vpunpckhdq {{.*#+}} xmm6 = xmm1[2],xmm5[2],xmm1[3],xmm5[3]
77; AVX1-NEXT:    vpunpckldq {{.*#+}} xmm1 = xmm1[0],xmm5[0],xmm1[1],xmm5[1]
78; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm5 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
79; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
80; AVX1-NEXT:    vinsertf128 $1, %xmm5, %ymm0, %ymm0
81; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm1 = xmm3[4],xmm6[4],xmm3[5],xmm6[5],xmm3[6],xmm6[6],xmm3[7],xmm6[7]
82; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm3 = xmm3[0],xmm6[0],xmm3[1],xmm6[1],xmm3[2],xmm6[2],xmm3[3],xmm6[3]
83; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm3, %ymm1
84; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm3 = xmm2[4],xmm4[4],xmm2[5],xmm4[5],xmm2[6],xmm4[6],xmm2[7],xmm4[7]
85; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm2 = xmm2[0],xmm4[0],xmm2[1],xmm4[1],xmm2[2],xmm4[2],xmm2[3],xmm4[3]
86; AVX1-NEXT:    vinsertf128 $1, %xmm3, %ymm2, %ymm2
87; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm3 = xmm9[4],xmm7[4],xmm9[5],xmm7[5],xmm9[6],xmm7[6],xmm9[7],xmm7[7]
88; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm4 = xmm9[0],xmm7[0],xmm9[1],xmm7[1],xmm9[2],xmm7[2],xmm9[3],xmm7[3]
89; AVX1-NEXT:    vinsertf128 $1, %xmm3, %ymm4, %ymm3
90; AVX1-NEXT:    retq
91;
92; AVX2-LABEL: interleave8x8:
93; AVX2:       # %bb.0:
94; AVX2-NEXT:    vpunpcklwd {{.*#+}} xmm8 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
95; AVX2-NEXT:    vpunpckhwd {{.*#+}} xmm0 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
96; AVX2-NEXT:    vpunpcklwd {{.*#+}} xmm1 = xmm2[0],xmm3[0],xmm2[1],xmm3[1],xmm2[2],xmm3[2],xmm2[3],xmm3[3]
97; AVX2-NEXT:    vpunpckhwd {{.*#+}} xmm2 = xmm2[4],xmm3[4],xmm2[5],xmm3[5],xmm2[6],xmm3[6],xmm2[7],xmm3[7]
98; AVX2-NEXT:    vpunpckhdq {{.*#+}} xmm9 = xmm0[2],xmm2[2],xmm0[3],xmm2[3]
99; AVX2-NEXT:    vpunpckldq {{.*#+}} xmm2 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
100; AVX2-NEXT:    vpunpckhdq {{.*#+}} xmm3 = xmm8[2],xmm1[2],xmm8[3],xmm1[3]
101; AVX2-NEXT:    vpunpckldq {{.*#+}} xmm0 = xmm8[0],xmm1[0],xmm8[1],xmm1[1]
102; AVX2-NEXT:    vpunpcklwd {{.*#+}} xmm1 = xmm4[0],xmm5[0],xmm4[1],xmm5[1],xmm4[2],xmm5[2],xmm4[3],xmm5[3]
103; AVX2-NEXT:    vpunpckhwd {{.*#+}} xmm4 = xmm4[4],xmm5[4],xmm4[5],xmm5[5],xmm4[6],xmm5[6],xmm4[7],xmm5[7]
104; AVX2-NEXT:    vpunpcklwd {{.*#+}} xmm5 = xmm7[0],xmm6[0],xmm7[1],xmm6[1],xmm7[2],xmm6[2],xmm7[3],xmm6[3]
105; AVX2-NEXT:    vpunpckhwd {{.*#+}} xmm6 = xmm7[4],xmm6[4],xmm7[5],xmm6[5],xmm7[6],xmm6[6],xmm7[7],xmm6[7]
106; AVX2-NEXT:    vpunpckhdq {{.*#+}} xmm7 = xmm4[2],xmm6[2],xmm4[3],xmm6[3]
107; AVX2-NEXT:    vpunpckldq {{.*#+}} xmm4 = xmm4[0],xmm6[0],xmm4[1],xmm6[1]
108; AVX2-NEXT:    vpunpckhdq {{.*#+}} xmm6 = xmm1[2],xmm5[2],xmm1[3],xmm5[3]
109; AVX2-NEXT:    vpunpckldq {{.*#+}} xmm1 = xmm1[0],xmm5[0],xmm1[1],xmm5[1]
110; AVX2-NEXT:    vpunpckhwd {{.*#+}} xmm5 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
111; AVX2-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
112; AVX2-NEXT:    vinserti128 $1, %xmm5, %ymm0, %ymm0
113; AVX2-NEXT:    vpunpckhwd {{.*#+}} xmm1 = xmm3[4],xmm6[4],xmm3[5],xmm6[5],xmm3[6],xmm6[6],xmm3[7],xmm6[7]
114; AVX2-NEXT:    vpunpcklwd {{.*#+}} xmm3 = xmm3[0],xmm6[0],xmm3[1],xmm6[1],xmm3[2],xmm6[2],xmm3[3],xmm6[3]
115; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm3, %ymm1
116; AVX2-NEXT:    vpunpckhwd {{.*#+}} xmm3 = xmm2[4],xmm4[4],xmm2[5],xmm4[5],xmm2[6],xmm4[6],xmm2[7],xmm4[7]
117; AVX2-NEXT:    vpunpcklwd {{.*#+}} xmm2 = xmm2[0],xmm4[0],xmm2[1],xmm4[1],xmm2[2],xmm4[2],xmm2[3],xmm4[3]
118; AVX2-NEXT:    vinserti128 $1, %xmm3, %ymm2, %ymm2
119; AVX2-NEXT:    vpunpckhwd {{.*#+}} xmm3 = xmm9[4],xmm7[4],xmm9[5],xmm7[5],xmm9[6],xmm7[6],xmm9[7],xmm7[7]
120; AVX2-NEXT:    vpunpcklwd {{.*#+}} xmm4 = xmm9[0],xmm7[0],xmm9[1],xmm7[1],xmm9[2],xmm7[2],xmm9[3],xmm7[3]
121; AVX2-NEXT:    vinserti128 $1, %xmm3, %ymm4, %ymm3
122; AVX2-NEXT:    retq
123  %ab = shufflevector <8 x i16> %a, <8 x i16> %b, <16 x i32> <i32 0, i32 8, i32 1, i32 9, i32 2, i32 10, i32 3, i32 11, i32 4, i32 12, i32 5, i32 13, i32 6, i32 14, i32 7, i32 15>
124  %cd = shufflevector <8 x i16> %c, <8 x i16> %d, <16 x i32> <i32 0, i32 8, i32 1, i32 9, i32 2, i32 10, i32 3, i32 11, i32 4, i32 12, i32 5, i32 13, i32 6, i32 14, i32 7, i32 15>
125  %ab32 = bitcast <16 x i16> %ab to <8 x i32>
126  %cd32 = bitcast <16 x i16> %cd to <8 x i32>
127  %abcd32 = shufflevector <8 x i32> %ab32, <8 x i32> %cd32, <16 x i32> <i32 0, i32 8, i32 1, i32 9, i32 2, i32 10, i32 3, i32 11, i32 4, i32 12, i32 5, i32 13, i32 6, i32 14, i32 7, i32 15>
128  %abcd = bitcast <16 x i32> %abcd32 to <32 x i16>
129
130  %ef = shufflevector <8 x i16> %e, <8 x i16> %f, <16 x i32> <i32 0, i32 8, i32 1, i32 9, i32 2, i32 10, i32 3, i32 11, i32 4, i32 12, i32 5, i32 13, i32 6, i32 14, i32 7, i32 15>
131  %gh = shufflevector <8 x i16> %g, <8 x i16> %h, <16 x i32> <i32 0, i32 8, i32 1, i32 9, i32 2, i32 10, i32 3, i32 11, i32 4, i32 12, i32 5, i32 13, i32 6, i32 14, i32 7, i32 15>
132  %ef32 = bitcast <16 x i16> %ef to <8 x i32>
133  %gh32 = bitcast <16 x i16> %gh to <8 x i32>
134  %efgh32 = shufflevector <8 x i32> %ef32, <8 x i32> %gh32, <16 x i32> <i32 0, i32 8, i32 1, i32 9, i32 2, i32 10, i32 3, i32 11, i32 4, i32 12, i32 5, i32 13, i32 6, i32 14, i32 7, i32 15>
135  %efgh = bitcast <16 x i32> %efgh32 to <32 x i16>
136
137  %result = shufflevector <32 x i16> %abcd, <32 x i16> %efgh, <64 x i32> <i32 0, i32 32, i32 1, i32 33, i32 2, i32 34, i32 3, i32 35, i32 4, i32 36, i32 5, i32 37, i32 6, i32 38, i32 7, i32 39, i32 8, i32 40, i32 9, i32 41, i32 10, i32 42, i32 11, i32 43, i32 12, i32 44, i32 13, i32 45, i32 14, i32 46, i32 15, i32 47, i32 16, i32 48, i32 17, i32 49, i32 18, i32 50, i32 19, i32 51, i32 20, i32 52, i32 21, i32 53, i32 22, i32 54, i32 23, i32 55, i32 24, i32 56, i32 25, i32 57, i32 26, i32 58, i32 27, i32 59, i32 28, i32 60, i32 29, i32 61, i32 30, i32 62, i32 31, i32 63>
138  ret <64 x i16> %result
139}
140
141define void @splat2_i8(ptr %s, ptr %d) {
142; SSE-LABEL: splat2_i8:
143; SSE:       # %bb.0:
144; SSE-NEXT:    movdqu (%rdi), %xmm0
145; SSE-NEXT:    movdqu 16(%rdi), %xmm1
146; SSE-NEXT:    movdqa %xmm0, %xmm2
147; SSE-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
148; SSE-NEXT:    punpckhbw {{.*#+}} xmm0 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
149; SSE-NEXT:    movdqa %xmm1, %xmm3
150; SSE-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1],xmm3[2],xmm1[2],xmm3[3],xmm1[3],xmm3[4],xmm1[4],xmm3[5],xmm1[5],xmm3[6],xmm1[6],xmm3[7],xmm1[7]
151; SSE-NEXT:    punpckhbw {{.*#+}} xmm1 = xmm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
152; SSE-NEXT:    movdqu %xmm1, 48(%rsi)
153; SSE-NEXT:    movdqu %xmm3, 32(%rsi)
154; SSE-NEXT:    movdqu %xmm0, 16(%rsi)
155; SSE-NEXT:    movdqu %xmm2, (%rsi)
156; SSE-NEXT:    retq
157;
158; AVX1-LABEL: splat2_i8:
159; AVX1:       # %bb.0:
160; AVX1-NEXT:    vmovdqu (%rdi), %xmm0
161; AVX1-NEXT:    vmovdqu 16(%rdi), %xmm1
162; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm2 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
163; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm0 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
164; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm3 = xmm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
165; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm1 = xmm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
166; AVX1-NEXT:    vmovdqu %xmm1, 48(%rsi)
167; AVX1-NEXT:    vmovdqu %xmm3, 32(%rsi)
168; AVX1-NEXT:    vmovdqu %xmm0, 16(%rsi)
169; AVX1-NEXT:    vmovdqu %xmm2, (%rsi)
170; AVX1-NEXT:    retq
171;
172; AVX2-LABEL: splat2_i8:
173; AVX2:       # %bb.0:
174; AVX2-NEXT:    vpermq $216, (%rdi), %ymm0 # ymm0 = mem[0,2,1,3]
175; AVX2-NEXT:    vpunpcklbw {{.*#+}} ymm1 = ymm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23]
176; AVX2-NEXT:    vpunpckhbw {{.*#+}} ymm0 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
177; AVX2-NEXT:    vmovdqu %ymm0, 32(%rsi)
178; AVX2-NEXT:    vmovdqu %ymm1, (%rsi)
179; AVX2-NEXT:    vzeroupper
180; AVX2-NEXT:    retq
181  %ld32 = load <32 x i8>, ptr %s, align 1
182  %cat = shufflevector <32 x i8> %ld32, <32 x i8> undef, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
183  %cat2 = shufflevector <64 x i8> %cat, <64 x i8> undef, <64 x i32> <i32 0, i32 32, i32 1, i32 33, i32 2, i32 34, i32 3, i32 35, i32 4, i32 36, i32 5, i32 37, i32 6, i32 38, i32 7, i32 39, i32 8, i32 40, i32 9, i32 41, i32 10, i32 42, i32 11, i32 43, i32 12, i32 44, i32 13, i32 45, i32 14, i32 46, i32 15, i32 47, i32 16, i32 48, i32 17, i32 49, i32 18, i32 50, i32 19, i32 51, i32 20, i32 52, i32 21, i32 53, i32 22, i32 54, i32 23, i32 55, i32 24, i32 56, i32 25, i32 57, i32 26, i32 58, i32 27, i32 59, i32 28, i32 60, i32 29, i32 61, i32 30, i32 62, i32 31, i32 63>
184  store <64 x i8> %cat2, ptr %d, align 1
185  ret void
186}
187
188define void @splat2_i16(ptr %s, ptr %d) {
189; SSE-LABEL: splat2_i16:
190; SSE:       # %bb.0:
191; SSE-NEXT:    movdqu (%rdi), %xmm0
192; SSE-NEXT:    movdqu 16(%rdi), %xmm1
193; SSE-NEXT:    movdqa %xmm0, %xmm2
194; SSE-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
195; SSE-NEXT:    punpckhwd {{.*#+}} xmm0 = xmm0[4,4,5,5,6,6,7,7]
196; SSE-NEXT:    movdqa %xmm1, %xmm3
197; SSE-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1],xmm3[2],xmm1[2],xmm3[3],xmm1[3]
198; SSE-NEXT:    punpckhwd {{.*#+}} xmm1 = xmm1[4,4,5,5,6,6,7,7]
199; SSE-NEXT:    movdqu %xmm1, 48(%rsi)
200; SSE-NEXT:    movdqu %xmm3, 32(%rsi)
201; SSE-NEXT:    movdqu %xmm0, 16(%rsi)
202; SSE-NEXT:    movdqu %xmm2, (%rsi)
203; SSE-NEXT:    retq
204;
205; AVX1-LABEL: splat2_i16:
206; AVX1:       # %bb.0:
207; AVX1-NEXT:    vmovdqu (%rdi), %xmm0
208; AVX1-NEXT:    vmovdqu 16(%rdi), %xmm1
209; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm2 = xmm0[0,0,1,1,2,2,3,3]
210; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm0 = xmm0[4,4,5,5,6,6,7,7]
211; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm3 = xmm1[0,0,1,1,2,2,3,3]
212; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm1 = xmm1[4,4,5,5,6,6,7,7]
213; AVX1-NEXT:    vmovdqu %xmm1, 48(%rsi)
214; AVX1-NEXT:    vmovdqu %xmm3, 32(%rsi)
215; AVX1-NEXT:    vmovdqu %xmm0, 16(%rsi)
216; AVX1-NEXT:    vmovdqu %xmm2, (%rsi)
217; AVX1-NEXT:    retq
218;
219; AVX2-LABEL: splat2_i16:
220; AVX2:       # %bb.0:
221; AVX2-NEXT:    vpermq $216, (%rdi), %ymm0 # ymm0 = mem[0,2,1,3]
222; AVX2-NEXT:    vpunpcklwd {{.*#+}} ymm1 = ymm0[0,0,1,1,2,2,3,3,8,8,9,9,10,10,11,11]
223; AVX2-NEXT:    vpunpckhwd {{.*#+}} ymm0 = ymm0[4,4,5,5,6,6,7,7,12,12,13,13,14,14,15,15]
224; AVX2-NEXT:    vmovdqu %ymm0, 32(%rsi)
225; AVX2-NEXT:    vmovdqu %ymm1, (%rsi)
226; AVX2-NEXT:    vzeroupper
227; AVX2-NEXT:    retq
228  %ld32 = load <16 x i16>, ptr %s, align 1
229  %cat = shufflevector <16 x i16> %ld32, <16 x i16> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
230  %cat2 = shufflevector <32 x i16> %cat, <32 x i16> undef, <32 x i32> <i32 0, i32 16, i32 1, i32 17, i32 2, i32 18, i32 3, i32 19, i32 4, i32 20, i32 5, i32 21, i32 6, i32 22, i32 7, i32 23, i32 8, i32 24, i32 9, i32 25, i32 10, i32 26, i32 11, i32 27, i32 12, i32 28, i32 13, i32 29, i32 14, i32 30, i32 15, i32 31>
231  store <32 x i16> %cat2, ptr %d, align 1
232  ret void
233}
234
235define void @splat2_i32(ptr %s, ptr %d) {
236; SSE-LABEL: splat2_i32:
237; SSE:       # %bb.0:
238; SSE-NEXT:    movdqu (%rdi), %xmm0
239; SSE-NEXT:    movdqu 16(%rdi), %xmm1
240; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[0,0,1,1]
241; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,2,3,3]
242; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm1[0,0,1,1]
243; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[2,2,3,3]
244; SSE-NEXT:    movdqu %xmm1, 48(%rsi)
245; SSE-NEXT:    movdqu %xmm3, 32(%rsi)
246; SSE-NEXT:    movdqu %xmm0, 16(%rsi)
247; SSE-NEXT:    movdqu %xmm2, (%rsi)
248; SSE-NEXT:    retq
249;
250; AVX1-LABEL: splat2_i32:
251; AVX1:       # %bb.0:
252; AVX1-NEXT:    vmovups (%rdi), %xmm0
253; AVX1-NEXT:    vmovups 16(%rdi), %xmm1
254; AVX1-NEXT:    vpermilps {{.*#+}} xmm2 = xmm0[0,0,1,1]
255; AVX1-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[2,2,3,3]
256; AVX1-NEXT:    vpermilps {{.*#+}} xmm3 = xmm1[0,0,1,1]
257; AVX1-NEXT:    vpermilps {{.*#+}} xmm1 = xmm1[2,2,3,3]
258; AVX1-NEXT:    vmovups %xmm1, 48(%rsi)
259; AVX1-NEXT:    vmovups %xmm3, 32(%rsi)
260; AVX1-NEXT:    vmovups %xmm0, 16(%rsi)
261; AVX1-NEXT:    vmovups %xmm2, (%rsi)
262; AVX1-NEXT:    retq
263;
264; AVX2-LABEL: splat2_i32:
265; AVX2:       # %bb.0:
266; AVX2-NEXT:    vpermpd $216, (%rdi), %ymm0 # ymm0 = mem[0,2,1,3]
267; AVX2-NEXT:    vpermilps {{.*#+}} ymm1 = ymm0[0,0,1,1,4,4,5,5]
268; AVX2-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[2,2,3,3,6,6,7,7]
269; AVX2-NEXT:    vmovups %ymm0, 32(%rsi)
270; AVX2-NEXT:    vmovups %ymm1, (%rsi)
271; AVX2-NEXT:    vzeroupper
272; AVX2-NEXT:    retq
273  %ld32 = load <8 x i32>, ptr %s, align 1
274  %cat = shufflevector <8 x i32> %ld32, <8 x i32> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
275  %cat2 = shufflevector <16 x i32> %cat, <16 x i32> undef, <16 x i32> <i32 0, i32 8, i32 1, i32 9, i32 2, i32 10, i32 3, i32 11, i32 4, i32 12, i32 5, i32 13, i32 6, i32 14, i32 7, i32 15>
276  store <16 x i32> %cat2, ptr %d, align 1
277  ret void
278}
279
280define void @splat2_i64(ptr %s, ptr %d) {
281; SSE-LABEL: splat2_i64:
282; SSE:       # %bb.0:
283; SSE-NEXT:    movdqu (%rdi), %xmm0
284; SSE-NEXT:    movdqu 16(%rdi), %xmm1
285; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[0,1,0,1]
286; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,3,2,3]
287; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm1[0,1,0,1]
288; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[2,3,2,3]
289; SSE-NEXT:    movdqu %xmm1, 48(%rsi)
290; SSE-NEXT:    movdqu %xmm3, 32(%rsi)
291; SSE-NEXT:    movdqu %xmm0, 16(%rsi)
292; SSE-NEXT:    movdqu %xmm2, (%rsi)
293; SSE-NEXT:    retq
294;
295; AVX1-LABEL: splat2_i64:
296; AVX1:       # %bb.0:
297; AVX1-NEXT:    vperm2f128 $51, (%rdi), %ymm0, %ymm0 # ymm0 = mem[2,3,2,3]
298; AVX1-NEXT:    vpermilpd {{.*#+}} ymm0 = ymm0[0,0,3,3]
299; AVX1-NEXT:    vbroadcastf128 (%rdi), %ymm1 # ymm1 = mem[0,1,0,1]
300; AVX1-NEXT:    vpermilpd {{.*#+}} ymm1 = ymm1[0,0,3,3]
301; AVX1-NEXT:    vmovupd %ymm0, 32(%rsi)
302; AVX1-NEXT:    vmovupd %ymm1, (%rsi)
303; AVX1-NEXT:    vzeroupper
304; AVX1-NEXT:    retq
305;
306; AVX2-LABEL: splat2_i64:
307; AVX2:       # %bb.0:
308; AVX2-NEXT:    vmovups (%rdi), %ymm0
309; AVX2-NEXT:    vpermpd {{.*#+}} ymm1 = ymm0[0,0,1,1]
310; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[2,2,3,3]
311; AVX2-NEXT:    vmovups %ymm0, 32(%rsi)
312; AVX2-NEXT:    vmovups %ymm1, (%rsi)
313; AVX2-NEXT:    vzeroupper
314; AVX2-NEXT:    retq
315  %ld32 = load <4 x i64>, ptr %s, align 1
316  %cat = shufflevector <4 x i64> %ld32, <4 x i64> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3>
317  %cat2 = shufflevector <8 x i64> %cat, <8 x i64> undef, <8 x i32> <i32 0, i32 4, i32 1, i32 5, i32 2, i32 6, i32 3, i32 7>
318  store <8 x i64> %cat2, ptr %d, align 1
319  ret void
320}
321