1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc < %s -mcpu=x86-64 -mattr=+avx512f,+avx512dq | FileCheck %s --check-prefix=ALL --check-prefix=AVX512 --check-prefix=AVX512F
3 ; RUN: llc < %s -mcpu=x86-64 -mattr=+avx512bw,+avx512dq | FileCheck %s --check-prefix=ALL --check-prefix=AVX512 --check-prefix=AVX512BW
5 target triple = "x86_64-unknown-unknown"
7 define <16 x float> @shuffle_v16f32_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00(<16 x float> %a, <16 x float> %b) {
8 ; ALL-LABEL: shuffle_v16f32_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
10 ; ALL-NEXT: vbroadcastss %xmm0, %zmm0
12 %shuffle = shufflevector <16 x float> %a, <16 x float> %b, <16 x i32><i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
13 ret <16 x float> %shuffle
16 define <16 x float> @shuffle_v16f32_08_08_08_08_08_08_08_08_08_08_08_08_08_08_08_08(<16 x float> %a, <16 x float> %b) {
17 ; ALL-LABEL: shuffle_v16f32_08_08_08_08_08_08_08_08_08_08_08_08_08_08_08_08:
19 ; ALL-NEXT: vextractf32x4 $2, %zmm0, %xmm0
20 ; ALL-NEXT: vbroadcastss %xmm0, %zmm0
22 %shuffle = shufflevector <16 x float> %a, <16 x float> %b, <16 x i32><i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8>
23 ret <16 x float> %shuffle
26 define <16 x float> @shuffle_v16f32_08_08_08_08_08_08_08_08_08_08_08_08_08_08_08_08_bc(<16 x i32> %a, <16 x i32> %b) {
27 ; ALL-LABEL: shuffle_v16f32_08_08_08_08_08_08_08_08_08_08_08_08_08_08_08_08_bc:
29 ; ALL-NEXT: vextractf32x4 $2, %zmm0, %xmm0
30 ; ALL-NEXT: vbroadcastss %xmm0, %zmm0
32 %tmp0 = bitcast <16 x i32> %a to <16 x float>
33 %tmp1 = bitcast <16 x i32> %b to <16 x float>
34 %shuffle = shufflevector <16 x float> %tmp0, <16 x float> %tmp1, <16 x i32><i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8>
35 ret <16 x float> %shuffle
38 define <16 x float> @shuffle_v16f32_00_10_01_11_04_14_05_15_08_18_09_19_0c_1c_0d_1d(<16 x float> %a, <16 x float> %b) {
39 ; ALL-LABEL: shuffle_v16f32_00_10_01_11_04_14_05_15_08_18_09_19_0c_1c_0d_1d:
41 ; ALL-NEXT: vunpcklps {{.*#+}} zmm0 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13]
43 %shuffle = shufflevector <16 x float> %a, <16 x float> %b, <16 x i32><i32 0, i32 16, i32 1, i32 17, i32 4, i32 20, i32 5, i32 21, i32 8, i32 24, i32 9, i32 25, i32 12, i32 28, i32 13, i32 29>
44 ret <16 x float> %shuffle
47 define <16 x float> @shuffle_v16f32_00_zz_01_zz_04_zz_05_zz_08_zz_09_zz_0c_zz_0d_zz(<16 x float> %a, <16 x float> %b) {
48 ; ALL-LABEL: shuffle_v16f32_00_zz_01_zz_04_zz_05_zz_08_zz_09_zz_0c_zz_0d_zz:
50 ; ALL-NEXT: vxorps %xmm1, %xmm1, %xmm1
51 ; ALL-NEXT: vunpcklps {{.*#+}} zmm0 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13]
53 %shuffle = shufflevector <16 x float> %a, <16 x float> zeroinitializer, <16 x i32><i32 0, i32 16, i32 1, i32 16, i32 4, i32 16, i32 5, i32 16, i32 8, i32 16, i32 9, i32 16, i32 12, i32 16, i32 13, i32 16>
54 ret <16 x float> %shuffle
57 define <16 x float> @shuffle_v16f32_vunpcklps_swap(<16 x float> %a, <16 x float> %b) {
58 ; ALL-LABEL: shuffle_v16f32_vunpcklps_swap:
60 ; ALL-NEXT: vunpcklps {{.*#+}} zmm0 = zmm1[0],zmm0[0],zmm1[1],zmm0[1],zmm1[4],zmm0[4],zmm1[5],zmm0[5],zmm1[8],zmm0[8],zmm1[9],zmm0[9],zmm1[12],zmm0[12],zmm1[13],zmm0[13]
62 %shuffle = shufflevector <16 x float> %a, <16 x float> %b, <16 x i32> <i32 16, i32 0, i32 17, i32 1, i32 20, i32 4, i32 21, i32 5, i32 24, i32 8, i32 25, i32 9, i32 28, i32 12, i32 29, i32 13>
63 ret <16 x float> %shuffle
67 define <16 x float> @shuffle_v16f32_01_01_03_00_06_04_05_07_08_08_09_09_15_14_14_12(<16 x float> %a0) {
68 ; ALL-LABEL: shuffle_v16f32_01_01_03_00_06_04_05_07_08_08_09_09_15_14_14_12:
70 ; ALL-NEXT: vpermilps {{.*#+}} zmm0 = zmm0[1,1,3,0,6,4,5,7,8,8,9,9,15,14,14,12]
72 %shuffle = shufflevector <16 x float> %a0, <16 x float> undef, <16 x i32> <i32 1, i32 1, i32 3, i32 0, i32 6, i32 4, i32 5, i32 7, i32 8, i32 8, i32 9, i32 9, i32 15, i32 14, i32 14, i32 12>
73 ret <16 x float> %shuffle
76 define <16 x i32> @shuffle_v16i32_00_10_01_11_04_14_05_15_08_18_09_19_0c_1c_0d_1d(<16 x i32> %a, <16 x i32> %b) {
77 ; ALL-LABEL: shuffle_v16i32_00_10_01_11_04_14_05_15_08_18_09_19_0c_1c_0d_1d:
79 ; ALL-NEXT: vunpcklps {{.*#+}} zmm0 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13]
81 %shuffle = shufflevector <16 x i32> %a, <16 x i32> %b, <16 x i32><i32 0, i32 16, i32 1, i32 17, i32 4, i32 20, i32 5, i32 21, i32 8, i32 24, i32 9, i32 25, i32 12, i32 28, i32 13, i32 29>
82 ret <16 x i32> %shuffle
85 define <16 x i32> @shuffle_v16i32_zz_10_zz_11_zz_14_zz_15_zz_18_zz_19_zz_1c_zz_1d(<16 x i32> %a, <16 x i32> %b) {
86 ; ALL-LABEL: shuffle_v16i32_zz_10_zz_11_zz_14_zz_15_zz_18_zz_19_zz_1c_zz_1d:
88 ; ALL-NEXT: vxorps %xmm0, %xmm0, %xmm0
89 ; ALL-NEXT: vunpcklps {{.*#+}} zmm0 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13]
91 %shuffle = shufflevector <16 x i32> zeroinitializer, <16 x i32> %b, <16 x i32><i32 15, i32 16, i32 13, i32 17, i32 11, i32 20, i32 9, i32 21, i32 7, i32 24, i32 5, i32 25, i32 3, i32 28, i32 1, i32 29>
92 ret <16 x i32> %shuffle
95 define <16 x float> @shuffle_v16f32_02_12_03_13_06_16_07_17_0a_1a_0b_1b_0e_1e_0f_1f(<16 x float> %a, <16 x float> %b) {
96 ; ALL-LABEL: shuffle_v16f32_02_12_03_13_06_16_07_17_0a_1a_0b_1b_0e_1e_0f_1f:
98 ; ALL-NEXT: vunpckhps {{.*#+}} zmm0 = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15]
100 %shuffle = shufflevector <16 x float> %a, <16 x float> %b, <16 x i32><i32 2, i32 18, i32 3, i32 19, i32 6, i32 22, i32 7, i32 23, i32 10, i32 26, i32 11, i32 27, i32 14, i32 30, i32 15, i32 31>
101 ret <16 x float> %shuffle
104 define <16 x float> @shuffle_v16f32_zz_12_zz_13_zz_16_zz_17_zz_1a_zz_1b_zz_1e_zz_1f(<16 x float> %a, <16 x float> %b) {
105 ; ALL-LABEL: shuffle_v16f32_zz_12_zz_13_zz_16_zz_17_zz_1a_zz_1b_zz_1e_zz_1f:
107 ; ALL-NEXT: vxorps %xmm0, %xmm0, %xmm0
108 ; ALL-NEXT: vunpckhps {{.*#+}} zmm0 = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15]
110 %shuffle = shufflevector <16 x float> zeroinitializer, <16 x float> %b, <16 x i32><i32 0, i32 18, i32 0, i32 19, i32 4, i32 22, i32 4, i32 23, i32 6, i32 26, i32 6, i32 27, i32 8, i32 30, i32 8, i32 31>
111 ret <16 x float> %shuffle
114 define <16 x float> @shuffle_v16f32_00_00_02_02_04_04_06_06_08_08_10_10_12_12_14_14(<16 x float> %a, <16 x float> %b) {
115 ; ALL-LABEL: shuffle_v16f32_00_00_02_02_04_04_06_06_08_08_10_10_12_12_14_14:
117 ; ALL-NEXT: vmovsldup {{.*#+}} zmm0 = zmm0[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
119 %shuffle = shufflevector <16 x float> %a, <16 x float> %b, <16 x i32><i32 0, i32 0, i32 2, i32 2, i32 4, i32 4, i32 6, i32 6, i32 8, i32 8, i32 10, i32 10, i32 12, i32 12, i32 14, i32 14>
120 ret <16 x float> %shuffle
123 define <16 x float> @shuffle_v16f32_01_01_03_03_05_05_07_07_09_09_11_11_13_13_15_15(<16 x float> %a, <16 x float> %b) {
124 ; ALL-LABEL: shuffle_v16f32_01_01_03_03_05_05_07_07_09_09_11_11_13_13_15_15:
126 ; ALL-NEXT: vmovshdup {{.*#+}} zmm0 = zmm0[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
128 %shuffle = shufflevector <16 x float> %a, <16 x float> %b, <16 x i32><i32 1, i32 1, i32 3, i32 3, i32 5, i32 5, i32 7, i32 7, i32 9, i32 9, i32 11, i32 11, i32 13, i32 13, i32 15, i32 15>
129 ret <16 x float> %shuffle
132 define <16 x float> @shuffle_v16f32_00_01_00_01_06_07_06_07_08_09_10_11_12_13_12_13(<16 x float> %a, <16 x float> %b) {
133 ; ALL-LABEL: shuffle_v16f32_00_01_00_01_06_07_06_07_08_09_10_11_12_13_12_13:
135 ; ALL-NEXT: vpermilpd {{.*#+}} zmm0 = zmm0[0,0,3,3,4,5,6,6]
137 %shuffle = shufflevector <16 x float> %a, <16 x float> %b, <16 x i32> <i32 0, i32 1, i32 0, i32 1, i32 6, i32 7, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 12, i32 13>
138 ret <16 x float> %shuffle
141 define <16 x float> @shuffle_v16f32_00_00_02_00_04_04_06_04_08_08_10_08_12_12_14_12(<16 x float> %a, <16 x float> %b) {
142 ; ALL-LABEL: shuffle_v16f32_00_00_02_00_04_04_06_04_08_08_10_08_12_12_14_12:
144 ; ALL-NEXT: vpermilps {{.*#+}} zmm0 = zmm0[0,0,2,0,4,4,6,4,8,8,10,8,12,12,14,12]
146 %shuffle = shufflevector <16 x float> %a, <16 x float> %b, <16 x i32> <i32 0, i32 0, i32 2, i32 0, i32 4, i32 4, i32 6, i32 4, i32 8, i32 8, i32 10, i32 8, i32 12, i32 12, i32 14, i32 12>
147 ret <16 x float> %shuffle
150 define <16 x float> @shuffle_v16f32_03_uu_uu_uu_uu_04_uu_uu_uu_uu_11_uu_uu_uu_uu_12(<16 x float> %a, <16 x float> %b) {
151 ; ALL-LABEL: shuffle_v16f32_03_uu_uu_uu_uu_04_uu_uu_uu_uu_11_uu_uu_uu_uu_12:
153 ; ALL-NEXT: vpermilps {{.*#+}} zmm0 = zmm0[3,0,3,0,7,4,7,4,11,8,11,8,15,12,15,12]
155 %shuffle = shufflevector <16 x float> %a, <16 x float> %b, <16 x i32> <i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 4, i32 undef, i32 undef, i32 undef, i32 undef, i32 11, i32 undef, i32 undef, i32 undef, i32 undef, i32 12>
156 ret <16 x float> %shuffle
159 define <16 x i32> @shuffle_v16i32_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00(<16 x i32> %a, <16 x i32> %b) {
160 ; ALL-LABEL: shuffle_v16i32_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
162 ; ALL-NEXT: vbroadcastss %xmm0, %zmm0
164 %shuffle = shufflevector <16 x i32> %a, <16 x i32> %b, <16 x i32><i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
165 ret <16 x i32> %shuffle
168 define <16 x i32> @shuffle_v16i32_04_04_04_04_04_04_04_04_04_04_04_04_04_04_04_04(<16 x i32> %a, <16 x i32> %b) {
169 ; ALL-LABEL: shuffle_v16i32_04_04_04_04_04_04_04_04_04_04_04_04_04_04_04_04:
171 ; ALL-NEXT: vextractf128 $1, %ymm0, %xmm0
172 ; ALL-NEXT: vbroadcastss %xmm0, %zmm0
174 %shuffle = shufflevector <16 x i32> %a, <16 x i32> %b, <16 x i32><i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4>
175 ret <16 x i32> %shuffle
178 define <16 x i32> @shuffle_v16i32_02_12_03_13_06_16_07_17_0a_1a_0b_1b_0e_1e_0f_1f(<16 x i32> %a, <16 x i32> %b) {
179 ; ALL-LABEL: shuffle_v16i32_02_12_03_13_06_16_07_17_0a_1a_0b_1b_0e_1e_0f_1f:
181 ; ALL-NEXT: vunpckhps {{.*#+}} zmm0 = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15]
183 %shuffle = shufflevector <16 x i32> %a, <16 x i32> %b, <16 x i32><i32 2, i32 18, i32 3, i32 19, i32 6, i32 22, i32 7, i32 23, i32 10, i32 26, i32 11, i32 27, i32 14, i32 30, i32 15, i32 31>
184 ret <16 x i32> %shuffle
187 define <16 x i32> @shuffle_v16i32_02_zz_03_zz_06_zz_07_zz_0a_zz_0b_zz_0e_zz_0f_zz(<16 x i32> %a, <16 x i32> %b) {
188 ; ALL-LABEL: shuffle_v16i32_02_zz_03_zz_06_zz_07_zz_0a_zz_0b_zz_0e_zz_0f_zz:
190 ; ALL-NEXT: vxorps %xmm1, %xmm1, %xmm1
191 ; ALL-NEXT: vunpckhps {{.*#+}} zmm0 = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15]
193 %shuffle = shufflevector <16 x i32> %a, <16 x i32> zeroinitializer, <16 x i32><i32 2, i32 30, i32 3, i32 28, i32 6, i32 26, i32 7, i32 24, i32 10, i32 22, i32 11, i32 20, i32 14, i32 18, i32 15, i32 16>
194 ret <16 x i32> %shuffle
197 define <16 x i32> @shuffle_v16i32_01_02_03_16_05_06_07_20_09_10_11_24_13_14_15_28(<16 x i32> %a, <16 x i32> %b) {
198 ; AVX512F-LABEL: shuffle_v16i32_01_02_03_16_05_06_07_20_09_10_11_24_13_14_15_28:
200 ; AVX512F-NEXT: vmovdqa64 {{.*#+}} zmm2 = [1,2,3,16,5,6,7,20,9,10,11,24,13,14,15,28]
201 ; AVX512F-NEXT: vpermt2d %zmm1, %zmm2, %zmm0
204 ; AVX512BW-LABEL: shuffle_v16i32_01_02_03_16_05_06_07_20_09_10_11_24_13_14_15_28:
206 ; AVX512BW-NEXT: vpalignr {{.*#+}} zmm0 = zmm0[4,5,6,7,8,9,10,11,12,13,14,15],zmm1[0,1,2,3],zmm0[20,21,22,23,24,25,26,27,28,29,30,31],zmm1[16,17,18,19],zmm0[36,37,38,39,40,41,42,43,44,45,46,47],zmm1[32,33,34,35],zmm0[52,53,54,55,56,57,58,59,60,61,62,63],zmm1[48,49,50,51]
207 ; AVX512BW-NEXT: retq
208 %shuffle = shufflevector <16 x i32> %a, <16 x i32> %b, <16 x i32><i32 1, i32 2, i32 3, i32 16, i32 5, i32 6, i32 7, i32 20, i32 9, i32 10, i32 11, i32 24, i32 13, i32 14, i32 15, i32 28>
209 ret <16 x i32> %shuffle
212 define <16 x float> @shuffle_v16f32_02_05_u_u_07_u_0a_01_00_05_u_04_07_u_0a_01(<16 x float> %a) {
213 ; ALL-LABEL: shuffle_v16f32_02_05_u_u_07_u_0a_01_00_05_u_04_07_u_0a_01:
215 ; ALL-NEXT: vmovaps {{.*#+}} zmm1 = <2,5,u,u,7,u,10,1,0,5,u,4,7,u,10,1>
216 ; ALL-NEXT: vpermps %zmm0, %zmm1, %zmm0
218 %c = shufflevector <16 x float> %a, <16 x float> undef, <16 x i32> <i32 2, i32 5, i32 undef, i32 undef, i32 7, i32 undef, i32 10, i32 1, i32 0, i32 5, i32 undef, i32 4, i32 7, i32 undef, i32 10, i32 1>
222 define <16 x i32> @shuffle_v16i32_02_05_u_u_07_u_0a_01_00_05_u_04_07_u_0a_01(<16 x i32> %a) {
223 ; ALL-LABEL: shuffle_v16i32_02_05_u_u_07_u_0a_01_00_05_u_04_07_u_0a_01:
225 ; ALL-NEXT: vmovaps {{.*#+}} zmm1 = <2,5,u,u,7,u,10,1,0,5,u,4,7,u,10,1>
226 ; ALL-NEXT: vpermps %zmm0, %zmm1, %zmm0
228 %c = shufflevector <16 x i32> %a, <16 x i32> undef, <16 x i32> <i32 2, i32 5, i32 undef, i32 undef, i32 7, i32 undef, i32 10, i32 1, i32 0, i32 5, i32 undef, i32 4, i32 7, i32 undef, i32 10, i32 1>
232 define <16 x i32> @shuffle_v16i32_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_18(<16 x i32> %a, <16 x i32> %b) {
233 ; ALL-LABEL: shuffle_v16i32_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_18:
235 ; ALL-NEXT: vmovdqa64 {{.*#+}} zmm2 = [15,31,14,22,13,29,4,28,11,27,10,26,9,25,8,24]
236 ; ALL-NEXT: vpermt2d %zmm1, %zmm2, %zmm0
238 %c = shufflevector <16 x i32> %a, <16 x i32> %b, <16 x i32> <i32 15, i32 31, i32 14, i32 22, i32 13, i32 29, i32 4, i32 28, i32 11, i32 27, i32 10, i32 26, i32 9, i32 25, i32 8, i32 24>
242 define <16 x float> @shuffle_v16f32_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_18(<16 x float> %a, <16 x float> %b) {
243 ; ALL-LABEL: shuffle_v16f32_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_18:
245 ; ALL-NEXT: vmovaps {{.*#+}} zmm2 = [15,31,14,22,13,29,4,28,11,27,10,26,9,25,8,24]
246 ; ALL-NEXT: vpermt2ps %zmm1, %zmm2, %zmm0
248 %c = shufflevector <16 x float> %a, <16 x float> %b, <16 x i32> <i32 15, i32 31, i32 14, i32 22, i32 13, i32 29, i32 4, i32 28, i32 11, i32 27, i32 10, i32 26, i32 9, i32 25, i32 8, i32 24>
252 define <16 x float> @shuffle_v16f32_load_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_18(<16 x float> %a, <16 x float>* %b) {
253 ; ALL-LABEL: shuffle_v16f32_load_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_18:
255 ; ALL-NEXT: vmovaps {{.*#+}} zmm1 = [15,31,14,22,13,29,4,28,11,27,10,26,9,25,8,24]
256 ; ALL-NEXT: vpermt2ps (%rdi), %zmm1, %zmm0
258 %c = load <16 x float>, <16 x float>* %b
259 %d = shufflevector <16 x float> %a, <16 x float> %c, <16 x i32> <i32 15, i32 31, i32 14, i32 22, i32 13, i32 29, i32 4, i32 28, i32 11, i32 27, i32 10, i32 26, i32 9, i32 25, i32 8, i32 24>
263 define <16 x i32> @shuffle_v16i32_load_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_18(<16 x i32> %a, <16 x i32>* %b) {
264 ; ALL-LABEL: shuffle_v16i32_load_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_18:
266 ; ALL-NEXT: vmovdqa64 {{.*#+}} zmm1 = [15,31,14,22,13,29,4,28,11,27,10,26,9,25,8,24]
267 ; ALL-NEXT: vpermt2d (%rdi), %zmm1, %zmm0
269 %c = load <16 x i32>, <16 x i32>* %b
270 %d = shufflevector <16 x i32> %a, <16 x i32> %c, <16 x i32> <i32 15, i32 31, i32 14, i32 22, i32 13, i32 29, i32 4, i32 28, i32 11, i32 27, i32 10, i32 26, i32 9, i32 25, i32 8, i32 24>
274 define <16 x i32> @shuffle_v16i32_0_1_2_19_u_u_u_u_u_u_u_u_u_u_u_u(<16 x i32> %a, <16 x i32> %b) {
275 ; ALL-LABEL: shuffle_v16i32_0_1_2_19_u_u_u_u_u_u_u_u_u_u_u_u:
277 ; ALL-NEXT: vblendps {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[3]
279 %c = shufflevector <16 x i32> %a, <16 x i32> %b, <16 x i32> <i32 0, i32 1, i32 2, i32 19, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
283 ;FIXME: can do better with vpcompress
284 define <8 x i32> @test_v16i32_1_3_5_7_9_11_13_15(<16 x i32> %v) {
285 ; ALL-LABEL: test_v16i32_1_3_5_7_9_11_13_15:
287 ; ALL-NEXT: vextractf64x4 $1, %zmm0, %ymm1
288 ; ALL-NEXT: vshufps {{.*#+}} ymm0 = ymm0[1,3],ymm1[1,3],ymm0[5,7],ymm1[5,7]
289 ; ALL-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,2,1,3]
291 %res = shufflevector <16 x i32> %v, <16 x i32> undef, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
295 ;FIXME: can do better with vpcompress
296 define <4 x i32> @test_v16i32_0_1_2_12 (<16 x i32> %v) {
297 ; ALL-LABEL: test_v16i32_0_1_2_12:
299 ; ALL-NEXT: vextractf64x4 $1, %zmm0, %ymm1
300 ; ALL-NEXT: vextractf128 $1, %ymm1, %xmm1
301 ; ALL-NEXT: vbroadcastss %xmm1, %xmm1
302 ; ALL-NEXT: vblendps {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[3]
303 ; ALL-NEXT: vzeroupper
305 %res = shufflevector <16 x i32> %v, <16 x i32> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 12>
310 ;FIXME: can do better with vpcompress
311 define <4 x i32> @test_v16i32_0_4_8_12(<16 x i32> %v) {
312 ; ALL-LABEL: test_v16i32_0_4_8_12:
314 ; ALL-NEXT: vextractf128 $1, %ymm0, %xmm1
315 ; ALL-NEXT: vunpcklps {{.*#+}} xmm1 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
316 ; ALL-NEXT: vextractf64x4 $1, %zmm0, %ymm0
317 ; ALL-NEXT: vmovaps {{.*#+}} ymm2 = <u,u,0,4,u,u,u,u>
318 ; ALL-NEXT: vpermps %ymm0, %ymm2, %ymm0
319 ; ALL-NEXT: vblendps {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3]
320 ; ALL-NEXT: vzeroupper
322 %res = shufflevector <16 x i32> %v, <16 x i32> undef, <4 x i32> <i32 0, i32 4, i32 8, i32 12>
326 define <8 x float> @shuffle_v16f32_extract_256(float* %RET, float* %a) {
327 ; ALL-LABEL: shuffle_v16f32_extract_256:
329 ; ALL-NEXT: vmovups 32(%rsi), %ymm0
331 %ptr_a = bitcast float* %a to <16 x float>*
332 %v_a = load <16 x float>, <16 x float>* %ptr_a, align 4
333 %v2 = shufflevector <16 x float> %v_a, <16 x float> undef, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
337 ;FIXME: can do better with vcompressp
338 define <8 x float> @test_v16f32_0_1_2_3_4_6_7_10 (<16 x float> %v) {
339 ; ALL-LABEL: test_v16f32_0_1_2_3_4_6_7_10:
341 ; ALL-NEXT: vextractf32x4 $2, %zmm0, %xmm1
342 ; ALL-NEXT: vmovsldup {{.*#+}} xmm1 = xmm1[0,0,2,2]
343 ; ALL-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1
344 ; ALL-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,6,7,u]
345 ; ALL-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5,6],ymm1[7]
347 %res = shufflevector <16 x float> %v, <16 x float> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 6, i32 7, i32 10>
351 ;FIXME: can do better with vcompressp
352 define <4 x float> @test_v16f32_0_1_3_6 (<16 x float> %v) {
353 ; ALL-LABEL: test_v16f32_0_1_3_6:
355 ; ALL-NEXT: vpermilps {{.*#+}} xmm1 = xmm0[0,1,3,3]
356 ; ALL-NEXT: vextractf128 $1, %ymm0, %xmm0
357 ; ALL-NEXT: vpermilpd {{.*#+}} xmm0 = xmm0[1,0]
358 ; ALL-NEXT: vinsertps {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[0]
359 ; ALL-NEXT: vzeroupper
361 %res = shufflevector <16 x float> %v, <16 x float> undef, <4 x i32> <i32 0, i32 1, i32 3, i32 6>
365 define <16 x i32> @shuffle_v16i16_1_0_0_0_5_4_4_4_9_8_8_8_13_12_12_12(<16 x i32> %a, <16 x i32> %b) {
366 ; ALL-LABEL: shuffle_v16i16_1_0_0_0_5_4_4_4_9_8_8_8_13_12_12_12:
368 ; ALL-NEXT: vpermilps {{.*#+}} zmm0 = zmm0[1,0,0,0,5,4,4,4,9,8,8,8,13,12,12,12]
370 %c = shufflevector <16 x i32> %a, <16 x i32> %b, <16 x i32> <i32 1, i32 0, i32 0, i32 0, i32 5, i32 4, i32 4, i32 4, i32 9, i32 8, i32 8, i32 8, i32 13, i32 12, i32 12, i32 12>
374 define <16 x i32> @shuffle_v16i16_3_3_0_0_7_7_4_4_11_11_8_8_15_15_12_12(<16 x i32> %a, <16 x i32> %b) {
375 ; ALL-LABEL: shuffle_v16i16_3_3_0_0_7_7_4_4_11_11_8_8_15_15_12_12:
377 ; ALL-NEXT: vpermilps {{.*#+}} zmm0 = zmm0[2,3,0,1,6,7,4,5,10,11,8,9,14,15,12,13]
379 %c = shufflevector <16 x i32> %a, <16 x i32> %b, <16 x i32> <i32 2, i32 3, i32 0, i32 1, i32 6, i32 7, i32 4, i32 5, i32 10, i32 11, i32 8, i32 9, i32 14, i32 15, i32 12, i32 13>
383 define <16 x float> @shuffle_v16f32_00_01_10_10_04_05_14_14_08_09_18_18_0c_0d_1c_1c(<16 x float> %a, <16 x float> %b) {
384 ; ALL-LABEL: shuffle_v16f32_00_01_10_10_04_05_14_14_08_09_18_18_0c_0d_1c_1c:
386 ; ALL-NEXT: vshufps {{.*#+}} zmm0 = zmm0[0,1],zmm1[0,0],zmm0[4,5],zmm1[4,4],zmm0[8,9],zmm1[8,8],zmm0[12,13],zmm1[12,12]
388 %shuffle = shufflevector <16 x float> %a, <16 x float> %b, <16 x i32> <i32 0, i32 1, i32 16, i32 16, i32 4, i32 5, i32 20, i32 20, i32 8, i32 9, i32 24, i32 24, i32 12, i32 13, i32 28, i32 28>
389 ret <16 x float> %shuffle
392 define <16 x i32> @insert_mem_and_zero_v16i32(i32* %ptr) {
393 ; ALL-LABEL: insert_mem_and_zero_v16i32:
395 ; ALL-NEXT: vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
397 %a = load i32, i32* %ptr
398 %v = insertelement <16 x i32> undef, i32 %a, i32 0
399 %shuffle = shufflevector <16 x i32> %v, <16 x i32> zeroinitializer, <16 x i32> <i32 0, 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>
400 ret <16 x i32> %shuffle
404 define <16 x i32> @shuffle_v16i32_0zzzzzzzzzzzzzzz(<16 x i32> %a) {
405 ; ALL-LABEL: shuffle_v16i32_0zzzzzzzzzzzzzzz:
407 ; ALL-NEXT: vxorps %xmm1, %xmm1, %xmm1
408 ; ALL-NEXT: vblendps {{.*#+}} xmm0 = xmm0[0],xmm1[1,2,3]
410 %shuffle = shufflevector <16 x i32> %a, <16 x i32> zeroinitializer, <16 x i32> <i32 0, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16>
411 ret <16 x i32> %shuffle
414 define <16 x float> @shuffle_v16f32_0zzzzzzzzzzzzzzz(<16 x float> %a) {
415 ; ALL-LABEL: shuffle_v16f32_0zzzzzzzzzzzzzzz:
417 ; ALL-NEXT: vxorps %xmm1, %xmm1, %xmm1
418 ; ALL-NEXT: vblendps {{.*#+}} xmm0 = xmm0[0],xmm1[1,2,3]
420 %shuffle = shufflevector <16 x float> %a, <16 x float> zeroinitializer, <16 x i32> <i32 0, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16>
421 ret <16 x float> %shuffle
424 define <16 x i32> @shuffle_v16i32_16_zz_17_zz_18_zz_19_zz_20_zz_21_zz_22_zz_23_zz(<16 x i32> %a) {
425 ; ALL-LABEL: shuffle_v16i32_16_zz_17_zz_18_zz_19_zz_20_zz_21_zz_22_zz_23_zz:
427 ; ALL-NEXT: vpmovzxdq {{.*#+}} zmm0 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero
429 %shuffle = shufflevector <16 x i32> zeroinitializer, <16 x i32> %a, <16 x i32> <i32 16, i32 0, i32 17, i32 0, i32 18, i32 0, i32 19, i32 0, i32 20, i32 0, i32 21, i32 0, i32 22, i32 0, i32 23, i32 0>
430 ret <16 x i32> %shuffle
433 define <16 x i32> @shuffle_v16i32_01_02_03_04_05_06_07_08_09_10_11_12_13_14_15_16(<16 x i32> %a, <16 x i32> %b) {
434 ; ALL-LABEL: shuffle_v16i32_01_02_03_04_05_06_07_08_09_10_11_12_13_14_15_16:
436 ; ALL-NEXT: valignd {{.*#+}} zmm0 = zmm0[1,2,3,4,5,6,7,8,9,10,11,12,13,14,15],zmm1[0]
438 %shuffle = shufflevector <16 x i32> %a, <16 x i32> %b, <16 x i32><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>
439 ret <16 x i32> %shuffle
442 define <16 x i32> @shuffle_v16i32_01_02_03_04_05_06_07_08_09_10_11_12_13_14_15_00(<16 x i32> %a) {
443 ; ALL-LABEL: shuffle_v16i32_01_02_03_04_05_06_07_08_09_10_11_12_13_14_15_00:
445 ; ALL-NEXT: valignd {{.*#+}} zmm0 = zmm0[1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,0]
447 %shuffle = shufflevector <16 x i32> %a, <16 x i32> undef, <16 x i32><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>
448 ret <16 x i32> %shuffle
451 define <16 x i32> @shuffle_v16i32_00_03_16_19_04_07_20_23_08_11_24_27_12_15_28_31(<16 x i32> %a, <16 x i32> %b) {
452 ; ALL-LABEL: shuffle_v16i32_00_03_16_19_04_07_20_23_08_11_24_27_12_15_28_31:
454 ; ALL-NEXT: vshufps {{.*#+}} zmm0 = zmm0[0,3],zmm1[0,3],zmm0[4,7],zmm1[4,7],zmm0[8,11],zmm1[8,11],zmm0[12,15],zmm1[12,15]
456 %shuffle = shufflevector <16 x i32> %a, <16 x i32> %b, <16 x i32> <i32 0, i32 3, i32 16, i32 19, i32 4, i32 7, i32 20, i32 23, i32 8, i32 11, i32 24, i32 27, i32 12, i32 15, i32 28, i32 31>
457 ret <16 x i32> %shuffle
460 define <16 x i32> @shuffle_v16i32_16_16_02_03_20_20_06_07_24_24_10_11_28_28_uu_uu(<16 x i32> %a, <16 x i32> %b) {
461 ; ALL-LABEL: shuffle_v16i32_16_16_02_03_20_20_06_07_24_24_10_11_28_28_uu_uu:
463 ; ALL-NEXT: vshufps {{.*#+}} zmm0 = zmm1[0,0],zmm0[2,3],zmm1[4,4],zmm0[6,7],zmm1[8,8],zmm0[10,11],zmm1[12,12],zmm0[14,15]
465 %shuffle = shufflevector <16 x i32> %a, <16 x i32> %b, <16 x i32> <i32 16, i32 16, i32 02, i32 03, i32 20, i32 20, i32 06, i32 07, i32 24, i32 24, i32 10, i32 11, i32 28, i32 28, i32 undef, i32 undef>
466 ret <16 x i32> %shuffle
469 define <16 x i32> @shuffle_v8i32_17_16_01_00_21_20_05_04_25_24_09_08_29_28_13_12(<16 x i32> %a, <16 x i32> %b) {
470 ; ALL-LABEL: shuffle_v8i32_17_16_01_00_21_20_05_04_25_24_09_08_29_28_13_12:
472 ; ALL-NEXT: vshufps {{.*#+}} zmm0 = zmm1[1,0],zmm0[1,0],zmm1[5,4],zmm0[5,4],zmm1[9,8],zmm0[9,8],zmm1[13,12],zmm0[13,12]
474 %shuffle = shufflevector <16 x i32> %a, <16 x i32> %b, <16 x i32> <i32 17, i32 16, i32 01, i32 00, i32 21, i32 20, i32 05, i32 04, i32 25, i32 24, i32 09, i32 08, i32 29, i32 28, i32 13, i32 12>
475 ret <16 x i32> %shuffle
478 define <16 x float> @shuffle_v8f32_v16f32_04_04_04_04_04_04_04_04_04_04_04_04_04_04_04_04(<8 x float> %a) {
479 ; ALL-LABEL: shuffle_v8f32_v16f32_04_04_04_04_04_04_04_04_04_04_04_04_04_04_04_04:
481 ; ALL-NEXT: vextractf128 $1, %ymm0, %xmm0
482 ; ALL-NEXT: vbroadcastss %xmm0, %zmm0
484 %shuffle = shufflevector <8 x float> %a, <8 x float> undef, <16 x i32> <i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4>
485 ret <16 x float> %shuffle
488 define <16 x i32> @mask_shuffle_v16i32_02_03_04_05_06_07_08_09_10_11_12_13_14_15_00_01(<16 x i32> %a, <16 x i32> %passthru, i16 %mask) {
489 ; AVX512F-LABEL: mask_shuffle_v16i32_02_03_04_05_06_07_08_09_10_11_12_13_14_15_00_01:
491 ; AVX512F-NEXT: kmovw %edi, %k1
492 ; AVX512F-NEXT: valignd {{.*#+}} zmm1 {%k1} = zmm0[2,3,4,5,6,7,8,9,10,11,12,13,14,15,0,1]
493 ; AVX512F-NEXT: vmovdqa64 %zmm1, %zmm0
496 ; AVX512BW-LABEL: mask_shuffle_v16i32_02_03_04_05_06_07_08_09_10_11_12_13_14_15_00_01:
498 ; AVX512BW-NEXT: kmovd %edi, %k1
499 ; AVX512BW-NEXT: valignd {{.*#+}} zmm1 {%k1} = zmm0[2,3,4,5,6,7,8,9,10,11,12,13,14,15,0,1]
500 ; AVX512BW-NEXT: vmovdqa64 %zmm1, %zmm0
501 ; AVX512BW-NEXT: retq
502 %shuffle = shufflevector <16 x i32> %a, <16 x i32> undef, <16 x i32><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>
503 %mask.cast = bitcast i16 %mask to <16 x i1>
504 %res = select <16 x i1> %mask.cast, <16 x i32> %shuffle, <16 x i32> %passthru
508 define <16 x i32> @mask_shuffle_v16i32_02_03_04_05_06_07_08_09_10_11_12_13_14_15_16_17(<16 x i32> %a, <16 x i32> %b, <16 x i32> %passthru, i16 %mask) {
509 ; AVX512F-LABEL: mask_shuffle_v16i32_02_03_04_05_06_07_08_09_10_11_12_13_14_15_16_17:
511 ; AVX512F-NEXT: kmovw %edi, %k1
512 ; AVX512F-NEXT: valignd {{.*#+}} zmm2 {%k1} = zmm0[2,3,4,5,6,7,8,9,10,11,12,13,14,15],zmm1[0,1]
513 ; AVX512F-NEXT: vmovdqa64 %zmm2, %zmm0
516 ; AVX512BW-LABEL: mask_shuffle_v16i32_02_03_04_05_06_07_08_09_10_11_12_13_14_15_16_17:
518 ; AVX512BW-NEXT: kmovd %edi, %k1
519 ; AVX512BW-NEXT: valignd {{.*#+}} zmm2 {%k1} = zmm0[2,3,4,5,6,7,8,9,10,11,12,13,14,15],zmm1[0,1]
520 ; AVX512BW-NEXT: vmovdqa64 %zmm2, %zmm0
521 ; AVX512BW-NEXT: retq
522 %shuffle = shufflevector <16 x i32> %a, <16 x i32> %b, <16 x i32><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>
523 %mask.cast = bitcast i16 %mask to <16 x i1>
524 %res = select <16 x i1> %mask.cast, <16 x i32> %shuffle, <16 x i32> %passthru
528 define <16 x i32> @maskz_shuffle_v16i32_02_03_04_05_06_07_08_09_10_11_12_13_14_15_00_01(<16 x i32> %a, i16 %mask) {
529 ; AVX512F-LABEL: maskz_shuffle_v16i32_02_03_04_05_06_07_08_09_10_11_12_13_14_15_00_01:
531 ; AVX512F-NEXT: kmovw %edi, %k1
532 ; AVX512F-NEXT: valignd {{.*#+}} zmm0 {%k1} {z} = zmm0[2,3,4,5,6,7,8,9,10,11,12,13,14,15,0,1]
535 ; AVX512BW-LABEL: maskz_shuffle_v16i32_02_03_04_05_06_07_08_09_10_11_12_13_14_15_00_01:
537 ; AVX512BW-NEXT: kmovd %edi, %k1
538 ; AVX512BW-NEXT: valignd {{.*#+}} zmm0 {%k1} {z} = zmm0[2,3,4,5,6,7,8,9,10,11,12,13,14,15,0,1]
539 ; AVX512BW-NEXT: retq
540 %shuffle = shufflevector <16 x i32> %a, <16 x i32> undef, <16 x i32><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>
541 %mask.cast = bitcast i16 %mask to <16 x i1>
542 %res = select <16 x i1> %mask.cast, <16 x i32> %shuffle, <16 x i32> zeroinitializer
546 define <16 x i32> @maskz_shuffle_v16i32_02_03_04_05_06_07_08_09_10_11_12_13_14_15_16_17(<16 x i32> %a, <16 x i32> %b, i16 %mask) {
547 ; AVX512F-LABEL: maskz_shuffle_v16i32_02_03_04_05_06_07_08_09_10_11_12_13_14_15_16_17:
549 ; AVX512F-NEXT: kmovw %edi, %k1
550 ; AVX512F-NEXT: valignd {{.*#+}} zmm0 {%k1} {z} = zmm0[2,3,4,5,6,7,8,9,10,11,12,13,14,15],zmm1[0,1]
553 ; AVX512BW-LABEL: maskz_shuffle_v16i32_02_03_04_05_06_07_08_09_10_11_12_13_14_15_16_17:
555 ; AVX512BW-NEXT: kmovd %edi, %k1
556 ; AVX512BW-NEXT: valignd {{.*#+}} zmm0 {%k1} {z} = zmm0[2,3,4,5,6,7,8,9,10,11,12,13,14,15],zmm1[0,1]
557 ; AVX512BW-NEXT: retq
558 %shuffle = shufflevector <16 x i32> %a, <16 x i32> %b, <16 x i32><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>
559 %mask.cast = bitcast i16 %mask to <16 x i1>
560 %res = select <16 x i1> %mask.cast, <16 x i32> %shuffle, <16 x i32> zeroinitializer
564 define <16 x float> @test_vshuff32x4_512(<16 x float> %x, <16 x float> %x1) nounwind {
565 ; ALL-LABEL: test_vshuff32x4_512:
567 ; ALL-NEXT: vshuff64x2 {{.*#+}} zmm0 = zmm0[0,1,2,3],zmm1[2,3,0,1]
569 %res = shufflevector <16 x float> %x, <16 x float> %x1, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 20, i32 21, i32 22, i32 23, i32 16, i32 17, i32 18, i32 19>
570 ret <16 x float> %res
573 define <16 x i32> @test_vshufi32x4_512(<16 x i32> %x, <16 x i32> %x1) nounwind {
574 ; ALL-LABEL: test_vshufi32x4_512:
576 ; ALL-NEXT: vshufi64x2 {{.*#+}} zmm0 = zmm0[0,1,2,3],zmm1[2,3,0,1]
578 %res = shufflevector <16 x i32> %x, <16 x i32> %x1, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 20, i32 21, i32 22, i32 23, i32 16, i32 17, i32 18, i32 19>
582 define <16 x float> @test_vshuff32x4_512_mask(<16 x float> %x, <16 x float> %x1, <16 x float> %y, <16 x i1> %mask) nounwind {
583 ; AVX512F-LABEL: test_vshuff32x4_512_mask:
585 ; AVX512F-NEXT: vpmovzxbd {{.*#+}} zmm3 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero,xmm3[4],zero,zero,zero,xmm3[5],zero,zero,zero,xmm3[6],zero,zero,zero,xmm3[7],zero,zero,zero,xmm3[8],zero,zero,zero,xmm3[9],zero,zero,zero,xmm3[10],zero,zero,zero,xmm3[11],zero,zero,zero,xmm3[12],zero,zero,zero,xmm3[13],zero,zero,zero,xmm3[14],zero,zero,zero,xmm3[15],zero,zero,zero
586 ; AVX512F-NEXT: vpslld $31, %zmm3, %zmm3
587 ; AVX512F-NEXT: vpmovd2m %zmm3, %k1
588 ; AVX512F-NEXT: vshuff32x4 {{.*#+}} zmm2 {%k1} = zmm0[0,1,2,3,4,5,6,7],zmm1[4,5,6,7,0,1,2,3]
589 ; AVX512F-NEXT: vmovaps %zmm2, %zmm0
592 ; AVX512BW-LABEL: test_vshuff32x4_512_mask:
594 ; AVX512BW-NEXT: vpsllw $7, %xmm3, %xmm3
595 ; AVX512BW-NEXT: vpmovb2m %zmm3, %k1
596 ; AVX512BW-NEXT: vshuff32x4 {{.*#+}} zmm2 {%k1} = zmm0[0,1,2,3,4,5,6,7],zmm1[4,5,6,7,0,1,2,3]
597 ; AVX512BW-NEXT: vmovaps %zmm2, %zmm0
598 ; AVX512BW-NEXT: retq
599 %x2 = shufflevector <16 x float> %x, <16 x float> %x1, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 20, i32 21, i32 22, i32 23, i32 16, i32 17, i32 18, i32 19>
600 %res = select <16 x i1> %mask, <16 x float> %x2, <16 x float> %y
601 ret <16 x float> %res
604 define <16 x i32> @test_vshufi32x4_512_mask(<16 x i32> %x, <16 x i32> %x1, <16 x i32> %y, <16 x i1> %mask) nounwind {
605 ; AVX512F-LABEL: test_vshufi32x4_512_mask:
607 ; AVX512F-NEXT: vpmovzxbd {{.*#+}} zmm3 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero,xmm3[4],zero,zero,zero,xmm3[5],zero,zero,zero,xmm3[6],zero,zero,zero,xmm3[7],zero,zero,zero,xmm3[8],zero,zero,zero,xmm3[9],zero,zero,zero,xmm3[10],zero,zero,zero,xmm3[11],zero,zero,zero,xmm3[12],zero,zero,zero,xmm3[13],zero,zero,zero,xmm3[14],zero,zero,zero,xmm3[15],zero,zero,zero
608 ; AVX512F-NEXT: vpslld $31, %zmm3, %zmm3
609 ; AVX512F-NEXT: vpmovd2m %zmm3, %k1
610 ; AVX512F-NEXT: vshufi32x4 {{.*#+}} zmm2 {%k1} = zmm0[0,1,2,3,4,5,6,7],zmm1[4,5,6,7,0,1,2,3]
611 ; AVX512F-NEXT: vmovdqa64 %zmm2, %zmm0
614 ; AVX512BW-LABEL: test_vshufi32x4_512_mask:
616 ; AVX512BW-NEXT: vpsllw $7, %xmm3, %xmm3
617 ; AVX512BW-NEXT: vpmovb2m %zmm3, %k1
618 ; AVX512BW-NEXT: vshufi32x4 {{.*#+}} zmm2 {%k1} = zmm0[0,1,2,3,4,5,6,7],zmm1[4,5,6,7,0,1,2,3]
619 ; AVX512BW-NEXT: vmovdqa64 %zmm2, %zmm0
620 ; AVX512BW-NEXT: retq
621 %x2 = shufflevector <16 x i32> %x, <16 x i32> %x1, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 20, i32 21, i32 22, i32 23, i32 16, i32 17, i32 18, i32 19>
622 %res = select <16 x i1> %mask, <16 x i32> %x2, <16 x i32> %y
626 define <16 x float> @mask_shuffle_v16f32_00_01_02_03_04_05_06_07_16_17_18_19_20_21_22_23(<16 x float> %a, <16 x float> %b, <16 x float> %passthru, i16 %mask) {
627 ; AVX512F-LABEL: mask_shuffle_v16f32_00_01_02_03_04_05_06_07_16_17_18_19_20_21_22_23:
629 ; AVX512F-NEXT: kmovw %edi, %k1
630 ; AVX512F-NEXT: vinsertf32x8 $1, %ymm1, %zmm0, %zmm2 {%k1}
631 ; AVX512F-NEXT: vmovaps %zmm2, %zmm0
634 ; AVX512BW-LABEL: mask_shuffle_v16f32_00_01_02_03_04_05_06_07_16_17_18_19_20_21_22_23:
636 ; AVX512BW-NEXT: kmovd %edi, %k1
637 ; AVX512BW-NEXT: vinsertf32x8 $1, %ymm1, %zmm0, %zmm2 {%k1}
638 ; AVX512BW-NEXT: vmovaps %zmm2, %zmm0
639 ; AVX512BW-NEXT: retq
640 %shuffle = shufflevector <16 x float> %a, <16 x float> %b, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23>
641 %mask.cast = bitcast i16 %mask to <16 x i1>
642 %res = select <16 x i1> %mask.cast, <16 x float> %shuffle, <16 x float> %passthru
643 ret <16 x float> %res
646 define <16 x float> @mask_shuffle_v16f32_00_01_02_03_16_17_18_19_08_09_10_11_12_13_14_15(<16 x float> %a, <16 x float> %b, <16 x float> %passthru, i16 %mask) {
647 ; AVX512F-LABEL: mask_shuffle_v16f32_00_01_02_03_16_17_18_19_08_09_10_11_12_13_14_15:
649 ; AVX512F-NEXT: kmovw %edi, %k1
650 ; AVX512F-NEXT: vinsertf32x4 $1, %xmm1, %zmm0, %zmm2 {%k1}
651 ; AVX512F-NEXT: vmovaps %zmm2, %zmm0
654 ; AVX512BW-LABEL: mask_shuffle_v16f32_00_01_02_03_16_17_18_19_08_09_10_11_12_13_14_15:
656 ; AVX512BW-NEXT: kmovd %edi, %k1
657 ; AVX512BW-NEXT: vinsertf32x4 $1, %xmm1, %zmm0, %zmm2 {%k1}
658 ; AVX512BW-NEXT: vmovaps %zmm2, %zmm0
659 ; AVX512BW-NEXT: retq
660 %shuffle = shufflevector <16 x float> %a, <16 x float> %b, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
661 %mask.cast = bitcast i16 %mask to <16 x i1>
662 %res = select <16 x i1> %mask.cast, <16 x float> %shuffle, <16 x float> %passthru
663 ret <16 x float> %res
666 define <16 x i32> @mask_shuffle_v16i32_00_01_02_03_04_05_06_07_16_17_18_19_20_21_22_23(<16 x i32> %a, <16 x i32> %b, <16 x i32> %passthru, i16 %mask) {
667 ; AVX512F-LABEL: mask_shuffle_v16i32_00_01_02_03_04_05_06_07_16_17_18_19_20_21_22_23:
669 ; AVX512F-NEXT: kmovw %edi, %k1
670 ; AVX512F-NEXT: vinserti32x8 $1, %ymm1, %zmm0, %zmm2 {%k1}
671 ; AVX512F-NEXT: vmovdqa64 %zmm2, %zmm0
674 ; AVX512BW-LABEL: mask_shuffle_v16i32_00_01_02_03_04_05_06_07_16_17_18_19_20_21_22_23:
676 ; AVX512BW-NEXT: kmovd %edi, %k1
677 ; AVX512BW-NEXT: vinserti32x8 $1, %ymm1, %zmm0, %zmm2 {%k1}
678 ; AVX512BW-NEXT: vmovdqa64 %zmm2, %zmm0
679 ; AVX512BW-NEXT: retq
680 %shuffle = shufflevector <16 x i32> %a, <16 x i32> %b, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23>
681 %mask.cast = bitcast i16 %mask to <16 x i1>
682 %res = select <16 x i1> %mask.cast, <16 x i32> %shuffle, <16 x i32> %passthru
686 define <16 x i32> @mask_shuffle_v16i32_00_01_02_03_16_17_18_19_08_09_10_11_12_13_14_15(<16 x i32> %a, <16 x i32> %b, <16 x i32> %passthru, i16 %mask) {
687 ; AVX512F-LABEL: mask_shuffle_v16i32_00_01_02_03_16_17_18_19_08_09_10_11_12_13_14_15:
689 ; AVX512F-NEXT: kmovw %edi, %k1
690 ; AVX512F-NEXT: vinserti32x4 $1, %xmm1, %zmm0, %zmm2 {%k1}
691 ; AVX512F-NEXT: vmovdqa64 %zmm2, %zmm0
694 ; AVX512BW-LABEL: mask_shuffle_v16i32_00_01_02_03_16_17_18_19_08_09_10_11_12_13_14_15:
696 ; AVX512BW-NEXT: kmovd %edi, %k1
697 ; AVX512BW-NEXT: vinserti32x4 $1, %xmm1, %zmm0, %zmm2 {%k1}
698 ; AVX512BW-NEXT: vmovdqa64 %zmm2, %zmm0
699 ; AVX512BW-NEXT: retq
700 %shuffle = shufflevector <16 x i32> %a, <16 x i32> %b, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
701 %mask.cast = bitcast i16 %mask to <16 x i1>
702 %res = select <16 x i1> %mask.cast, <16 x i32> %shuffle, <16 x i32> %passthru
706 define <16 x i32> @mask_shuffle_v4i32_v16i32_00_01_02_03_00_01_02_03_00_01_02_03_00_01_02_03(<4 x i32> %a) {
707 ; ALL-LABEL: mask_shuffle_v4i32_v16i32_00_01_02_03_00_01_02_03_00_01_02_03_00_01_02_03:
709 ; ALL-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0
710 ; ALL-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
711 ; ALL-NEXT: vinsertf64x4 $1, %ymm0, %zmm0, %zmm0
713 %res = shufflevector <4 x i32> %a, <4 x i32> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3>
717 define <16 x float> @mask_shuffle_v4f32_v16f32_00_01_02_03_00_01_02_03_00_01_02_03_00_01_02_03(<4 x float> %a) {
718 ; ALL-LABEL: mask_shuffle_v4f32_v16f32_00_01_02_03_00_01_02_03_00_01_02_03_00_01_02_03:
720 ; ALL-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0
721 ; ALL-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
722 ; ALL-NEXT: vinsertf64x4 $1, %ymm0, %zmm0, %zmm0
724 %res = shufflevector <4 x float> %a, <4 x float> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3>
725 ret <16 x float> %res