1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py 2 ; RUN: llc < %s -mcpu=x86-64 -mattr=+avx512f | FileCheck %s --check-prefix=ALL --check-prefix=AVX512 --check-prefix=AVX512F 3 ; RUN: llc < %s -mcpu=x86-64 -mattr=+avx512bw | FileCheck %s --check-prefix=ALL --check-prefix=AVX512 --check-prefix=AVX512BW 4 5 target triple = "x86_64-unknown-unknown" 6 7 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) { 8 ; ALL-LABEL: shuffle_v16f32_00_10_01_11_04_14_05_15_08_18_09_19_0c_1c_0d_1d: 9 ; ALL: # BB#0: 10 ; 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] 11 ; ALL-NEXT: retq 12 %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> 13 ret <16 x float> %shuffle 14 } 15 16 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) { 17 ; ALL-LABEL: shuffle_v16f32_00_zz_01_zz_04_zz_05_zz_08_zz_09_zz_0c_zz_0d_zz: 18 ; ALL: # BB#0: 19 ; ALL-NEXT: vpxord %zmm1, %zmm1, %zmm1 20 ; 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] 21 ; ALL-NEXT: retq 22 %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> 23 ret <16 x float> %shuffle 24 } 25 26 define <16 x float> @shuffle_v16f32_vunpcklps_swap(<16 x float> %a, <16 x float> %b) { 27 ; ALL-LABEL: shuffle_v16f32_vunpcklps_swap: 28 ; ALL: # BB#0: 29 ; 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] 30 ; ALL-NEXT: retq 31 %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> 32 ret <16 x float> %shuffle 33 } 34 35 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) { 36 ; ALL-LABEL: shuffle_v16i32_00_10_01_11_04_14_05_15_08_18_09_19_0c_1c_0d_1d: 37 ; ALL: # BB#0: 38 ; ALL-NEXT: vpunpckldq {{.*#+}} 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] 39 ; ALL-NEXT: retq 40 %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> 41 ret <16 x i32> %shuffle 42 } 43 44 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) { 45 ; ALL-LABEL: shuffle_v16i32_zz_10_zz_11_zz_14_zz_15_zz_18_zz_19_zz_1c_zz_1d: 46 ; ALL: # BB#0: 47 ; ALL-NEXT: vpxord %zmm0, %zmm0, %zmm0 48 ; ALL-NEXT: vpunpckldq {{.*#+}} 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] 49 ; ALL-NEXT: retq 50 %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> 51 ret <16 x i32> %shuffle 52 } 53 54 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) { 55 ; ALL-LABEL: shuffle_v16f32_02_12_03_13_06_16_07_17_0a_1a_0b_1b_0e_1e_0f_1f: 56 ; ALL: # BB#0: 57 ; 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] 58 ; ALL-NEXT: retq 59 %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> 60 ret <16 x float> %shuffle 61 } 62 63 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) { 64 ; ALL-LABEL: shuffle_v16f32_zz_12_zz_13_zz_16_zz_17_zz_1a_zz_1b_zz_1e_zz_1f: 65 ; ALL: # BB#0: 66 ; ALL-NEXT: vpxord %zmm0, %zmm0, %zmm0 67 ; 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] 68 ; ALL-NEXT: retq 69 %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> 70 ret <16 x float> %shuffle 71 } 72 73 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) { 74 ; ALL-LABEL: shuffle_v16i32_02_12_03_13_06_16_07_17_0a_1a_0b_1b_0e_1e_0f_1f: 75 ; ALL: # BB#0: 76 ; ALL-NEXT: vpunpckhdq {{.*#+}} 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] 77 ; ALL-NEXT: retq 78 %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> 79 ret <16 x i32> %shuffle 80 } 81 82 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) { 83 ; ALL-LABEL: shuffle_v16i32_02_zz_03_zz_06_zz_07_zz_0a_zz_0b_zz_0e_zz_0f_zz: 84 ; ALL: # BB#0: 85 ; ALL-NEXT: vpxord %zmm1, %zmm1, %zmm1 86 ; ALL-NEXT: vpunpckhdq {{.*#+}} 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] 87 ; ALL-NEXT: retq 88 %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> 89 ret <16 x i32> %shuffle 90 } 91 92 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) { 93 ; ALL-LABEL: shuffle_v16f32_02_05_u_u_07_u_0a_01_00_05_u_04_07_u_0a_01: 94 ; ALL: # BB#0: 95 ; ALL-NEXT: vmovdqa32 {{.*#+}} zmm1 = <2,5,u,u,7,u,10,1,0,5,u,4,7,u,10,1> 96 ; ALL-NEXT: vpermps %zmm0, %zmm1, %zmm0 97 ; ALL-NEXT: retq 98 %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> 99 ret <16 x float> %c 100 } 101 102 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) { 103 ; ALL-LABEL: shuffle_v16i32_02_05_u_u_07_u_0a_01_00_05_u_04_07_u_0a_01: 104 ; ALL: # BB#0: 105 ; ALL-NEXT: vmovdqa32 {{.*#+}} zmm1 = <2,5,u,u,7,u,10,1,0,5,u,4,7,u,10,1> 106 ; ALL-NEXT: vpermd %zmm0, %zmm1, %zmm0 107 ; ALL-NEXT: retq 108 %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> 109 ret <16 x i32> %c 110 } 111 112 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) { 113 ; ALL-LABEL: shuffle_v16i32_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_18: 114 ; ALL: # BB#0: 115 ; ALL-NEXT: vmovdqa32 {{.*#+}} zmm2 = [15,31,14,22,13,29,4,28,11,27,10,26,9,25,8,24] 116 ; ALL-NEXT: vpermt2d %zmm1, %zmm2, %zmm0 117 ; ALL-NEXT: retq 118 %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> 119 ret <16 x i32> %c 120 } 121 122 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) { 123 ; ALL-LABEL: shuffle_v16f32_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_18: 124 ; ALL: # BB#0: 125 ; ALL-NEXT: vmovdqa32 {{.*#+}} zmm2 = [15,31,14,22,13,29,4,28,11,27,10,26,9,25,8,24] 126 ; ALL-NEXT: vpermt2ps %zmm1, %zmm2, %zmm0 127 ; ALL-NEXT: retq 128 %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> 129 ret <16 x float> %c 130 } 131 132 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) { 133 ; ALL-LABEL: shuffle_v16f32_load_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_18: 134 ; ALL: # BB#0: 135 ; ALL-NEXT: vmovdqa32 {{.*#+}} zmm1 = [15,31,14,22,13,29,4,28,11,27,10,26,9,25,8,24] 136 ; ALL-NEXT: vpermt2ps (%rdi), %zmm1, %zmm0 137 ; ALL-NEXT: retq 138 %c = load <16 x float>, <16 x float>* %b 139 %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> 140 ret <16 x float> %d 141 } 142 143 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) { 144 ; ALL-LABEL: shuffle_v16i32_load_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_18: 145 ; ALL: # BB#0: 146 ; ALL-NEXT: vmovdqa32 {{.*#+}} zmm1 = [15,31,14,22,13,29,4,28,11,27,10,26,9,25,8,24] 147 ; ALL-NEXT: vpermt2d (%rdi), %zmm1, %zmm0 148 ; ALL-NEXT: retq 149 %c = load <16 x i32>, <16 x i32>* %b 150 %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> 151 ret <16 x i32> %d 152 } 153 154 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) { 155 ; ALL-LABEL: shuffle_v16i32_0_1_2_19_u_u_u_u_u_u_u_u_u_u_u_u: 156 ; ALL: # BB#0: 157 ; ALL-NEXT: vmovdqa32 {{.*#+}} zmm2 = <0,1,2,19,u,u,u,u,u,u,u,u,u,u,u,u> 158 ; ALL-NEXT: vpermt2d %zmm1, %zmm2, %zmm0 159 ; ALL-NEXT: retq 160 %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> 161 ret <16 x i32> %c 162 } 163 164 define <8 x float> @shuffle_v16f32_extract_256(float* %RET, float* %a) { 165 ; ALL-LABEL: shuffle_v16f32_extract_256: 166 ; ALL: # BB#0: 167 ; ALL-NEXT: vmovups (%rsi), %zmm0 168 ; ALL-NEXT: vextractf64x4 $1, %zmm0, %ymm0 169 ; ALL-NEXT: retq 170 %ptr_a = bitcast float* %a to <16 x float>* 171 %v_a = load <16 x float>, <16 x float>* %ptr_a, align 4 172 %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> 173 ret <8 x float> %v2 174 } 175
