1; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py 2; RUN: llc -mtriple=thumbv8.1m.main-none-none-eabi -mattr=+mve.fp,+fp64 -mve-max-interleave-factor=4 -verify-machineinstrs %s -o - | FileCheck %s 3 4; i32 5 6define ptr @vst4_v4i32(ptr %src, ptr %dst) { 7; CHECK-LABEL: vst4_v4i32: 8; CHECK: @ %bb.0: @ %entry 9; CHECK-NEXT: vldrw.u32 q3, [r0, #48] 10; CHECK-NEXT: vldrw.u32 q2, [r0, #32] 11; CHECK-NEXT: vldrw.u32 q1, [r0, #16] 12; CHECK-NEXT: vldrw.u32 q0, [r0] 13; CHECK-NEXT: vst40.32 {q0, q1, q2, q3}, [r1] 14; CHECK-NEXT: vst41.32 {q0, q1, q2, q3}, [r1] 15; CHECK-NEXT: vst42.32 {q0, q1, q2, q3}, [r1] 16; CHECK-NEXT: vst43.32 {q0, q1, q2, q3}, [r1]! 17; CHECK-NEXT: mov r0, r1 18; CHECK-NEXT: bx lr 19entry: 20 %l1 = load <4 x i32>, ptr %src, align 4 21 %s2 = getelementptr <4 x i32>, ptr %src, i32 1 22 %l2 = load <4 x i32>, ptr %s2, align 4 23 %s3 = getelementptr <4 x i32>, ptr %src, i32 2 24 %l3 = load <4 x i32>, ptr %s3, align 4 25 %s4 = getelementptr <4 x i32>, ptr %src, i32 3 26 %l4 = load <4 x i32>, ptr %s4, align 4 27 %t1 = shufflevector <4 x i32> %l1, <4 x i32> %l2, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7> 28 %t2 = shufflevector <4 x i32> %l3, <4 x i32> %l4, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7> 29 %s = shufflevector <8 x i32> %t1, <8 x i32> %t2, <16 x i32> <i32 0, i32 4, i32 8, i32 12, i32 1, i32 5, i32 9, i32 13, i32 2, i32 6, i32 10, i32 14, i32 3, i32 7, i32 11, i32 15> 30 store <16 x i32> %s, ptr %dst 31 %ret = getelementptr inbounds <16 x i32>, ptr %dst, i32 1 32 ret ptr %ret 33} 34 35; i16 36 37define ptr @vst4_v8i16(ptr %src, ptr %dst) { 38; CHECK-LABEL: vst4_v8i16: 39; CHECK: @ %bb.0: @ %entry 40; CHECK-NEXT: vldrw.u32 q3, [r0, #48] 41; CHECK-NEXT: vldrw.u32 q2, [r0, #32] 42; CHECK-NEXT: vldrw.u32 q1, [r0, #16] 43; CHECK-NEXT: vldrw.u32 q0, [r0] 44; CHECK-NEXT: vst40.16 {q0, q1, q2, q3}, [r1] 45; CHECK-NEXT: vst41.16 {q0, q1, q2, q3}, [r1] 46; CHECK-NEXT: vst42.16 {q0, q1, q2, q3}, [r1] 47; CHECK-NEXT: vst43.16 {q0, q1, q2, q3}, [r1]! 48; CHECK-NEXT: mov r0, r1 49; CHECK-NEXT: bx lr 50entry: 51 %l1 = load <8 x i16>, ptr %src, align 4 52 %s2 = getelementptr <8 x i16>, ptr %src, i32 1 53 %l2 = load <8 x i16>, ptr %s2, align 4 54 %s3 = getelementptr <8 x i16>, ptr %src, i32 2 55 %l3 = load <8 x i16>, ptr %s3, align 4 56 %s4 = getelementptr <8 x i16>, ptr %src, i32 3 57 %l4 = load <8 x i16>, ptr %s4, align 4 58 %t1 = shufflevector <8 x i16> %l1, <8 x i16> %l2, <16 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> 59 %t2 = shufflevector <8 x i16> %l3, <8 x i16> %l4, <16 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> 60 %s = shufflevector <16 x i16> %t1, <16 x i16> %t2, <32 x i32> <i32 0, i32 8, i32 16, i32 24, i32 1, i32 9, i32 17, i32 25, i32 2, i32 10, i32 18, i32 26, i32 3, i32 11, i32 19, i32 27, i32 4, i32 12, i32 20, i32 28, i32 5, i32 13, i32 21, i32 29, i32 6, i32 14, i32 22, i32 30, i32 7, i32 15, i32 23, i32 31> 61 store <32 x i16> %s, ptr %dst 62 %ret = getelementptr inbounds <32 x i16>, ptr %dst, i32 1 63 ret ptr %ret 64} 65 66; i8 67 68define ptr @vst4_v16i8(ptr %src, ptr %dst) { 69; CHECK-LABEL: vst4_v16i8: 70; CHECK: @ %bb.0: @ %entry 71; CHECK-NEXT: vldrw.u32 q3, [r0, #48] 72; CHECK-NEXT: vldrw.u32 q2, [r0, #32] 73; CHECK-NEXT: vldrw.u32 q1, [r0, #16] 74; CHECK-NEXT: vldrw.u32 q0, [r0] 75; CHECK-NEXT: vst40.8 {q0, q1, q2, q3}, [r1] 76; CHECK-NEXT: vst41.8 {q0, q1, q2, q3}, [r1] 77; CHECK-NEXT: vst42.8 {q0, q1, q2, q3}, [r1] 78; CHECK-NEXT: vst43.8 {q0, q1, q2, q3}, [r1]! 79; CHECK-NEXT: mov r0, r1 80; CHECK-NEXT: bx lr 81entry: 82 %l1 = load <16 x i8>, ptr %src, align 4 83 %s2 = getelementptr <16 x i8>, ptr %src, i32 1 84 %l2 = load <16 x i8>, ptr %s2, align 4 85 %s3 = getelementptr <16 x i8>, ptr %src, i32 2 86 %l3 = load <16 x i8>, ptr %s3, align 4 87 %s4 = getelementptr <16 x i8>, ptr %src, i32 3 88 %l4 = load <16 x i8>, ptr %s4, align 4 89 %t1 = shufflevector <16 x i8> %l1, <16 x i8> %l2, <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 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> 90 %t2 = shufflevector <16 x i8> %l3, <16 x i8> %l4, <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 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> 91 %s = shufflevector <32 x i8> %t1, <32 x i8> %t2, <64 x i32> <i32 0, i32 16, i32 32, i32 48, i32 1, i32 17, i32 33, i32 49, i32 2, i32 18, i32 34, i32 50, i32 3, i32 19, i32 35, i32 51, i32 4, i32 20, i32 36, i32 52, i32 5, i32 21, i32 37, i32 53, i32 6, i32 22, i32 38, i32 54, i32 7, i32 23, i32 39, i32 55, i32 8, i32 24, i32 40, i32 56, i32 9, i32 25, i32 41, i32 57, i32 10, i32 26, i32 42, i32 58, i32 11, i32 27, i32 43, i32 59, i32 12, i32 28, i32 44, i32 60, i32 13, i32 29, i32 45, i32 61, i32 14, i32 30, i32 46, i32 62, i32 15, i32 31, i32 47, i32 63> 92 store <64 x i8> %s, ptr %dst 93 %ret = getelementptr inbounds <64 x i8>, ptr %dst, i32 1 94 ret ptr %ret 95} 96 97; i64 98 99define ptr @vst4_v2i64(ptr %src, ptr %dst) { 100; CHECK-LABEL: vst4_v2i64: 101; CHECK: @ %bb.0: @ %entry 102; CHECK-NEXT: .vsave {d8, d9} 103; CHECK-NEXT: vpush {d8, d9} 104; CHECK-NEXT: vldrw.u32 q0, [r0, #16] 105; CHECK-NEXT: vldrw.u32 q3, [r0] 106; CHECK-NEXT: vldrw.u32 q2, [r0, #48] 107; CHECK-NEXT: vldrw.u32 q4, [r0, #32] 108; CHECK-NEXT: vmov.f64 d2, d6 109; CHECK-NEXT: vmov.f64 d3, d0 110; CHECK-NEXT: vmov.f64 d0, d7 111; CHECK-NEXT: vmov.f64 d7, d4 112; CHECK-NEXT: vstrw.32 q0, [r1, #32] 113; CHECK-NEXT: vmov.f64 d6, d8 114; CHECK-NEXT: vmov.f64 d4, d9 115; CHECK-NEXT: vstrw.32 q3, [r1, #16] 116; CHECK-NEXT: vstrw.32 q2, [r1, #48] 117; CHECK-NEXT: vstrw.32 q1, [r1], #64 118; CHECK-NEXT: mov r0, r1 119; CHECK-NEXT: vpop {d8, d9} 120; CHECK-NEXT: bx lr 121entry: 122 %l1 = load <2 x i64>, ptr %src, align 4 123 %s2 = getelementptr <2 x i64>, ptr %src, i32 1 124 %l2 = load <2 x i64>, ptr %s2, align 4 125 %s3 = getelementptr <2 x i64>, ptr %src, i32 2 126 %l3 = load <2 x i64>, ptr %s3, align 4 127 %s4 = getelementptr <2 x i64>, ptr %src, i32 3 128 %l4 = load <2 x i64>, ptr %s4, align 4 129 %t1 = shufflevector <2 x i64> %l1, <2 x i64> %l2, <4 x i32> <i32 0, i32 1, i32 2, i32 3> 130 %t2 = shufflevector <2 x i64> %l3, <2 x i64> %l4, <4 x i32> <i32 0, i32 1, i32 2, i32 3> 131 %s = shufflevector <4 x i64> %t1, <4 x i64> %t2, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 1, i32 3, i32 5, i32 7> 132 store <8 x i64> %s, ptr %dst 133 %ret = getelementptr inbounds <8 x i64>, ptr %dst, i32 1 134 ret ptr %ret 135} 136 137; f32 138 139define ptr @vst4_v4f32(ptr %src, ptr %dst) { 140; CHECK-LABEL: vst4_v4f32: 141; CHECK: @ %bb.0: @ %entry 142; CHECK-NEXT: vldrw.u32 q3, [r0, #48] 143; CHECK-NEXT: vldrw.u32 q2, [r0, #32] 144; CHECK-NEXT: vldrw.u32 q1, [r0, #16] 145; CHECK-NEXT: vldrw.u32 q0, [r0] 146; CHECK-NEXT: vst40.32 {q0, q1, q2, q3}, [r1] 147; CHECK-NEXT: vst41.32 {q0, q1, q2, q3}, [r1] 148; CHECK-NEXT: vst42.32 {q0, q1, q2, q3}, [r1] 149; CHECK-NEXT: vst43.32 {q0, q1, q2, q3}, [r1]! 150; CHECK-NEXT: mov r0, r1 151; CHECK-NEXT: bx lr 152entry: 153 %l1 = load <4 x float>, ptr %src, align 4 154 %s2 = getelementptr <4 x float>, ptr %src, i32 1 155 %l2 = load <4 x float>, ptr %s2, align 4 156 %s3 = getelementptr <4 x float>, ptr %src, i32 2 157 %l3 = load <4 x float>, ptr %s3, align 4 158 %s4 = getelementptr <4 x float>, ptr %src, i32 3 159 %l4 = load <4 x float>, ptr %s4, align 4 160 %t1 = shufflevector <4 x float> %l1, <4 x float> %l2, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7> 161 %t2 = shufflevector <4 x float> %l3, <4 x float> %l4, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7> 162 %s = shufflevector <8 x float> %t1, <8 x float> %t2, <16 x i32> <i32 0, i32 4, i32 8, i32 12, i32 1, i32 5, i32 9, i32 13, i32 2, i32 6, i32 10, i32 14, i32 3, i32 7, i32 11, i32 15> 163 store <16 x float> %s, ptr %dst 164 %ret = getelementptr inbounds <16 x float>, ptr %dst, i32 1 165 ret ptr %ret 166} 167 168; f16 169 170define ptr @vst4_v8f16(ptr %src, ptr %dst) { 171; CHECK-LABEL: vst4_v8f16: 172; CHECK: @ %bb.0: @ %entry 173; CHECK-NEXT: vldrw.u32 q3, [r0, #48] 174; CHECK-NEXT: vldrw.u32 q2, [r0, #32] 175; CHECK-NEXT: vldrw.u32 q1, [r0, #16] 176; CHECK-NEXT: vldrw.u32 q0, [r0] 177; CHECK-NEXT: vst40.16 {q0, q1, q2, q3}, [r1] 178; CHECK-NEXT: vst41.16 {q0, q1, q2, q3}, [r1] 179; CHECK-NEXT: vst42.16 {q0, q1, q2, q3}, [r1] 180; CHECK-NEXT: vst43.16 {q0, q1, q2, q3}, [r1]! 181; CHECK-NEXT: mov r0, r1 182; CHECK-NEXT: bx lr 183entry: 184 %l1 = load <8 x half>, ptr %src, align 4 185 %s2 = getelementptr <8 x half>, ptr %src, i32 1 186 %l2 = load <8 x half>, ptr %s2, align 4 187 %s3 = getelementptr <8 x half>, ptr %src, i32 2 188 %l3 = load <8 x half>, ptr %s3, align 4 189 %s4 = getelementptr <8 x half>, ptr %src, i32 3 190 %l4 = load <8 x half>, ptr %s4, align 4 191 %t1 = shufflevector <8 x half> %l1, <8 x half> %l2, <16 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> 192 %t2 = shufflevector <8 x half> %l3, <8 x half> %l4, <16 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> 193 %s = shufflevector <16 x half> %t1, <16 x half> %t2, <32 x i32> <i32 0, i32 8, i32 16, i32 24, i32 1, i32 9, i32 17, i32 25, i32 2, i32 10, i32 18, i32 26, i32 3, i32 11, i32 19, i32 27, i32 4, i32 12, i32 20, i32 28, i32 5, i32 13, i32 21, i32 29, i32 6, i32 14, i32 22, i32 30, i32 7, i32 15, i32 23, i32 31> 194 store <32 x half> %s, ptr %dst 195 %ret = getelementptr inbounds <32 x half>, ptr %dst, i32 1 196 ret ptr %ret 197} 198 199; f64 200 201define ptr @vst4_v2f64(ptr %src, ptr %dst) { 202; CHECK-LABEL: vst4_v2f64: 203; CHECK: @ %bb.0: @ %entry 204; CHECK-NEXT: .vsave {d8, d9} 205; CHECK-NEXT: vpush {d8, d9} 206; CHECK-NEXT: vldrw.u32 q0, [r0, #16] 207; CHECK-NEXT: vldrw.u32 q3, [r0] 208; CHECK-NEXT: vldrw.u32 q2, [r0, #48] 209; CHECK-NEXT: vldrw.u32 q4, [r0, #32] 210; CHECK-NEXT: vmov.f64 d2, d6 211; CHECK-NEXT: vmov.f64 d3, d0 212; CHECK-NEXT: vmov.f64 d0, d7 213; CHECK-NEXT: vmov.f64 d7, d4 214; CHECK-NEXT: vstrw.32 q0, [r1, #32] 215; CHECK-NEXT: vmov.f64 d6, d8 216; CHECK-NEXT: vmov.f64 d4, d9 217; CHECK-NEXT: vstrw.32 q3, [r1, #16] 218; CHECK-NEXT: vstrw.32 q2, [r1, #48] 219; CHECK-NEXT: vstrw.32 q1, [r1], #64 220; CHECK-NEXT: mov r0, r1 221; CHECK-NEXT: vpop {d8, d9} 222; CHECK-NEXT: bx lr 223entry: 224 %l1 = load <2 x double>, ptr %src, align 4 225 %s2 = getelementptr <2 x double>, ptr %src, i32 1 226 %l2 = load <2 x double>, ptr %s2, align 4 227 %s3 = getelementptr <2 x double>, ptr %src, i32 2 228 %l3 = load <2 x double>, ptr %s3, align 4 229 %s4 = getelementptr <2 x double>, ptr %src, i32 3 230 %l4 = load <2 x double>, ptr %s4, align 4 231 %t1 = shufflevector <2 x double> %l1, <2 x double> %l2, <4 x i32> <i32 0, i32 1, i32 2, i32 3> 232 %t2 = shufflevector <2 x double> %l3, <2 x double> %l4, <4 x i32> <i32 0, i32 1, i32 2, i32 3> 233 %s = shufflevector <4 x double> %t1, <4 x double> %t2, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 1, i32 3, i32 5, i32 7> 234 store <8 x double> %s, ptr %dst 235 %ret = getelementptr inbounds <8 x double>, ptr %dst, i32 1 236 ret ptr %ret 237} 238