1; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py 2; RUN: llc -mtriple=riscv32 -mattr=+m,+experimental-v -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=8 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,LMULMAX8 3; RUN: llc -mtriple=riscv64 -mattr=+m,+experimental-v -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=8 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,LMULMAX8 4; RUN: llc -mtriple=riscv32 -mattr=+m,+experimental-v -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=2 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,LMULMAX2 5; RUN: llc -mtriple=riscv64 -mattr=+m,+experimental-v -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=2 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,LMULMAX2 6; RUN: llc -mtriple=riscv32 -mattr=+m,+experimental-v -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=1 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,LMULMAX1 7; RUN: llc -mtriple=riscv64 -mattr=+m,+experimental-v -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=1 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,LMULMAX1 8 9define void @sext_v4i8_v4i32(<4 x i8>* %x, <4 x i32>* %z) { 10; CHECK-LABEL: sext_v4i8_v4i32: 11; CHECK: # %bb.0: 12; CHECK-NEXT: vsetivli zero, 4, e8, mf4, ta, mu 13; CHECK-NEXT: vle8.v v25, (a0) 14; CHECK-NEXT: vsetvli zero, zero, e32, m1, ta, mu 15; CHECK-NEXT: vsext.vf4 v26, v25 16; CHECK-NEXT: vse32.v v26, (a1) 17; CHECK-NEXT: ret 18 %a = load <4 x i8>, <4 x i8>* %x 19 %b = sext <4 x i8> %a to <4 x i32> 20 store <4 x i32> %b, <4 x i32>* %z 21 ret void 22} 23 24define void @zext_v4i8_v4i32(<4 x i8>* %x, <4 x i32>* %z) { 25; CHECK-LABEL: zext_v4i8_v4i32: 26; CHECK: # %bb.0: 27; CHECK-NEXT: vsetivli zero, 4, e8, mf4, ta, mu 28; CHECK-NEXT: vle8.v v25, (a0) 29; CHECK-NEXT: vsetvli zero, zero, e32, m1, ta, mu 30; CHECK-NEXT: vzext.vf4 v26, v25 31; CHECK-NEXT: vse32.v v26, (a1) 32; CHECK-NEXT: ret 33 %a = load <4 x i8>, <4 x i8>* %x 34 %b = zext <4 x i8> %a to <4 x i32> 35 store <4 x i32> %b, <4 x i32>* %z 36 ret void 37} 38 39define void @sext_v8i8_v8i32(<8 x i8>* %x, <8 x i32>* %z) { 40; LMULMAX8-LABEL: sext_v8i8_v8i32: 41; LMULMAX8: # %bb.0: 42; LMULMAX8-NEXT: vsetivli zero, 8, e8, mf2, ta, mu 43; LMULMAX8-NEXT: vle8.v v25, (a0) 44; LMULMAX8-NEXT: vsetvli zero, zero, e32, m2, ta, mu 45; LMULMAX8-NEXT: vsext.vf4 v26, v25 46; LMULMAX8-NEXT: vse32.v v26, (a1) 47; LMULMAX8-NEXT: ret 48; 49; LMULMAX2-LABEL: sext_v8i8_v8i32: 50; LMULMAX2: # %bb.0: 51; LMULMAX2-NEXT: vsetivli zero, 8, e8, mf2, ta, mu 52; LMULMAX2-NEXT: vle8.v v25, (a0) 53; LMULMAX2-NEXT: vsetvli zero, zero, e32, m2, ta, mu 54; LMULMAX2-NEXT: vsext.vf4 v26, v25 55; LMULMAX2-NEXT: vse32.v v26, (a1) 56; LMULMAX2-NEXT: ret 57; 58; LMULMAX1-LABEL: sext_v8i8_v8i32: 59; LMULMAX1: # %bb.0: 60; LMULMAX1-NEXT: vsetivli zero, 8, e8, mf2, ta, mu 61; LMULMAX1-NEXT: vle8.v v25, (a0) 62; LMULMAX1-NEXT: vsetivli zero, 4, e8, mf2, ta, mu 63; LMULMAX1-NEXT: vslidedown.vi v26, v25, 4 64; LMULMAX1-NEXT: vsetivli zero, 4, e32, m1, ta, mu 65; LMULMAX1-NEXT: vsext.vf4 v27, v26 66; LMULMAX1-NEXT: vsext.vf4 v26, v25 67; LMULMAX1-NEXT: addi a0, a1, 16 68; LMULMAX1-NEXT: vse32.v v27, (a0) 69; LMULMAX1-NEXT: vse32.v v26, (a1) 70; LMULMAX1-NEXT: ret 71 %a = load <8 x i8>, <8 x i8>* %x 72 %b = sext <8 x i8> %a to <8 x i32> 73 store <8 x i32> %b, <8 x i32>* %z 74 ret void 75} 76 77define void @sext_v32i8_v32i32(<32 x i8>* %x, <32 x i32>* %z) { 78; LMULMAX8-LABEL: sext_v32i8_v32i32: 79; LMULMAX8: # %bb.0: 80; LMULMAX8-NEXT: addi a2, zero, 32 81; LMULMAX8-NEXT: vsetvli zero, a2, e8, m2, ta, mu 82; LMULMAX8-NEXT: vle8.v v26, (a0) 83; LMULMAX8-NEXT: vsetvli zero, zero, e32, m8, ta, mu 84; LMULMAX8-NEXT: vsext.vf4 v8, v26 85; LMULMAX8-NEXT: vse32.v v8, (a1) 86; LMULMAX8-NEXT: ret 87; 88; LMULMAX2-LABEL: sext_v32i8_v32i32: 89; LMULMAX2: # %bb.0: 90; LMULMAX2-NEXT: addi a2, zero, 32 91; LMULMAX2-NEXT: vsetvli zero, a2, e8, m2, ta, mu 92; LMULMAX2-NEXT: vle8.v v26, (a0) 93; LMULMAX2-NEXT: vsetivli zero, 8, e8, m1, ta, mu 94; LMULMAX2-NEXT: vslidedown.vi v25, v26, 8 95; LMULMAX2-NEXT: vsetivli zero, 8, e32, m2, ta, mu 96; LMULMAX2-NEXT: vsext.vf4 v28, v25 97; LMULMAX2-NEXT: vsetivli zero, 16, e8, m2, ta, mu 98; LMULMAX2-NEXT: vslidedown.vi v30, v26, 16 99; LMULMAX2-NEXT: vsetivli zero, 8, e8, m1, ta, mu 100; LMULMAX2-NEXT: vslidedown.vi v25, v30, 8 101; LMULMAX2-NEXT: vsetivli zero, 8, e32, m2, ta, mu 102; LMULMAX2-NEXT: vsext.vf4 v8, v25 103; LMULMAX2-NEXT: vsext.vf4 v10, v26 104; LMULMAX2-NEXT: vsext.vf4 v26, v30 105; LMULMAX2-NEXT: addi a0, a1, 64 106; LMULMAX2-NEXT: vse32.v v26, (a0) 107; LMULMAX2-NEXT: vse32.v v10, (a1) 108; LMULMAX2-NEXT: addi a0, a1, 96 109; LMULMAX2-NEXT: vse32.v v8, (a0) 110; LMULMAX2-NEXT: addi a0, a1, 32 111; LMULMAX2-NEXT: vse32.v v28, (a0) 112; LMULMAX2-NEXT: ret 113; 114; LMULMAX1-LABEL: sext_v32i8_v32i32: 115; LMULMAX1: # %bb.0: 116; LMULMAX1-NEXT: vsetivli zero, 16, e8, m1, ta, mu 117; LMULMAX1-NEXT: addi a2, a0, 16 118; LMULMAX1-NEXT: vle8.v v25, (a2) 119; LMULMAX1-NEXT: vle8.v v26, (a0) 120; LMULMAX1-NEXT: vsetivli zero, 4, e8, mf2, ta, mu 121; LMULMAX1-NEXT: vslidedown.vi v27, v25, 4 122; LMULMAX1-NEXT: vsetivli zero, 4, e32, m1, ta, mu 123; LMULMAX1-NEXT: vsext.vf4 v28, v27 124; LMULMAX1-NEXT: vsetivli zero, 8, e8, m1, ta, mu 125; LMULMAX1-NEXT: vslidedown.vi v27, v25, 8 126; LMULMAX1-NEXT: vsetivli zero, 4, e8, mf2, ta, mu 127; LMULMAX1-NEXT: vslidedown.vi v29, v27, 4 128; LMULMAX1-NEXT: vsetivli zero, 4, e32, m1, ta, mu 129; LMULMAX1-NEXT: vsext.vf4 v30, v29 130; LMULMAX1-NEXT: vsetivli zero, 4, e8, mf2, ta, mu 131; LMULMAX1-NEXT: vslidedown.vi v29, v26, 4 132; LMULMAX1-NEXT: vsetivli zero, 4, e32, m1, ta, mu 133; LMULMAX1-NEXT: vsext.vf4 v31, v29 134; LMULMAX1-NEXT: vsetivli zero, 8, e8, m1, ta, mu 135; LMULMAX1-NEXT: vslidedown.vi v29, v26, 8 136; LMULMAX1-NEXT: vsetivli zero, 4, e8, mf2, ta, mu 137; LMULMAX1-NEXT: vslidedown.vi v8, v29, 4 138; LMULMAX1-NEXT: vsetivli zero, 4, e32, m1, ta, mu 139; LMULMAX1-NEXT: vsext.vf4 v9, v8 140; LMULMAX1-NEXT: vsext.vf4 v8, v27 141; LMULMAX1-NEXT: vsext.vf4 v27, v29 142; LMULMAX1-NEXT: vsext.vf4 v29, v25 143; LMULMAX1-NEXT: vsext.vf4 v25, v26 144; LMULMAX1-NEXT: addi a0, a1, 32 145; LMULMAX1-NEXT: vse32.v v27, (a0) 146; LMULMAX1-NEXT: vse32.v v25, (a1) 147; LMULMAX1-NEXT: addi a0, a1, 96 148; LMULMAX1-NEXT: vse32.v v8, (a0) 149; LMULMAX1-NEXT: addi a0, a1, 64 150; LMULMAX1-NEXT: vse32.v v29, (a0) 151; LMULMAX1-NEXT: addi a0, a1, 48 152; LMULMAX1-NEXT: vse32.v v9, (a0) 153; LMULMAX1-NEXT: addi a0, a1, 16 154; LMULMAX1-NEXT: vse32.v v31, (a0) 155; LMULMAX1-NEXT: addi a0, a1, 112 156; LMULMAX1-NEXT: vse32.v v30, (a0) 157; LMULMAX1-NEXT: addi a0, a1, 80 158; LMULMAX1-NEXT: vse32.v v28, (a0) 159; LMULMAX1-NEXT: ret 160 %a = load <32 x i8>, <32 x i8>* %x 161 %b = sext <32 x i8> %a to <32 x i32> 162 store <32 x i32> %b, <32 x i32>* %z 163 ret void 164} 165 166define void @trunc_v4i8_v4i32(<4 x i32>* %x, <4 x i8>* %z) { 167; CHECK-LABEL: trunc_v4i8_v4i32: 168; CHECK: # %bb.0: 169; CHECK-NEXT: vsetivli zero, 4, e32, m1, ta, mu 170; CHECK-NEXT: vle32.v v25, (a0) 171; CHECK-NEXT: vsetvli zero, zero, e16, mf2, ta, mu 172; CHECK-NEXT: vnsrl.wi v25, v25, 0 173; CHECK-NEXT: vsetvli zero, zero, e8, mf4, ta, mu 174; CHECK-NEXT: vnsrl.wi v25, v25, 0 175; CHECK-NEXT: vse8.v v25, (a1) 176; CHECK-NEXT: ret 177 %a = load <4 x i32>, <4 x i32>* %x 178 %b = trunc <4 x i32> %a to <4 x i8> 179 store <4 x i8> %b, <4 x i8>* %z 180 ret void 181} 182 183define void @trunc_v8i8_v8i32(<8 x i32>* %x, <8 x i8>* %z) { 184; LMULMAX8-LABEL: trunc_v8i8_v8i32: 185; LMULMAX8: # %bb.0: 186; LMULMAX8-NEXT: vsetivli zero, 8, e32, m2, ta, mu 187; LMULMAX8-NEXT: vle32.v v26, (a0) 188; LMULMAX8-NEXT: vsetvli zero, zero, e16, m1, ta, mu 189; LMULMAX8-NEXT: vnsrl.wi v25, v26, 0 190; LMULMAX8-NEXT: vsetvli zero, zero, e8, mf2, ta, mu 191; LMULMAX8-NEXT: vnsrl.wi v25, v25, 0 192; LMULMAX8-NEXT: vse8.v v25, (a1) 193; LMULMAX8-NEXT: ret 194; 195; LMULMAX2-LABEL: trunc_v8i8_v8i32: 196; LMULMAX2: # %bb.0: 197; LMULMAX2-NEXT: vsetivli zero, 8, e32, m2, ta, mu 198; LMULMAX2-NEXT: vle32.v v26, (a0) 199; LMULMAX2-NEXT: vsetvli zero, zero, e16, m1, ta, mu 200; LMULMAX2-NEXT: vnsrl.wi v25, v26, 0 201; LMULMAX2-NEXT: vsetvli zero, zero, e8, mf2, ta, mu 202; LMULMAX2-NEXT: vnsrl.wi v25, v25, 0 203; LMULMAX2-NEXT: vse8.v v25, (a1) 204; LMULMAX2-NEXT: ret 205; 206; LMULMAX1-LABEL: trunc_v8i8_v8i32: 207; LMULMAX1: # %bb.0: 208; LMULMAX1-NEXT: vsetivli zero, 4, e32, m1, ta, mu 209; LMULMAX1-NEXT: vle32.v v25, (a0) 210; LMULMAX1-NEXT: addi a0, a0, 16 211; LMULMAX1-NEXT: vle32.v v26, (a0) 212; LMULMAX1-NEXT: vsetvli zero, zero, e16, mf2, ta, mu 213; LMULMAX1-NEXT: vnsrl.wi v25, v25, 0 214; LMULMAX1-NEXT: vsetvli zero, zero, e8, mf4, ta, mu 215; LMULMAX1-NEXT: vnsrl.wi v25, v25, 0 216; LMULMAX1-NEXT: vsetivli zero, 8, e8, mf2, ta, mu 217; LMULMAX1-NEXT: vmv.v.i v27, 0 218; LMULMAX1-NEXT: vsetivli zero, 4, e8, mf2, tu, mu 219; LMULMAX1-NEXT: vslideup.vi v27, v25, 0 220; LMULMAX1-NEXT: vsetivli zero, 4, e16, mf2, ta, mu 221; LMULMAX1-NEXT: vnsrl.wi v25, v26, 0 222; LMULMAX1-NEXT: vsetvli zero, zero, e8, mf4, ta, mu 223; LMULMAX1-NEXT: vnsrl.wi v25, v25, 0 224; LMULMAX1-NEXT: vsetivli zero, 8, e8, mf2, tu, mu 225; LMULMAX1-NEXT: vslideup.vi v27, v25, 4 226; LMULMAX1-NEXT: vsetvli zero, zero, e8, mf2, ta, mu 227; LMULMAX1-NEXT: vse8.v v27, (a1) 228; LMULMAX1-NEXT: ret 229 %a = load <8 x i32>, <8 x i32>* %x 230 %b = trunc <8 x i32> %a to <8 x i8> 231 store <8 x i8> %b, <8 x i8>* %z 232 ret void 233} 234