1 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py 2 ; RUN: opt < %s -mtriple=x86_64-unknown -basicaa -slp-vectorizer -S | FileCheck %s --check-prefix=CHECK --check-prefix=SSE 3 ; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=corei7-avx -basicaa -slp-vectorizer -S | FileCheck %s --check-prefix=CHECK --check-prefix=AVX 4 ; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=core-avx2 -basicaa -slp-vectorizer -S | FileCheck %s --check-prefix=CHECK --check-prefix=AVX 5 6 target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128" 7 8 @src64 = common global [4 x i64] zeroinitializer, align 32 9 @dst64 = common global [4 x i64] zeroinitializer, align 32 10 @src32 = common global [8 x i32] zeroinitializer, align 32 11 @dst32 = common global [8 x i32] zeroinitializer, align 32 12 @src16 = common global [16 x i16] zeroinitializer, align 32 13 @dst16 = common global [16 x i16] zeroinitializer, align 32 14 @src8 = common global [32 x i8] zeroinitializer, align 32 15 @dst8 = common global [32 x i8] zeroinitializer, align 32 16 17 declare i64 @llvm.ctpop.i64(i64) 18 declare i32 @llvm.ctpop.i32(i32) 19 declare i16 @llvm.ctpop.i16(i16) 20 declare i8 @llvm.ctpop.i8(i8) 21 22 define void @ctpop_2i64() #0 { 23 ; CHECK-LABEL: @ctpop_2i64( 24 ; CHECK-NEXT: [[TMP1:%.*]] = load <2 x i64>, <2 x i64>* bitcast ([4 x i64]* @src64 to <2 x i64>*), align 8 25 ; CHECK-NEXT: [[TMP2:%.*]] = call <2 x i64> @llvm.ctpop.v2i64(<2 x i64> [[TMP1]]) 26 ; CHECK-NEXT: store <2 x i64> [[TMP2]], <2 x i64>* bitcast ([4 x i64]* @dst64 to <2 x i64>*), align 8 27 ; CHECK-NEXT: ret void 28 ; 29 %ld0 = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i32 0, i64 0), align 8 30 %ld1 = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i32 0, i64 1), align 8 31 %ctpop0 = call i64 @llvm.ctpop.i64(i64 %ld0) 32 %ctpop1 = call i64 @llvm.ctpop.i64(i64 %ld1) 33 store i64 %ctpop0, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i32 0, i64 0), align 8 34 store i64 %ctpop1, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i32 0, i64 1), align 8 35 ret void 36 } 37 38 define void @ctpop_4i64() #0 { 39 ; SSE-LABEL: @ctpop_4i64( 40 ; SSE-NEXT: [[TMP1:%.*]] = load <2 x i64>, <2 x i64>* bitcast ([4 x i64]* @src64 to <2 x i64>*), align 4 41 ; SSE-NEXT: [[TMP2:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i64 0, i64 2) to <2 x i64>*), align 4 42 ; SSE-NEXT: [[TMP3:%.*]] = call <2 x i64> @llvm.ctpop.v2i64(<2 x i64> [[TMP1]]) 43 ; SSE-NEXT: [[TMP4:%.*]] = call <2 x i64> @llvm.ctpop.v2i64(<2 x i64> [[TMP2]]) 44 ; SSE-NEXT: store <2 x i64> [[TMP3]], <2 x i64>* bitcast ([4 x i64]* @dst64 to <2 x i64>*), align 4 45 ; SSE-NEXT: store <2 x i64> [[TMP4]], <2 x i64>* bitcast (i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 2) to <2 x i64>*), align 4 46 ; SSE-NEXT: ret void 47 ; 48 ; AVX-LABEL: @ctpop_4i64( 49 ; AVX-NEXT: [[TMP1:%.*]] = load <4 x i64>, <4 x i64>* bitcast ([4 x i64]* @src64 to <4 x i64>*), align 4 50 ; AVX-NEXT: [[TMP2:%.*]] = call <4 x i64> @llvm.ctpop.v4i64(<4 x i64> [[TMP1]]) 51 ; AVX-NEXT: store <4 x i64> [[TMP2]], <4 x i64>* bitcast ([4 x i64]* @dst64 to <4 x i64>*), align 4 52 ; AVX-NEXT: ret void 53 ; 54 %ld0 = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i64 0, i64 0), align 4 55 %ld1 = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i64 0, i64 1), align 4 56 %ld2 = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i64 0, i64 2), align 4 57 %ld3 = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i64 0, i64 3), align 4 58 %ctpop0 = call i64 @llvm.ctpop.i64(i64 %ld0) 59 %ctpop1 = call i64 @llvm.ctpop.i64(i64 %ld1) 60 %ctpop2 = call i64 @llvm.ctpop.i64(i64 %ld2) 61 %ctpop3 = call i64 @llvm.ctpop.i64(i64 %ld3) 62 store i64 %ctpop0, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 0), align 4 63 store i64 %ctpop1, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 1), align 4 64 store i64 %ctpop2, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 2), align 4 65 store i64 %ctpop3, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 3), align 4 66 ret void 67 } 68 69 define void @ctpop_4i32() #0 { 70 ; CHECK-LABEL: @ctpop_4i32( 71 ; CHECK-NEXT: [[TMP1:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([8 x i32]* @src32 to <4 x i32>*), align 4 72 ; CHECK-NEXT: [[TMP2:%.*]] = call <4 x i32> @llvm.ctpop.v4i32(<4 x i32> [[TMP1]]) 73 ; CHECK-NEXT: store <4 x i32> [[TMP2]], <4 x i32>* bitcast ([8 x i32]* @dst32 to <4 x i32>*), align 4 74 ; CHECK-NEXT: ret void 75 ; 76 %ld0 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 0), align 4 77 %ld1 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 1), align 4 78 %ld2 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 2), align 4 79 %ld3 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 3), align 4 80 %ctpop0 = call i32 @llvm.ctpop.i32(i32 %ld0) 81 %ctpop1 = call i32 @llvm.ctpop.i32(i32 %ld1) 82 %ctpop2 = call i32 @llvm.ctpop.i32(i32 %ld2) 83 %ctpop3 = call i32 @llvm.ctpop.i32(i32 %ld3) 84 store i32 %ctpop0, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 4 85 store i32 %ctpop1, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 4 86 store i32 %ctpop2, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 4 87 store i32 %ctpop3, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 3), align 4 88 ret void 89 } 90 91 define void @ctpop_8i32() #0 { 92 ; SSE-LABEL: @ctpop_8i32( 93 ; SSE-NEXT: [[TMP1:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([8 x i32]* @src32 to <4 x i32>*), align 2 94 ; SSE-NEXT: [[TMP2:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 4) to <4 x i32>*), align 2 95 ; SSE-NEXT: [[TMP3:%.*]] = call <4 x i32> @llvm.ctpop.v4i32(<4 x i32> [[TMP1]]) 96 ; SSE-NEXT: [[TMP4:%.*]] = call <4 x i32> @llvm.ctpop.v4i32(<4 x i32> [[TMP2]]) 97 ; SSE-NEXT: store <4 x i32> [[TMP3]], <4 x i32>* bitcast ([8 x i32]* @dst32 to <4 x i32>*), align 2 98 ; SSE-NEXT: store <4 x i32> [[TMP4]], <4 x i32>* bitcast (i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 4) to <4 x i32>*), align 2 99 ; SSE-NEXT: ret void 100 ; 101 ; AVX-LABEL: @ctpop_8i32( 102 ; AVX-NEXT: [[TMP1:%.*]] = load <8 x i32>, <8 x i32>* bitcast ([8 x i32]* @src32 to <8 x i32>*), align 2 103 ; AVX-NEXT: [[TMP2:%.*]] = call <8 x i32> @llvm.ctpop.v8i32(<8 x i32> [[TMP1]]) 104 ; AVX-NEXT: store <8 x i32> [[TMP2]], <8 x i32>* bitcast ([8 x i32]* @dst32 to <8 x i32>*), align 2 105 ; AVX-NEXT: ret void 106 ; 107 %ld0 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 0), align 2 108 %ld1 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 1), align 2 109 %ld2 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 2), align 2 110 %ld3 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 3), align 2 111 %ld4 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 4), align 2 112 %ld5 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 5), align 2 113 %ld6 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 6), align 2 114 %ld7 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 7), align 2 115 %ctpop0 = call i32 @llvm.ctpop.i32(i32 %ld0) 116 %ctpop1 = call i32 @llvm.ctpop.i32(i32 %ld1) 117 %ctpop2 = call i32 @llvm.ctpop.i32(i32 %ld2) 118 %ctpop3 = call i32 @llvm.ctpop.i32(i32 %ld3) 119 %ctpop4 = call i32 @llvm.ctpop.i32(i32 %ld4) 120 %ctpop5 = call i32 @llvm.ctpop.i32(i32 %ld5) 121 %ctpop6 = call i32 @llvm.ctpop.i32(i32 %ld6) 122 %ctpop7 = call i32 @llvm.ctpop.i32(i32 %ld7) 123 store i32 %ctpop0, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 2 124 store i32 %ctpop1, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 2 125 store i32 %ctpop2, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 2 126 store i32 %ctpop3, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 3), align 2 127 store i32 %ctpop4, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 4), align 2 128 store i32 %ctpop5, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 5), align 2 129 store i32 %ctpop6, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 6), align 2 130 store i32 %ctpop7, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 7), align 2 131 ret void 132 } 133 134 define void @ctpop_8i16() #0 { 135 ; CHECK-LABEL: @ctpop_8i16( 136 ; CHECK-NEXT: [[TMP1:%.*]] = load <8 x i16>, <8 x i16>* bitcast ([16 x i16]* @src16 to <8 x i16>*), align 2 137 ; CHECK-NEXT: [[TMP2:%.*]] = call <8 x i16> @llvm.ctpop.v8i16(<8 x i16> [[TMP1]]) 138 ; CHECK-NEXT: store <8 x i16> [[TMP2]], <8 x i16>* bitcast ([16 x i16]* @dst16 to <8 x i16>*), align 2 139 ; CHECK-NEXT: ret void 140 ; 141 %ld0 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 0), align 2 142 %ld1 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 1), align 2 143 %ld2 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 2), align 2 144 %ld3 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 3), align 2 145 %ld4 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 4), align 2 146 %ld5 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 5), align 2 147 %ld6 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 6), align 2 148 %ld7 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 7), align 2 149 %ctpop0 = call i16 @llvm.ctpop.i16(i16 %ld0) 150 %ctpop1 = call i16 @llvm.ctpop.i16(i16 %ld1) 151 %ctpop2 = call i16 @llvm.ctpop.i16(i16 %ld2) 152 %ctpop3 = call i16 @llvm.ctpop.i16(i16 %ld3) 153 %ctpop4 = call i16 @llvm.ctpop.i16(i16 %ld4) 154 %ctpop5 = call i16 @llvm.ctpop.i16(i16 %ld5) 155 %ctpop6 = call i16 @llvm.ctpop.i16(i16 %ld6) 156 %ctpop7 = call i16 @llvm.ctpop.i16(i16 %ld7) 157 store i16 %ctpop0, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2 158 store i16 %ctpop1, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2 159 store i16 %ctpop2, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2 160 store i16 %ctpop3, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2 161 store i16 %ctpop4, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2 162 store i16 %ctpop5, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2 163 store i16 %ctpop6, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2 164 store i16 %ctpop7, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 7), align 2 165 ret void 166 } 167 168 define void @ctpop_16i16() #0 { 169 ; SSE-LABEL: @ctpop_16i16( 170 ; SSE-NEXT: [[TMP1:%.*]] = load <8 x i16>, <8 x i16>* bitcast ([16 x i16]* @src16 to <8 x i16>*), align 2 171 ; SSE-NEXT: [[TMP2:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 8) to <8 x i16>*), align 2 172 ; SSE-NEXT: [[TMP3:%.*]] = call <8 x i16> @llvm.ctpop.v8i16(<8 x i16> [[TMP1]]) 173 ; SSE-NEXT: [[TMP4:%.*]] = call <8 x i16> @llvm.ctpop.v8i16(<8 x i16> [[TMP2]]) 174 ; SSE-NEXT: store <8 x i16> [[TMP3]], <8 x i16>* bitcast ([16 x i16]* @dst16 to <8 x i16>*), align 2 175 ; SSE-NEXT: store <8 x i16> [[TMP4]], <8 x i16>* bitcast (i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 8) to <8 x i16>*), align 2 176 ; SSE-NEXT: ret void 177 ; 178 ; AVX-LABEL: @ctpop_16i16( 179 ; AVX-NEXT: [[TMP1:%.*]] = load <16 x i16>, <16 x i16>* bitcast ([16 x i16]* @src16 to <16 x i16>*), align 2 180 ; AVX-NEXT: [[TMP2:%.*]] = call <16 x i16> @llvm.ctpop.v16i16(<16 x i16> [[TMP1]]) 181 ; AVX-NEXT: store <16 x i16> [[TMP2]], <16 x i16>* bitcast ([16 x i16]* @dst16 to <16 x i16>*), align 2 182 ; AVX-NEXT: ret void 183 ; 184 %ld0 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 0), align 2 185 %ld1 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 1), align 2 186 %ld2 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 2), align 2 187 %ld3 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 3), align 2 188 %ld4 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 4), align 2 189 %ld5 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 5), align 2 190 %ld6 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 6), align 2 191 %ld7 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 7), align 2 192 %ld8 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 8), align 2 193 %ld9 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 9), align 2 194 %ld10 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 10), align 2 195 %ld11 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 11), align 2 196 %ld12 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 12), align 2 197 %ld13 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 13), align 2 198 %ld14 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 14), align 2 199 %ld15 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 15), align 2 200 %ctpop0 = call i16 @llvm.ctpop.i16(i16 %ld0) 201 %ctpop1 = call i16 @llvm.ctpop.i16(i16 %ld1) 202 %ctpop2 = call i16 @llvm.ctpop.i16(i16 %ld2) 203 %ctpop3 = call i16 @llvm.ctpop.i16(i16 %ld3) 204 %ctpop4 = call i16 @llvm.ctpop.i16(i16 %ld4) 205 %ctpop5 = call i16 @llvm.ctpop.i16(i16 %ld5) 206 %ctpop6 = call i16 @llvm.ctpop.i16(i16 %ld6) 207 %ctpop7 = call i16 @llvm.ctpop.i16(i16 %ld7) 208 %ctpop8 = call i16 @llvm.ctpop.i16(i16 %ld8) 209 %ctpop9 = call i16 @llvm.ctpop.i16(i16 %ld9) 210 %ctpop10 = call i16 @llvm.ctpop.i16(i16 %ld10) 211 %ctpop11 = call i16 @llvm.ctpop.i16(i16 %ld11) 212 %ctpop12 = call i16 @llvm.ctpop.i16(i16 %ld12) 213 %ctpop13 = call i16 @llvm.ctpop.i16(i16 %ld13) 214 %ctpop14 = call i16 @llvm.ctpop.i16(i16 %ld14) 215 %ctpop15 = call i16 @llvm.ctpop.i16(i16 %ld15) 216 store i16 %ctpop0 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2 217 store i16 %ctpop1 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2 218 store i16 %ctpop2 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2 219 store i16 %ctpop3 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2 220 store i16 %ctpop4 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2 221 store i16 %ctpop5 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2 222 store i16 %ctpop6 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2 223 store i16 %ctpop7 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 7), align 2 224 store i16 %ctpop8 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 8), align 2 225 store i16 %ctpop9 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 9), align 2 226 store i16 %ctpop10, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 10), align 2 227 store i16 %ctpop11, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 11), align 2 228 store i16 %ctpop12, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 12), align 2 229 store i16 %ctpop13, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 13), align 2 230 store i16 %ctpop14, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 14), align 2 231 store i16 %ctpop15, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 15), align 2 232 ret void 233 } 234 235 define void @ctpop_16i8() #0 { 236 ; CHECK-LABEL: @ctpop_16i8( 237 ; CHECK-NEXT: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* bitcast ([32 x i8]* @src8 to <16 x i8>*), align 1 238 ; CHECK-NEXT: [[TMP2:%.*]] = call <16 x i8> @llvm.ctpop.v16i8(<16 x i8> [[TMP1]]) 239 ; CHECK-NEXT: store <16 x i8> [[TMP2]], <16 x i8>* bitcast ([32 x i8]* @dst8 to <16 x i8>*), align 1 240 ; CHECK-NEXT: ret void 241 ; 242 %ld0 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 0), align 1 243 %ld1 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 1), align 1 244 %ld2 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 2), align 1 245 %ld3 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 3), align 1 246 %ld4 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 4), align 1 247 %ld5 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 5), align 1 248 %ld6 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 6), align 1 249 %ld7 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 7), align 1 250 %ld8 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 8), align 1 251 %ld9 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 9), align 1 252 %ld10 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 10), align 1 253 %ld11 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 11), align 1 254 %ld12 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 12), align 1 255 %ld13 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 13), align 1 256 %ld14 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 14), align 1 257 %ld15 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 15), align 1 258 %ctpop0 = call i8 @llvm.ctpop.i8(i8 %ld0) 259 %ctpop1 = call i8 @llvm.ctpop.i8(i8 %ld1) 260 %ctpop2 = call i8 @llvm.ctpop.i8(i8 %ld2) 261 %ctpop3 = call i8 @llvm.ctpop.i8(i8 %ld3) 262 %ctpop4 = call i8 @llvm.ctpop.i8(i8 %ld4) 263 %ctpop5 = call i8 @llvm.ctpop.i8(i8 %ld5) 264 %ctpop6 = call i8 @llvm.ctpop.i8(i8 %ld6) 265 %ctpop7 = call i8 @llvm.ctpop.i8(i8 %ld7) 266 %ctpop8 = call i8 @llvm.ctpop.i8(i8 %ld8) 267 %ctpop9 = call i8 @llvm.ctpop.i8(i8 %ld9) 268 %ctpop10 = call i8 @llvm.ctpop.i8(i8 %ld10) 269 %ctpop11 = call i8 @llvm.ctpop.i8(i8 %ld11) 270 %ctpop12 = call i8 @llvm.ctpop.i8(i8 %ld12) 271 %ctpop13 = call i8 @llvm.ctpop.i8(i8 %ld13) 272 %ctpop14 = call i8 @llvm.ctpop.i8(i8 %ld14) 273 %ctpop15 = call i8 @llvm.ctpop.i8(i8 %ld15) 274 store i8 %ctpop0 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 0), align 1 275 store i8 %ctpop1 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 1), align 1 276 store i8 %ctpop2 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 2), align 1 277 store i8 %ctpop3 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 3), align 1 278 store i8 %ctpop4 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 4), align 1 279 store i8 %ctpop5 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 5), align 1 280 store i8 %ctpop6 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 6), align 1 281 store i8 %ctpop7 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 7), align 1 282 store i8 %ctpop8 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 8), align 1 283 store i8 %ctpop9 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 9), align 1 284 store i8 %ctpop10, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 10), align 1 285 store i8 %ctpop11, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 11), align 1 286 store i8 %ctpop12, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 12), align 1 287 store i8 %ctpop13, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 13), align 1 288 store i8 %ctpop14, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 14), align 1 289 store i8 %ctpop15, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 15), align 1 290 ret void 291 } 292 293 define void @ctpop_32i8() #0 { 294 ; CHECK-LABEL: @ctpop_32i8( 295 ; CHECK-NEXT: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* bitcast ([32 x i8]* @src8 to <16 x i8>*), align 1 296 ; CHECK-NEXT: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 16) to <16 x i8>*), align 1 297 ; CHECK-NEXT: [[TMP3:%.*]] = call <16 x i8> @llvm.ctpop.v16i8(<16 x i8> [[TMP1]]) 298 ; CHECK-NEXT: [[TMP4:%.*]] = call <16 x i8> @llvm.ctpop.v16i8(<16 x i8> [[TMP2]]) 299 ; CHECK-NEXT: store <16 x i8> [[TMP3]], <16 x i8>* bitcast ([32 x i8]* @dst8 to <16 x i8>*), align 1 300 ; CHECK-NEXT: store <16 x i8> [[TMP4]], <16 x i8>* bitcast (i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 16) to <16 x i8>*), align 1 301 ; CHECK-NEXT: ret void 302 ; 303 %ld0 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 0), align 1 304 %ld1 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 1), align 1 305 %ld2 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 2), align 1 306 %ld3 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 3), align 1 307 %ld4 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 4), align 1 308 %ld5 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 5), align 1 309 %ld6 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 6), align 1 310 %ld7 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 7), align 1 311 %ld8 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 8), align 1 312 %ld9 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 9), align 1 313 %ld10 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 10), align 1 314 %ld11 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 11), align 1 315 %ld12 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 12), align 1 316 %ld13 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 13), align 1 317 %ld14 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 14), align 1 318 %ld15 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 15), align 1 319 %ld16 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 16), align 1 320 %ld17 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 17), align 1 321 %ld18 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 18), align 1 322 %ld19 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 19), align 1 323 %ld20 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 20), align 1 324 %ld21 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 21), align 1 325 %ld22 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 22), align 1 326 %ld23 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 23), align 1 327 %ld24 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 24), align 1 328 %ld25 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 25), align 1 329 %ld26 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 26), align 1 330 %ld27 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 27), align 1 331 %ld28 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 28), align 1 332 %ld29 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 29), align 1 333 %ld30 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 30), align 1 334 %ld31 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 31), align 1 335 %ctpop0 = call i8 @llvm.ctpop.i8(i8 %ld0) 336 %ctpop1 = call i8 @llvm.ctpop.i8(i8 %ld1) 337 %ctpop2 = call i8 @llvm.ctpop.i8(i8 %ld2) 338 %ctpop3 = call i8 @llvm.ctpop.i8(i8 %ld3) 339 %ctpop4 = call i8 @llvm.ctpop.i8(i8 %ld4) 340 %ctpop5 = call i8 @llvm.ctpop.i8(i8 %ld5) 341 %ctpop6 = call i8 @llvm.ctpop.i8(i8 %ld6) 342 %ctpop7 = call i8 @llvm.ctpop.i8(i8 %ld7) 343 %ctpop8 = call i8 @llvm.ctpop.i8(i8 %ld8) 344 %ctpop9 = call i8 @llvm.ctpop.i8(i8 %ld9) 345 %ctpop10 = call i8 @llvm.ctpop.i8(i8 %ld10) 346 %ctpop11 = call i8 @llvm.ctpop.i8(i8 %ld11) 347 %ctpop12 = call i8 @llvm.ctpop.i8(i8 %ld12) 348 %ctpop13 = call i8 @llvm.ctpop.i8(i8 %ld13) 349 %ctpop14 = call i8 @llvm.ctpop.i8(i8 %ld14) 350 %ctpop15 = call i8 @llvm.ctpop.i8(i8 %ld15) 351 %ctpop16 = call i8 @llvm.ctpop.i8(i8 %ld16) 352 %ctpop17 = call i8 @llvm.ctpop.i8(i8 %ld17) 353 %ctpop18 = call i8 @llvm.ctpop.i8(i8 %ld18) 354 %ctpop19 = call i8 @llvm.ctpop.i8(i8 %ld19) 355 %ctpop20 = call i8 @llvm.ctpop.i8(i8 %ld20) 356 %ctpop21 = call i8 @llvm.ctpop.i8(i8 %ld21) 357 %ctpop22 = call i8 @llvm.ctpop.i8(i8 %ld22) 358 %ctpop23 = call i8 @llvm.ctpop.i8(i8 %ld23) 359 %ctpop24 = call i8 @llvm.ctpop.i8(i8 %ld24) 360 %ctpop25 = call i8 @llvm.ctpop.i8(i8 %ld25) 361 %ctpop26 = call i8 @llvm.ctpop.i8(i8 %ld26) 362 %ctpop27 = call i8 @llvm.ctpop.i8(i8 %ld27) 363 %ctpop28 = call i8 @llvm.ctpop.i8(i8 %ld28) 364 %ctpop29 = call i8 @llvm.ctpop.i8(i8 %ld29) 365 %ctpop30 = call i8 @llvm.ctpop.i8(i8 %ld30) 366 %ctpop31 = call i8 @llvm.ctpop.i8(i8 %ld31) 367 store i8 %ctpop0 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 0), align 1 368 store i8 %ctpop1 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 1), align 1 369 store i8 %ctpop2 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 2), align 1 370 store i8 %ctpop3 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 3), align 1 371 store i8 %ctpop4 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 4), align 1 372 store i8 %ctpop5 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 5), align 1 373 store i8 %ctpop6 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 6), align 1 374 store i8 %ctpop7 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 7), align 1 375 store i8 %ctpop8 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 8), align 1 376 store i8 %ctpop9 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 9), align 1 377 store i8 %ctpop10, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 10), align 1 378 store i8 %ctpop11, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 11), align 1 379 store i8 %ctpop12, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 12), align 1 380 store i8 %ctpop13, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 13), align 1 381 store i8 %ctpop14, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 14), align 1 382 store i8 %ctpop15, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 15), align 1 383 store i8 %ctpop16, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 16), align 1 384 store i8 %ctpop17, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 17), align 1 385 store i8 %ctpop18, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 18), align 1 386 store i8 %ctpop19, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 19), align 1 387 store i8 %ctpop20, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 20), align 1 388 store i8 %ctpop21, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 21), align 1 389 store i8 %ctpop22, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 22), align 1 390 store i8 %ctpop23, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 23), align 1 391 store i8 %ctpop24, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 24), align 1 392 store i8 %ctpop25, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 25), align 1 393 store i8 %ctpop26, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 26), align 1 394 store i8 %ctpop27, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 27), align 1 395 store i8 %ctpop28, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 28), align 1 396 store i8 %ctpop29, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 29), align 1 397 store i8 %ctpop30, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 30), align 1 398 store i8 %ctpop31, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 31), align 1 399 ret void 400 } 401 402 attributes #0 = { nounwind } 403 404
