| 1 |
/* PR target/113609 */ |
| 2 |
/* { dg-do compile } */ |
| 3 |
/* { dg-options "-O2 -march=x86-64-v4" } */ |
| 4 |
/* { dg-final { scan-assembler-not "^cmp" } } */ |
| 5 |
/* { dg-final { scan-assembler-not "\[ \\t\]+sete" { target { ! ia32 } } } } */ |
| 6 |
/* { dg-final { scan-assembler-not "\[ \\t\]+setne" { target { ! ia32 } } } } */ |
| 7 |
/* { dg-final { scan-assembler-not "\[ \\t\]+je" { target { ! ia32 } } } } */ |
| 8 |
/* { dg-final { scan-assembler-not "\[ \\t\]+jne" { target { ! ia32 } } } } */ |
| 9 |
/* { dg-final { scan-assembler-times "\[ \\t\]+sete" 1 { target { ia32 } } } } */ |
| 10 |
/* { dg-final { scan-assembler-times "\[ \\t\]+setne" 1 { target { ia32 } } } } */ |
| 11 |
/* { dg-final { scan-assembler-times "\[ \\t\]+je" 1 { target { ia32 } } } } */ |
| 12 |
/* { dg-final { scan-assembler-times "\[ \\t\]+jne" 2 { target { ia32 } } } } */ |
| 13 |
/* { dg-final { scan-assembler-times "kortest" 12 { target { ia32 } } } } */ |
| 14 |
/* { dg-final { scan-assembler-times "kortest" 17 { target { ! ia32 } } } } */ |
| 15 |
|
| 16 |
#include <immintrin.h> |
| 17 |
|
| 18 |
unsigned int |
| 19 |
cmp_vector_sete_mask8(__m128i a, __m128i b) |
| 20 |
{ |
| 21 |
__mmask8 k = _mm_cmpeq_epi16_mask (a, b); |
| 22 |
if (k == (__mmask8) -1) |
| 23 |
return 1; |
| 24 |
else |
| 25 |
return 0; |
| 26 |
} |
| 27 |
|
| 28 |
unsigned int |
| 29 |
cmp_vector_sete_mask16(__m128i a, __m128i b) |
| 30 |
{ |
| 31 |
__mmask16 k = _mm_cmpeq_epi8_mask (a, b); |
| 32 |
if (k == (__mmask16) -1) |
| 33 |
return 1; |
| 34 |
else |
| 35 |
return 0; |
| 36 |
} |
| 37 |
|
| 38 |
unsigned int |
| 39 |
cmp_vector_sete_mask32(__m256i a, __m256i b) |
| 40 |
{ |
| 41 |
__mmask32 k = _mm256_cmpeq_epi8_mask (a, b); |
| 42 |
if (k == (__mmask32) -1) |
| 43 |
return 1; |
| 44 |
else |
| 45 |
return 0; |
| 46 |
} |
| 47 |
|
| 48 |
unsigned int |
| 49 |
cmp_vector_sete_mask64(__m512i a, __m512i b) |
| 50 |
{ |
| 51 |
__mmask64 k = _mm512_cmpeq_epi8_mask (a, b); |
| 52 |
if (k == (__mmask64) -1) |
| 53 |
return 1; |
| 54 |
else |
| 55 |
return 0; |
| 56 |
} |
| 57 |
|
| 58 |
unsigned int |
| 59 |
cmp_vector_setne_mask8(__m128i a, __m128i b) |
| 60 |
{ |
| 61 |
__mmask8 k = _mm_cmpeq_epi16_mask (a, b); |
| 62 |
if (k != (__mmask8) -1) |
| 63 |
return 1; |
| 64 |
else |
| 65 |
return 0; |
| 66 |
} |
| 67 |
|
| 68 |
unsigned int |
| 69 |
cmp_vector_setne_mask16(__m128i a, __m128i b) |
| 70 |
{ |
| 71 |
__mmask16 k = _mm_cmpeq_epi8_mask (a, b); |
| 72 |
if (k != (__mmask16) -1) |
| 73 |
return 1; |
| 74 |
else |
| 75 |
return 0; |
| 76 |
} |
| 77 |
|
| 78 |
unsigned int |
| 79 |
cmp_vector_setne_mask32(__m256i a, __m256i b) |
| 80 |
{ |
| 81 |
__mmask32 k = _mm256_cmpeq_epi8_mask (a, b); |
| 82 |
if (k != (__mmask32) -1) |
| 83 |
return 1; |
| 84 |
else |
| 85 |
return 0; |
| 86 |
} |
| 87 |
|
| 88 |
unsigned int |
| 89 |
cmp_vector_setne_mask64(__m512i a, __m512i b) |
| 90 |
{ |
| 91 |
__mmask64 k = _mm512_cmpeq_epi8_mask (a, b); |
| 92 |
if (k != (__mmask64) -1) |
| 93 |
return 1; |
| 94 |
else |
| 95 |
return 0; |
| 96 |
} |
| 97 |
|
| 98 |
__m128i |
| 99 |
cmp_vector_je_mask8(__m128i a, __m128i b) { |
| 100 |
__mmask8 k = _mm_cmpeq_epi16_mask (a, b); |
| 101 |
if (k == (__mmask8) -1) { |
| 102 |
a[0] = a[0] + 1; |
| 103 |
} |
| 104 |
else { |
| 105 |
a[0] = a[0] - 1; |
| 106 |
} |
| 107 |
return a; |
| 108 |
} |
| 109 |
|
| 110 |
__m128i |
| 111 |
cmp_vector_je_mask16(__m128i a, __m128i b) { |
| 112 |
__mmask16 k = _mm_cmpeq_epi8_mask (a, b); |
| 113 |
if (k == (__mmask16) -1) { |
| 114 |
a[0] = a[0] + 1; |
| 115 |
} |
| 116 |
else { |
| 117 |
a[0] = a[0] - 1; |
| 118 |
} |
| 119 |
return a; |
| 120 |
} |
| 121 |
|
| 122 |
__m256i |
| 123 |
cmp_vector_je_mask32(__m256i a, __m256i b) { |
| 124 |
__mmask32 k = _mm256_cmpeq_epi8_mask (a, b); |
| 125 |
if (k == (__mmask32) -1) { |
| 126 |
a[0] = a[0] + 1; |
| 127 |
} |
| 128 |
else { |
| 129 |
a[0] = a[0] - 1; |
| 130 |
} |
| 131 |
return a; |
| 132 |
} |
| 133 |
|
| 134 |
__m512i |
| 135 |
cmp_vector_je_mask64(__m512i a, __m512i b) { |
| 136 |
__mmask64 k = _mm512_cmpeq_epi8_mask (a, b); |
| 137 |
if (k == (__mmask64) -1) { |
| 138 |
a[0] = a[0] + 1; |
| 139 |
} |
| 140 |
else { |
| 141 |
a[0] = a[0] - 5; |
| 142 |
} |
| 143 |
return a; |
| 144 |
} |
| 145 |
|
| 146 |
__m128i |
| 147 |
cmp_vector_jne_mask8(__m128i a, __m128i b) { |
| 148 |
__mmask8 k = _mm_cmpeq_epi16_mask (a, b); |
| 149 |
if (k == (__mmask8) -1) { |
| 150 |
a[0] = a[0] + 1; |
| 151 |
} |
| 152 |
a[0] = a[0] - 4; |
| 153 |
return a; |
| 154 |
} |
| 155 |
|
| 156 |
__m128i |
| 157 |
cmp_vector_jne_mask16(__m128i a, __m128i b) { |
| 158 |
__mmask16 k = _mm_cmpeq_epi8_mask (a, b); |
| 159 |
if (k == (__mmask16) -1) { |
| 160 |
a[0] = a[0] + 1; |
| 161 |
} |
| 162 |
a[0] = a[0] - 4; |
| 163 |
return a; |
| 164 |
} |
| 165 |
|
| 166 |
__m256i |
| 167 |
cmp_vector_jne_mask32(__m256i a, __m256i b) { |
| 168 |
__mmask32 k = _mm256_cmpeq_epi8_mask (a, b); |
| 169 |
if (k == (__mmask32) -1) { |
| 170 |
a[0] = a[0] + 1; |
| 171 |
} |
| 172 |
a[0] = a[0] - 4; |
| 173 |
return a; |
| 174 |
} |
| 175 |
|
| 176 |
__m512i |
| 177 |
cmp_vector_jne_mask64(__m512i a, __m512i b) { |
| 178 |
__mmask64 k = _mm512_cmpeq_epi8_mask (a, b); |
| 179 |
if (k == (__mmask64) -1) { |
| 180 |
a[0] = a[0] + 1; |
| 181 |
} |
| 182 |
a[0] = a[0] - 4; |
| 183 |
return a; |
| 184 |
} |
| 185 |
|
| 186 |
__m512i |
| 187 |
mask_cmp_vector_jne_mask64(__m512i a, __m512i b) { |
| 188 |
__mmask64 k = _mm512_mask_cmpeq_epi8_mask ((__mmask64)0xffffffefffffffff, a, b); |
| 189 |
if (k == (__mmask64) -1) { |
| 190 |
a[0] = a[0] + 1; |
| 191 |
} |
| 192 |
a[0] = a[0] - 4; |
| 193 |
return a; |
| 194 |
} |