| 1 |
/* { dg-skip-if "" { arm*-*-* } } */ |
| 2 |
|
| 3 |
#include <arm_neon.h> |
| 4 |
#include "arm-neon-ref.h" |
| 5 |
#include "compute-ref-data.h" |
| 6 |
|
| 7 |
/* Expected results. */ |
| 8 |
VECT_VAR_DECL(expected,int,8,8) [] = { 0xf0, 0xf2, 0xf4, 0xf6, |
| 9 |
0x11, 0x11, 0x11, 0x11 }; |
| 10 |
VECT_VAR_DECL(expected,int,16,4) [] = { 0xfff0, 0xfff2, 0x22, 0x22 }; |
| 11 |
VECT_VAR_DECL(expected,int,32,2) [] = { 0xfffffff0, 0x33 }; |
| 12 |
VECT_VAR_DECL(expected,int,64,1) [] = { 0xfffffffffffffff0 }; |
| 13 |
VECT_VAR_DECL(expected,uint,8,8) [] = { 0xf0, 0xf2, 0xf4, 0xf6, |
| 14 |
0x55, 0x55, 0x55, 0x55 }; |
| 15 |
VECT_VAR_DECL(expected,uint,16,4) [] = { 0xfff0, 0xfff2, 0x66, 0x66 }; |
| 16 |
VECT_VAR_DECL(expected,uint,32,2) [] = { 0xfffffff0, 0x77 }; |
| 17 |
VECT_VAR_DECL(expected,uint,64,1) [] = { 0xfffffffffffffff0 }; |
| 18 |
VECT_VAR_DECL(expected,poly,8,8) [] = { 0xf0, 0xf2, 0xf4, 0xf6, |
| 19 |
0x55, 0x55, 0x55, 0x55 }; |
| 20 |
VECT_VAR_DECL(expected,poly,16,4) [] = { 0xfff0, 0xfff2, 0x66, 0x66 }; |
| 21 |
#if MFLOAT8_SUPPORTED |
| 22 |
VECT_VAR_DECL(expected,hmfloat,8,8) [] = { 0xf0, 0xf2, 0xf4, 0xf6, |
| 23 |
0x7b, 0x7b, 0x7b, 0x7b }; |
| 24 |
#endif |
| 25 |
VECT_VAR_DECL(expected,hfloat,32,2) [] = { 0xc1800000, 0x42066666 }; |
| 26 |
#if defined (FP16_SUPPORTED) |
| 27 |
VECT_VAR_DECL (expected, hfloat, 16, 4) [] = { 0xcc00, 0xcb00, |
| 28 |
0x4b4d, 0x4b4d }; |
| 29 |
#endif |
| 30 |
VECT_VAR_DECL(expected,int,8,16) [] = { 0xf0, 0xf2, 0xf4, 0xf6, |
| 31 |
0xf8, 0xfa, 0xfc, 0xfe, |
| 32 |
0x11, 0x11, 0x11, 0x11, |
| 33 |
0x11, 0x11, 0x11, 0x11 }; |
| 34 |
VECT_VAR_DECL(expected,int,16,8) [] = { 0xfff0, 0xfff2, 0xfff4, 0xfff6, |
| 35 |
0x22, 0x22, 0x22, 0x22 }; |
| 36 |
VECT_VAR_DECL(expected,int,32,4) [] = { 0xfffffff0, 0xfffffff2, |
| 37 |
0x33, 0x33 }; |
| 38 |
VECT_VAR_DECL(expected,int,64,2) [] = { 0xfffffffffffffff0, |
| 39 |
0x44 }; |
| 40 |
VECT_VAR_DECL(expected,uint,8,16) [] = { 0xf0, 0xf2, 0xf4, 0xf6, |
| 41 |
0xf8, 0xfa, 0xfc, 0xfe, |
| 42 |
0x55, 0x55, 0x55, 0x55, |
| 43 |
0x55, 0x55, 0x55, 0x55 }; |
| 44 |
VECT_VAR_DECL(expected,uint,16,8) [] = { 0xfff0, 0xfff2, 0xfff4, 0xfff6, |
| 45 |
0x66, 0x66, 0x66, 0x66 }; |
| 46 |
VECT_VAR_DECL(expected,uint,32,4) [] = { 0xfffffff0, 0xfffffff2, 0x77, 0x77 }; |
| 47 |
VECT_VAR_DECL(expected,uint,64,2) [] = { 0xfffffffffffffff0, |
| 48 |
0x88 }; |
| 49 |
VECT_VAR_DECL(expected,poly,8,16) [] = { 0xf0, 0xf2, 0xf4, 0xf6, |
| 50 |
0xf8, 0xfa, 0xfc, 0xfe, |
| 51 |
0x55, 0x55, 0x55, 0x55, |
| 52 |
0x55, 0x55, 0x55, 0x55 }; |
| 53 |
VECT_VAR_DECL(expected,poly,16,8) [] = { 0xfff0, 0xfff2, 0xfff4, 0xfff6, |
| 54 |
0x66, 0x66, 0x66, 0x66 }; |
| 55 |
#if MFLOAT8_SUPPORTED |
| 56 |
VECT_VAR_DECL(expected,hmfloat,8,16) [] = { 0xf0, 0xf2, 0xf4, 0xf6, |
| 57 |
0xf8, 0xfa, 0xfc, 0xfe, |
| 58 |
0x92, 0x92, 0x92, 0x92, |
| 59 |
0x92, 0x92, 0x92, 0x92 }; |
| 60 |
#endif |
| 61 |
#if defined (FP16_SUPPORTED) |
| 62 |
VECT_VAR_DECL (expected, hfloat, 16, 8) [] = { 0xcc00, 0xcb00, 0xca00, 0xc900, |
| 63 |
0x4b4d, 0x4b4d, 0x4b4d, 0x4b4d }; |
| 64 |
#endif |
| 65 |
VECT_VAR_DECL(expected,hfloat,32,4) [] = { 0xc1800000, 0xc1600000, |
| 66 |
0x42073333, 0x42073333 }; |
| 67 |
|
| 68 |
#define TEST_MSG "VUZP1" |
| 69 |
void exec_vuzp_half (void) |
| 70 |
{ |
| 71 |
#define TEST_VUZP(PART, Q, T1, T2, W, N) \ |
| 72 |
VECT_VAR(vector_res, T1, W, N) = \ |
| 73 |
vuzp##PART##Q##_##T2##W(VECT_VAR(vector, T1, W, N), \ |
| 74 |
VECT_VAR(vector2, T1, W, N)); \ |
| 75 |
vst1##Q##_##T2##W(VECT_VAR(result, T1, W, N), VECT_VAR(vector_res, T1, W, N)) |
| 76 |
|
| 77 |
#define TEST_VUZP1(Q, T1, T2, W, N) TEST_VUZP(1, Q, T1, T2, W, N) |
| 78 |
|
| 79 |
/* Input vector can only have 64 bits. */ |
| 80 |
DECL_VARIABLE_ALL_VARIANTS(vector); |
| 81 |
DECL_VARIABLE_ALL_VARIANTS(vector2); |
| 82 |
|
| 83 |
DECL_VARIABLE_ALL_VARIANTS(vector_res); |
| 84 |
|
| 85 |
clean_results (); |
| 86 |
/* We don't have vuzp1_T64x1, so set expected to the clean value. */ |
| 87 |
CLEAN(expected, int, 64, 1); |
| 88 |
CLEAN(expected, uint, 64, 1); |
| 89 |
|
| 90 |
TEST_MACRO_ALL_VARIANTS_2_5(VLOAD, vector, buffer); |
| 91 |
#if MFLOAT8_SUPPORTED |
| 92 |
VLOAD(vector, buffer, , mfloat, mf, 8, 8); |
| 93 |
VLOAD(vector, buffer, q, mfloat, mf, 8, 16); |
| 94 |
#endif |
| 95 |
#if defined (FP16_SUPPORTED) |
| 96 |
VLOAD(vector, buffer, , float, f, 16, 4); |
| 97 |
VLOAD(vector, buffer, q, float, f, 16, 8); |
| 98 |
#endif |
| 99 |
VLOAD(vector, buffer, , float, f, 32, 2); |
| 100 |
VLOAD(vector, buffer, q, float, f, 32, 4); |
| 101 |
VLOAD(vector, buffer, q, float, f, 64, 2); |
| 102 |
|
| 103 |
/* Choose arbitrary initialization values. */ |
| 104 |
VDUP(vector2, , int, s, 8, 8, 0x11); |
| 105 |
VDUP(vector2, , int, s, 16, 4, 0x22); |
| 106 |
VDUP(vector2, , int, s, 32, 2, 0x33); |
| 107 |
VDUP(vector2, , uint, u, 8, 8, 0x55); |
| 108 |
VDUP(vector2, , uint, u, 16, 4, 0x66); |
| 109 |
VDUP(vector2, , uint, u, 32, 2, 0x77); |
| 110 |
VDUP(vector2, , poly, p, 8, 8, 0x55); |
| 111 |
VDUP(vector2, , poly, p, 16, 4, 0x66); |
| 112 |
MFLOAT8_ONLY(VDUP(vector2, , mfloat, mf, 8, 8, MFLOAT8(0x7b))); |
| 113 |
#if defined (FP16_SUPPORTED) |
| 114 |
VDUP (vector2, , float, f, 16, 4, 14.6f); /* 14.6f is 0x4b4d. */ |
| 115 |
#endif |
| 116 |
VDUP(vector2, , float, f, 32, 2, 33.6f); |
| 117 |
|
| 118 |
VDUP(vector2, q, int, s, 8, 16, 0x11); |
| 119 |
VDUP(vector2, q, int, s, 16, 8, 0x22); |
| 120 |
VDUP(vector2, q, int, s, 32, 4, 0x33); |
| 121 |
VDUP(vector2, q, int, s, 64, 2, 0x44); |
| 122 |
VDUP(vector2, q, uint, u, 8, 16, 0x55); |
| 123 |
VDUP(vector2, q, uint, u, 16, 8, 0x66); |
| 124 |
VDUP(vector2, q, uint, u, 32, 4, 0x77); |
| 125 |
VDUP(vector2, q, uint, u, 64, 2, 0x88); |
| 126 |
VDUP(vector2, q, poly, p, 8, 16, 0x55); |
| 127 |
VDUP(vector2, q, poly, p, 16, 8, 0x66); |
| 128 |
MFLOAT8_ONLY(VDUP(vector2, q, mfloat, mf, 8, 16, MFLOAT8(0x92))); |
| 129 |
#if defined (FP16_SUPPORTED) |
| 130 |
VDUP (vector2, q, float, f, 16, 8, 14.6f); |
| 131 |
#endif |
| 132 |
VDUP(vector2, q, float, f, 32, 4, 33.8f); |
| 133 |
VDUP(vector2, q, float, f, 64, 2, 33.8f); |
| 134 |
|
| 135 |
TEST_VUZP1(, int, s, 8, 8); |
| 136 |
TEST_VUZP1(, int, s, 16, 4); |
| 137 |
TEST_VUZP1(, int, s, 32, 2); |
| 138 |
TEST_VUZP1(, uint, u, 8, 8); |
| 139 |
TEST_VUZP1(, uint, u, 16, 4); |
| 140 |
TEST_VUZP1(, uint, u, 32, 2); |
| 141 |
TEST_VUZP1(, poly, p, 8, 8); |
| 142 |
TEST_VUZP1(, poly, p, 16, 4); |
| 143 |
MFLOAT8_ONLY(TEST_VUZP1(, mfloat, mf, 8, 8)); |
| 144 |
#if defined (FP16_SUPPORTED) |
| 145 |
TEST_VUZP1(, float, f, 16, 4); |
| 146 |
#endif |
| 147 |
TEST_VUZP1(, float, f, 32, 2); |
| 148 |
|
| 149 |
TEST_VUZP1(q, int, s, 8, 16); |
| 150 |
TEST_VUZP1(q, int, s, 16, 8); |
| 151 |
TEST_VUZP1(q, int, s, 32, 4); |
| 152 |
TEST_VUZP1(q, int, s, 64, 2); |
| 153 |
TEST_VUZP1(q, uint, u, 8, 16); |
| 154 |
TEST_VUZP1(q, uint, u, 16, 8); |
| 155 |
TEST_VUZP1(q, uint, u, 32, 4); |
| 156 |
TEST_VUZP1(q, uint, u, 64, 2); |
| 157 |
TEST_VUZP1(q, poly, p, 8, 16); |
| 158 |
TEST_VUZP1(q, poly, p, 16, 8); |
| 159 |
MFLOAT8_ONLY(TEST_VUZP1(q, mfloat, mf, 8, 16)); |
| 160 |
#if defined (FP16_SUPPORTED) |
| 161 |
TEST_VUZP1(q, float, f, 16, 8); |
| 162 |
#endif |
| 163 |
TEST_VUZP1(q, float, f, 32, 4); |
| 164 |
TEST_VUZP1(q, float, f, 64, 2); |
| 165 |
|
| 166 |
#if defined (FP16_SUPPORTED) |
| 167 |
CHECK_RESULTS (TEST_MSG, ""); |
| 168 |
#else |
| 169 |
CHECK_RESULTS_NO_FP16 (TEST_MSG, ""); |
| 170 |
#endif |
| 171 |
|
| 172 |
#undef TEST_MSG |
| 173 |
#define TEST_MSG "VUZP2" |
| 174 |
|
| 175 |
#define TEST_VUZP2(Q, T1, T2, W, N) TEST_VUZP(2, Q, T1, T2, W, N) |
| 176 |
|
| 177 |
/* Expected results. */ |
| 178 |
VECT_VAR_DECL(expected2,int,8,8) [] = { 0xf1, 0xf3, 0xf5, 0xf7, |
| 179 |
0x11, 0x11, 0x11, 0x11 }; |
| 180 |
VECT_VAR_DECL(expected2,int,16,4) [] = { 0xfff1, 0xfff3, 0x22, 0x22 }; |
| 181 |
VECT_VAR_DECL(expected2,int,32,2) [] = { 0xfffffff1, 0x33 }; |
| 182 |
VECT_VAR_DECL(expected2,int,64,1) [] = { 0xfffffffffffffff1 }; |
| 183 |
VECT_VAR_DECL(expected2,uint,8,8) [] = { 0xf1, 0xf3, 0xf5, 0xf7, |
| 184 |
0x55, 0x55, 0x55, 0x55 }; |
| 185 |
VECT_VAR_DECL(expected2,uint,16,4) [] = { 0xfff1, 0xfff3, 0x66, 0x66 }; |
| 186 |
VECT_VAR_DECL(expected2,uint,32,2) [] = { 0xfffffff1, 0x77 }; |
| 187 |
VECT_VAR_DECL(expected2,uint,64,1) [] = { 0xfffffffffffffff1 }; |
| 188 |
VECT_VAR_DECL(expected2,poly,8,8) [] = { 0xf1, 0xf3, 0xf5, 0xf7, |
| 189 |
0x55, 0x55, 0x55, 0x55 }; |
| 190 |
VECT_VAR_DECL(expected2,poly,16,4) [] = { 0xfff1, 0xfff3, 0x66, 0x66 }; |
| 191 |
#if MFLOAT8_SUPPORTED |
| 192 |
VECT_VAR_DECL(expected2,hmfloat,8,8) [] = { 0xf1, 0xf3, 0xf5, 0xf7, |
| 193 |
0x7b, 0x7b, 0x7b, 0x7b }; |
| 194 |
#endif |
| 195 |
VECT_VAR_DECL(expected2,hfloat,32,2) [] = { 0xc1700000, 0x42066666 }; |
| 196 |
#if defined (FP16_SUPPORTED) |
| 197 |
VECT_VAR_DECL (expected2, hfloat, 16, 4) [] = { 0xcb80, 0xca80, |
| 198 |
0x4b4d, 0x4b4d }; |
| 199 |
#endif |
| 200 |
VECT_VAR_DECL(expected2,int,8,16) [] = { 0xf1, 0xf3, 0xf5, 0xf7, |
| 201 |
0xf9, 0xfb, 0xfd, 0xff, |
| 202 |
0x11, 0x11, 0x11, 0x11, |
| 203 |
0x11, 0x11, 0x11, 0x11 }; |
| 204 |
VECT_VAR_DECL(expected2,int,16,8) [] = { 0xfff1, 0xfff3, 0xfff5, 0xfff7, |
| 205 |
0x22, 0x22, 0x22, 0x22 }; |
| 206 |
VECT_VAR_DECL(expected2,int,32,4) [] = { 0xfffffff1, 0xfffffff3, |
| 207 |
0x33, 0x33 }; |
| 208 |
VECT_VAR_DECL(expected2,int,64,2) [] = { 0xfffffffffffffff1, |
| 209 |
0x44 }; |
| 210 |
VECT_VAR_DECL(expected2,uint,8,16) [] = { 0xf1, 0xf3, 0xf5, 0xf7, |
| 211 |
0xf9, 0xfb, 0xfd, 0xff, |
| 212 |
0x55, 0x55, 0x55, 0x55, |
| 213 |
0x55, 0x55, 0x55, 0x55 }; |
| 214 |
VECT_VAR_DECL(expected2,uint,16,8) [] = { 0xfff1, 0xfff3, 0xfff5, 0xfff7, |
| 215 |
0x66, 0x66, 0x66, 0x66 }; |
| 216 |
VECT_VAR_DECL(expected2,uint,32,4) [] = { 0xfffffff1, 0xfffffff3, 0x77, 0x77 }; |
| 217 |
VECT_VAR_DECL(expected2,uint,64,2) [] = { 0xfffffffffffffff1, |
| 218 |
0x88 }; |
| 219 |
VECT_VAR_DECL(expected2,poly,8,16) [] = { 0xf1, 0xf3, 0xf5, 0xf7, |
| 220 |
0xf9, 0xfb, 0xfd, 0xff, |
| 221 |
0x55, 0x55, 0x55, 0x55, |
| 222 |
0x55, 0x55, 0x55, 0x55 }; |
| 223 |
VECT_VAR_DECL(expected2,poly,16,8) [] = { 0xfff1, 0xfff3, 0xfff5, 0xfff7, |
| 224 |
0x66, 0x66, 0x66, 0x66 }; |
| 225 |
#if MFLOAT8_SUPPORTED |
| 226 |
VECT_VAR_DECL(expected2,hmfloat,8,16) [] = { 0xf1, 0xf3, 0xf5, 0xf7, |
| 227 |
0xf9, 0xfb, 0xfd, 0xff, |
| 228 |
0x92, 0x92, 0x92, 0x92, |
| 229 |
0x92, 0x92, 0x92, 0x92 }; |
| 230 |
#endif |
| 231 |
#if defined (FP16_SUPPORTED) |
| 232 |
VECT_VAR_DECL (expected2, hfloat, 16, 8) [] = { 0xcb80, 0xca80, 0xc980, 0xc880, |
| 233 |
0x4b4d, 0x4b4d, 0x4b4d, 0x4b4d |
| 234 |
}; |
| 235 |
#endif |
| 236 |
VECT_VAR_DECL(expected2,hfloat,32,4) [] = { 0xc1700000, 0xc1500000, |
| 237 |
0x42073333, 0x42073333 }; |
| 238 |
|
| 239 |
clean_results (); |
| 240 |
CLEAN(expected2, int, 64, 1); |
| 241 |
CLEAN(expected2, uint, 64, 1); |
| 242 |
|
| 243 |
TEST_VUZP2(, int, s, 8, 8); |
| 244 |
TEST_VUZP2(, int, s, 16, 4); |
| 245 |
TEST_VUZP2(, int, s, 32, 2); |
| 246 |
TEST_VUZP2(, uint, u, 8, 8); |
| 247 |
TEST_VUZP2(, uint, u, 16, 4); |
| 248 |
TEST_VUZP2(, uint, u, 32, 2); |
| 249 |
TEST_VUZP2(, poly, p, 8, 8); |
| 250 |
TEST_VUZP2(, poly, p, 16, 4); |
| 251 |
MFLOAT8_ONLY(TEST_VUZP2(, mfloat, mf, 8, 8)); |
| 252 |
#if defined (FP16_SUPPORTED) |
| 253 |
TEST_VUZP2(, float, f, 16, 4); |
| 254 |
#endif |
| 255 |
TEST_VUZP2(, float, f, 32, 2); |
| 256 |
|
| 257 |
TEST_VUZP2(q, int, s, 8, 16); |
| 258 |
TEST_VUZP2(q, int, s, 16, 8); |
| 259 |
TEST_VUZP2(q, int, s, 32, 4); |
| 260 |
TEST_VUZP2(q, int, s, 64, 2); |
| 261 |
TEST_VUZP2(q, uint, u, 8, 16); |
| 262 |
TEST_VUZP2(q, uint, u, 16, 8); |
| 263 |
TEST_VUZP2(q, uint, u, 32, 4); |
| 264 |
TEST_VUZP2(q, uint, u, 64, 2); |
| 265 |
TEST_VUZP2(q, poly, p, 8, 16); |
| 266 |
TEST_VUZP2(q, poly, p, 16, 8); |
| 267 |
MFLOAT8_ONLY(TEST_VUZP2(q, mfloat, mf, 8, 16)); |
| 268 |
#if defined (FP16_SUPPORTED) |
| 269 |
TEST_VUZP2(q, float, f, 16, 8); |
| 270 |
#endif |
| 271 |
TEST_VUZP2(q, float, f, 32, 4); |
| 272 |
TEST_VUZP2(q, float, f, 64, 2); |
| 273 |
|
| 274 |
CHECK_RESULTS_NAMED (TEST_MSG, expected2, ""); |
| 275 |
#if defined (FP16_SUPPORTED) |
| 276 |
CHECK_FP(TEST_MSG, float, 16, 4, PRIx16, expected2, ""); |
| 277 |
CHECK_FP(TEST_MSG, float, 16, 8, PRIx16, expected2, ""); |
| 278 |
#endif |
| 279 |
} |
| 280 |
|
| 281 |
int main (void) |
| 282 |
{ |
| 283 |
exec_vuzp_half (); |
| 284 |
return 0; |
| 285 |
} |