| 1 |
/* { dg-do compile } */ |
| 2 |
/* { dg-options "-march=rv64gcv_zvfhmin -mabi=lp64d -O3 -mrvv-vector-bits=zvl" } */ |
| 3 |
|
| 4 |
#include <stdint-gcc.h> |
| 5 |
|
| 6 |
typedef int8_t vnx128qi __attribute__((vector_size (128))); |
| 7 |
typedef int16_t vnx64hi __attribute__((vector_size (128))); |
| 8 |
typedef int32_t vnx32si __attribute__((vector_size (128))); |
| 9 |
typedef int64_t vnx16di __attribute__((vector_size (128))); |
| 10 |
typedef uint8_t vnx128uqi __attribute__((vector_size (128))); |
| 11 |
typedef uint16_t vnx64uhi __attribute__((vector_size (128))); |
| 12 |
typedef uint32_t vnx32usi __attribute__((vector_size (128))); |
| 13 |
typedef uint64_t vnx16udi __attribute__((vector_size (128))); |
| 14 |
|
| 15 |
typedef _Float16 vnx64hf __attribute__((vector_size (128))); |
| 16 |
typedef float vnx32sf __attribute__((vector_size (128))); |
| 17 |
typedef double vnx16df __attribute__((vector_size (128))); |
| 18 |
|
| 19 |
#define MASK_128 0, 129, 2, 131, 4, 133, 6, 135, 8, 137, \ |
| 20 |
10, 139, 12, 141, 14, 143, 16, 145, 18, 147, \ |
| 21 |
20, 149, 22, 151, 24, 153, 26, 155, 28, 157, 30, 159, \ |
| 22 |
32, 161, 34, 163, 36, 165, 38, 167, 40, 169, \ |
| 23 |
42, 171, 44, 173, 46, 175, 48, 177, 50, 179, \ |
| 24 |
52, 181, 54, 183, 56, 185, 58, 187, 60, 189, \ |
| 25 |
62, 191, \ |
| 26 |
64, 193, 66, 195, 68, 197, 70, 199, 72, 201, \ |
| 27 |
74, 203, 76, 205, 78, 207, 80, 209, 82, 211, \ |
| 28 |
84, 213, 86, 215, 88, 217, 90, 219, 92, 221, 94, 223, \ |
| 29 |
96, 225, 98, 227, 100, 229, 102, 231, 104, 233, \ |
| 30 |
106, 235, 108, 237, 110, 239, 112, 241, 114, 243, \ |
| 31 |
116, 245, 118, 247, 120, 249, 122, 251, 124, 253, \ |
| 32 |
126, 255 |
| 33 |
#define MASK_64 0, 65, 2, 67, 4, 69, 6, 71, 8, 73, \ |
| 34 |
10, 75, 12, 77, 14, 79, 16, 81, 18, 83, \ |
| 35 |
20, 85, 22, 87, 24, 89, 26, 91, 28, 93, 30, 95, \ |
| 36 |
32, 97, 34, 99, 36, 101, 38, 103, 40, 105, \ |
| 37 |
42, 107, 44, 109, 46, 111, 48, 113, 50, 115, \ |
| 38 |
52, 117, 54, 119, 56, 121, 58, 123, 60, 125, \ |
| 39 |
62, 127 |
| 40 |
#define MASK_32 0, 33, 2, 35, 4, 37, 6, 39, 8, 41, \ |
| 41 |
10, 43, 12, 45, 14, 47, 16, 49, 18, 51, \ |
| 42 |
20, 53, 22, 55, 24, 57, 26, 59, 28, 61, 30, 63 |
| 43 |
#define MASK_16 0, 17, 2, 19, 4, 21, 6, 23, 8, 25, 10, 27, 12, 29, 14, 31 |
| 44 |
|
| 45 |
void __attribute__ ((noipa)) |
| 46 |
merge0 (vnx128qi x, vnx128qi y, vnx128qi *out) |
| 47 |
{ |
| 48 |
vnx128qi v = __builtin_shufflevector ((vnx128qi) x, (vnx128qi) y, MASK_128); |
| 49 |
*(vnx128qi*)out = v; |
| 50 |
} |
| 51 |
|
| 52 |
void __attribute__ ((noipa)) |
| 53 |
merge1 (vnx128uqi x, vnx128uqi y, vnx128uqi *out) |
| 54 |
{ |
| 55 |
vnx128uqi v = __builtin_shufflevector ((vnx128uqi) x, (vnx128uqi) y, MASK_128); |
| 56 |
*(vnx128uqi*)out = v; |
| 57 |
} |
| 58 |
|
| 59 |
void __attribute__ ((noipa)) |
| 60 |
merge2 (vnx64hi x, vnx64hi y, vnx64hi *out) |
| 61 |
{ |
| 62 |
vnx64hi v = __builtin_shufflevector ((vnx64hi) x, (vnx64hi) y, MASK_64); |
| 63 |
*(vnx64hi*)out = v; |
| 64 |
} |
| 65 |
|
| 66 |
void __attribute__ ((noipa)) |
| 67 |
merge3 (vnx64uhi x, vnx64uhi y, vnx64uhi *out) |
| 68 |
{ |
| 69 |
vnx64uhi v = __builtin_shufflevector ((vnx64uhi) x, (vnx64uhi) y, MASK_64); |
| 70 |
*(vnx64uhi*)out = v; |
| 71 |
} |
| 72 |
|
| 73 |
void __attribute__ ((noipa)) |
| 74 |
merge4 (vnx32si x, vnx32si y, vnx32si *out) |
| 75 |
{ |
| 76 |
vnx32si v = __builtin_shufflevector ((vnx32si) x, (vnx32si) y, MASK_32); |
| 77 |
*(vnx32si*)out = v; |
| 78 |
} |
| 79 |
|
| 80 |
void __attribute__ ((noipa)) |
| 81 |
merge5 (vnx32usi x, vnx32usi y, vnx32usi *out) |
| 82 |
{ |
| 83 |
vnx32usi v = __builtin_shufflevector ((vnx32usi) x, (vnx32usi) y, MASK_32); |
| 84 |
*(vnx32usi*)out = v; |
| 85 |
} |
| 86 |
|
| 87 |
void __attribute__ ((noipa)) |
| 88 |
merge6 (vnx16di x, vnx16di y, vnx16di *out) |
| 89 |
{ |
| 90 |
vnx16di v = __builtin_shufflevector ((vnx16di) x, (vnx16di) y, MASK_16); |
| 91 |
*(vnx16di*)out = v; |
| 92 |
} |
| 93 |
|
| 94 |
void __attribute__ ((noipa)) |
| 95 |
merge7 (vnx16udi x, vnx16udi y, vnx16udi *out) |
| 96 |
{ |
| 97 |
vnx16udi v = __builtin_shufflevector ((vnx16udi) x, (vnx16udi) y, MASK_16); |
| 98 |
*(vnx16udi*)out = v; |
| 99 |
} |
| 100 |
|
| 101 |
void __attribute__ ((noipa)) |
| 102 |
merge8 (vnx64hf x, vnx64hf y, vnx64hf *out) |
| 103 |
{ |
| 104 |
vnx64hf v = __builtin_shufflevector ((vnx64hf) x, (vnx64hf) y, MASK_64); |
| 105 |
*(vnx64hf*)out = v; |
| 106 |
} |
| 107 |
|
| 108 |
void __attribute__ ((noipa)) |
| 109 |
merge9 (vnx32sf x, vnx32sf y, vnx32sf *out) |
| 110 |
{ |
| 111 |
vnx32sf v = __builtin_shufflevector ((vnx32sf) x, (vnx32sf) y, MASK_32); |
| 112 |
*(vnx32sf*)out = v; |
| 113 |
} |
| 114 |
|
| 115 |
void __attribute__ ((noipa)) |
| 116 |
merge10 (vnx16df x, vnx16df y, vnx16df *out) |
| 117 |
{ |
| 118 |
vnx16df v = __builtin_shufflevector ((vnx16df) x, (vnx16df) y, MASK_16); |
| 119 |
*(vnx16df*)out = v; |
| 120 |
} |
| 121 |
|
| 122 |
/* { dg-final { scan-assembler-times {\tvmerge.vvm} 11 } } */ |