| 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 vnx32qi __attribute__((vector_size (32))); |
| 7 |
typedef int16_t vnx16hi __attribute__((vector_size (32))); |
| 8 |
typedef int32_t vnx8si __attribute__((vector_size (32))); |
| 9 |
typedef int64_t vnx4di __attribute__((vector_size (32))); |
| 10 |
typedef uint8_t vnx32uqi __attribute__((vector_size (32))); |
| 11 |
typedef uint16_t vnx16uhi __attribute__((vector_size (32))); |
| 12 |
typedef uint32_t vnx8usi __attribute__((vector_size (32))); |
| 13 |
typedef uint64_t vnx4udi __attribute__((vector_size (32))); |
| 14 |
|
| 15 |
typedef _Float16 vnx16hf __attribute__((vector_size (32))); |
| 16 |
typedef float vnx8sf __attribute__((vector_size (32))); |
| 17 |
typedef double vnx4df __attribute__((vector_size (32))); |
| 18 |
|
| 19 |
#define MASK_32 0, 33, 2, 35, 4, 37, 6, 39, 8, 41, \ |
| 20 |
10, 43, 12, 45, 14, 47, 16, 49, 18, 51, \ |
| 21 |
20, 53, 22, 55, 24, 57, 26, 59, 28, 61, 30, 63 |
| 22 |
#define MASK_16 0, 17, 2, 19, 4, 21, 6, 23, 8, 25, 10, 27, 12, 29, 14, 31 |
| 23 |
#define MASK_8 0, 9, 2, 11, 4, 13, 6, 15 |
| 24 |
#define MASK_4 0, 5, 2, 7 |
| 25 |
|
| 26 |
void __attribute__ ((noipa)) |
| 27 |
merge0 (vnx32qi x, vnx32qi y, vnx32qi *out) |
| 28 |
{ |
| 29 |
vnx32qi v = __builtin_shufflevector ((vnx32qi) x, (vnx32qi) y, MASK_32); |
| 30 |
*(vnx32qi*)out = v; |
| 31 |
} |
| 32 |
|
| 33 |
void __attribute__ ((noipa)) |
| 34 |
merge1 (vnx32uqi x, vnx32uqi y, vnx32uqi *out) |
| 35 |
{ |
| 36 |
vnx32uqi v = __builtin_shufflevector ((vnx32uqi) x, (vnx32uqi) y, MASK_32); |
| 37 |
*(vnx32uqi*)out = v; |
| 38 |
} |
| 39 |
|
| 40 |
void __attribute__ ((noipa)) |
| 41 |
merge2 (vnx16hi x, vnx16hi y, vnx16hi *out) |
| 42 |
{ |
| 43 |
vnx16hi v = __builtin_shufflevector ((vnx16hi) x, (vnx16hi) y, MASK_16); |
| 44 |
*(vnx16hi*)out = v; |
| 45 |
} |
| 46 |
|
| 47 |
void __attribute__ ((noipa)) |
| 48 |
merge3 (vnx16uhi x, vnx16uhi y, vnx16uhi *out) |
| 49 |
{ |
| 50 |
vnx16uhi v = __builtin_shufflevector ((vnx16uhi) x, (vnx16uhi) y, MASK_16); |
| 51 |
*(vnx16uhi*)out = v; |
| 52 |
} |
| 53 |
|
| 54 |
void __attribute__ ((noipa)) |
| 55 |
merge4 (vnx8si x, vnx8si y, vnx8si *out) |
| 56 |
{ |
| 57 |
vnx8si v = __builtin_shufflevector ((vnx8si) x, (vnx8si) y, MASK_8); |
| 58 |
*(vnx8si*)out = v; |
| 59 |
} |
| 60 |
|
| 61 |
void __attribute__ ((noipa)) |
| 62 |
merge5 (vnx8usi x, vnx8usi y, vnx8usi *out) |
| 63 |
{ |
| 64 |
vnx8usi v = __builtin_shufflevector ((vnx8usi) x, (vnx8usi) y, MASK_8); |
| 65 |
*(vnx8usi*)out = v; |
| 66 |
} |
| 67 |
|
| 68 |
void __attribute__ ((noipa)) |
| 69 |
merge6 (vnx4di x, vnx4di y, vnx4di *out) |
| 70 |
{ |
| 71 |
vnx4di v = __builtin_shufflevector ((vnx4di) x, (vnx4di) y, MASK_4); |
| 72 |
*(vnx4di*)out = v; |
| 73 |
} |
| 74 |
|
| 75 |
void __attribute__ ((noipa)) |
| 76 |
merge7 (vnx4udi x, vnx4udi y, vnx4udi *out) |
| 77 |
{ |
| 78 |
vnx4udi v = __builtin_shufflevector ((vnx4udi) x, (vnx4udi) y, MASK_4); |
| 79 |
*(vnx4udi*)out = v; |
| 80 |
} |
| 81 |
|
| 82 |
void __attribute__ ((noipa)) |
| 83 |
merge8 (vnx16hf x, vnx16hf y, vnx16hf *out) |
| 84 |
{ |
| 85 |
vnx16hf v = __builtin_shufflevector ((vnx16hf) x, (vnx16hf) y, MASK_16); |
| 86 |
*(vnx16hf*)out = v; |
| 87 |
} |
| 88 |
|
| 89 |
void __attribute__ ((noipa)) |
| 90 |
merge9 (vnx8sf x, vnx8sf y, vnx8sf *out) |
| 91 |
{ |
| 92 |
vnx8sf v = __builtin_shufflevector ((vnx8sf) x, (vnx8sf) y, MASK_8); |
| 93 |
*(vnx8sf*)out = v; |
| 94 |
} |
| 95 |
|
| 96 |
void __attribute__ ((noipa)) |
| 97 |
merge10 (vnx4df x, vnx4df y, vnx4df *out) |
| 98 |
{ |
| 99 |
vnx4df v = __builtin_shufflevector ((vnx4df) x, (vnx4df) y, MASK_4); |
| 100 |
*(vnx4df*)out = v; |
| 101 |
} |
| 102 |
|
| 103 |
/* { dg-final { scan-assembler-times {\tvmerge.vvm} 11 } } */ |