| 1 |
/* { dg-do compile } */ |
| 2 |
/* { dg-additional-options "-march=rv32gcv_zvfh -mabi=ilp32d -mrvv-vector-bits=scalable -ffast-math -fno-vect-cost-model" } */ |
| 3 |
|
| 4 |
#include <stdint-gcc.h> |
| 5 |
|
| 6 |
#define NUM_ELEMS(TYPE) (1024 / sizeof (TYPE)) |
| 7 |
|
| 8 |
#define DEF_REDUC_PLUS(TYPE) \ |
| 9 |
void __attribute__ ((noinline, noclone)) \ |
| 10 |
reduc_plus_##TYPE (TYPE (*restrict a)[NUM_ELEMS (TYPE)], \ |
| 11 |
TYPE *restrict r, int n) \ |
| 12 |
{ \ |
| 13 |
for (int i = 0; i < n; i++) \ |
| 14 |
{ \ |
| 15 |
r[i] = 0; \ |
| 16 |
for (int j = 0; j < NUM_ELEMS (TYPE); j++) \ |
| 17 |
r[i] += a[i][j]; \ |
| 18 |
} \ |
| 19 |
} |
| 20 |
|
| 21 |
#define TEST_PLUS(T) \ |
| 22 |
T (int8_t) \ |
| 23 |
T (int16_t) \ |
| 24 |
T (int32_t) \ |
| 25 |
T (int64_t) \ |
| 26 |
T (uint8_t) \ |
| 27 |
T (uint16_t) \ |
| 28 |
T (uint32_t) \ |
| 29 |
T (uint64_t) \ |
| 30 |
T (_Float16) \ |
| 31 |
T (float) \ |
| 32 |
T (double) |
| 33 |
|
| 34 |
TEST_PLUS (DEF_REDUC_PLUS) |
| 35 |
|
| 36 |
#define DEF_REDUC_MAXMIN(TYPE, NAME, CMP_OP) \ |
| 37 |
void __attribute__ ((noinline, noclone)) \ |
| 38 |
reduc_##NAME##_##TYPE (TYPE (*restrict a)[NUM_ELEMS (TYPE)], \ |
| 39 |
TYPE *restrict r, int n) \ |
| 40 |
{ \ |
| 41 |
for (int i = 0; i < n; i++) \ |
| 42 |
{ \ |
| 43 |
r[i] = a[i][0]; \ |
| 44 |
for (int j = 0; j < NUM_ELEMS (TYPE); j++) \ |
| 45 |
r[i] = a[i][j] CMP_OP r[i] ? a[i][j] : r[i]; \ |
| 46 |
} \ |
| 47 |
} |
| 48 |
|
| 49 |
#define TEST_MAXMIN(T) \ |
| 50 |
T (int8_t, max, >) \ |
| 51 |
T (int16_t, max, >) \ |
| 52 |
T (int32_t, max, >) \ |
| 53 |
T (int64_t, max, >) \ |
| 54 |
T (uint8_t, max, >) \ |
| 55 |
T (uint16_t, max, >) \ |
| 56 |
T (uint32_t, max, >) \ |
| 57 |
T (uint64_t, max, >) \ |
| 58 |
T (_Float16, max, >) \ |
| 59 |
T (float, max, >) \ |
| 60 |
T (double, max, >) \ |
| 61 |
\ |
| 62 |
T (int8_t, min, <) \ |
| 63 |
T (int16_t, min, <) \ |
| 64 |
T (int32_t, min, <) \ |
| 65 |
T (int64_t, min, <) \ |
| 66 |
T (uint8_t, min, <) \ |
| 67 |
T (uint16_t, min, <) \ |
| 68 |
T (uint32_t, min, <) \ |
| 69 |
T (uint64_t, min, <) \ |
| 70 |
T (_Float16, min, <) \ |
| 71 |
T (float, min, <) \ |
| 72 |
T (double, min, <) |
| 73 |
|
| 74 |
TEST_MAXMIN (DEF_REDUC_MAXMIN) |
| 75 |
|
| 76 |
#define DEF_REDUC_BITWISE(TYPE,NAME,BIT_OP) \ |
| 77 |
void __attribute__ ((noinline, noclone)) \ |
| 78 |
reduc_##NAME##TYPE (TYPE (*restrict a)[NUM_ELEMS(TYPE)], \ |
| 79 |
TYPE *restrict r, int n) \ |
| 80 |
{ \ |
| 81 |
for (int i = 0; i < n; i++) \ |
| 82 |
{ \ |
| 83 |
r[i] = a[i][0]; \ |
| 84 |
for (int j = 0; j < NUM_ELEMS(TYPE); j++) \ |
| 85 |
r[i] BIT_OP a[i][j]; \ |
| 86 |
} \ |
| 87 |
} |
| 88 |
|
| 89 |
#define TEST_BITWISE(T) \ |
| 90 |
T (int8_t, and, &=) \ |
| 91 |
T (int16_t, and, &=) \ |
| 92 |
T (int32_t, and, &=) \ |
| 93 |
T (int64_t, and, &=) \ |
| 94 |
T (uint8_t, and, &=) \ |
| 95 |
T (uint16_t, and, &=) \ |
| 96 |
T (uint32_t, and, &=) \ |
| 97 |
T (uint64_t, and, &=) \ |
| 98 |
\ |
| 99 |
T (int8_t, ior, |=) \ |
| 100 |
T (int16_t, ior, |=) \ |
| 101 |
T (int32_t, ior, |=) \ |
| 102 |
T (int64_t, ior, |=) \ |
| 103 |
T (uint8_t, ior, |=) \ |
| 104 |
T (uint16_t, ior, |=) \ |
| 105 |
T (uint32_t, ior, |=) \ |
| 106 |
T (uint64_t, ior, |=) \ |
| 107 |
\ |
| 108 |
T (int8_t, xor, ^=) \ |
| 109 |
T (int16_t, xor, ^=) \ |
| 110 |
T (int32_t, xor, ^=) \ |
| 111 |
T (int64_t, xor, ^=) \ |
| 112 |
T (uint8_t, xor, ^=) \ |
| 113 |
T (uint16_t, xor, ^=) \ |
| 114 |
T (uint32_t, xor, ^=) \ |
| 115 |
T (uint64_t, xor, ^=) |
| 116 |
|
| 117 |
TEST_BITWISE (DEF_REDUC_BITWISE) |
| 118 |
|
| 119 |
/* { dg-final { scan-assembler-times {vredsum\.vs\s+v[0-9]+,\s*v[0-9]+,\s*v[0-9]+} 8 } } */ |
| 120 |
/* { dg-final { scan-assembler-times {vredmax\.vs\s+v[0-9]+,\s*v[0-9]+,\s*v[0-9]+} 4 } } */ |
| 121 |
/* { dg-final { scan-assembler-times {vredmaxu\.vs\s+v[0-9]+,\s*v[0-9]+,\s*v[0-9]+} 4 } } */ |
| 122 |
/* { dg-final { scan-assembler-times {vredmin\.vs\s+v[0-9]+,\s*v[0-9]+,\s*v[0-9]+} 4 } } */ |
| 123 |
/* { dg-final { scan-assembler-times {vredminu\.vs\s+v[0-9]+,\s*v[0-9]+,\s*v[0-9]+} 4 } } */ |
| 124 |
/* { dg-final { scan-assembler-times {vredand\.vs\s+v[0-9]+,\s*v[0-9]+,\s*v[0-9]+} 8 } } */ |
| 125 |
/* { dg-final { scan-assembler-times {vredor\.vs\s+v[0-9]+,\s*v[0-9]+,\s*v[0-9]+} 8 } } */ |
| 126 |
/* { dg-final { scan-assembler-times {vredxor\.vs\s+v[0-9]+,\s*v[0-9]+,\s*v[0-9]+} 8 } } */ |
| 127 |
/* { dg-final { scan-assembler-times {vfredusum\.vs\s+v[0-9]+,\s*v[0-9]+,\s*v[0-9]+} 3 } } */ |
| 128 |
/* { dg-final { scan-assembler-times {vfredmax\.vs\s+v[0-9]+,\s*v[0-9]+,\s*v[0-9]+} 3 } } */ |
| 129 |
/* { dg-final { scan-assembler-times {vfredmin\.vs\s+v[0-9]+,\s*v[0-9]+,\s*v[0-9]+} 3 } } */ |