| 1 |
/* { dg-do assemble } */ |
| 2 |
/* { dg-require-effective-target arm_hard_ok } */ |
| 3 |
/* { dg-require-effective-target arm_v8_2a_bf16_neon_ok } */ |
| 4 |
/* { dg-add-options arm_v8_2a_bf16_neon } */ |
| 5 |
/* { dg-additional-options "-save-temps -O2 -mfloat-abi=hard" } */ |
| 6 |
/* { dg-final { check-function-bodies "**" "" } } */ |
| 7 |
|
| 8 |
#include "arm_neon.h" |
| 9 |
|
| 10 |
|
| 11 |
/* |
| 12 |
**test_vld2_bf16: |
| 13 |
** ... |
| 14 |
** vld[0-9]+.16 {d[0-9]+-d[0-9]+}, \[r[0-9]+\] |
| 15 |
** ... |
| 16 |
*/ |
| 17 |
bfloat16x4x2_t |
| 18 |
test_vld2_bf16 (bfloat16_t * ptr) |
| 19 |
{ |
| 20 |
return vld2_bf16 (ptr); |
| 21 |
} |
| 22 |
|
| 23 |
/* |
| 24 |
**test_vld2q_bf16: |
| 25 |
** ... |
| 26 |
** vld[0-9]+.16 {d[0-9]+-d[0-9]+}, \[r[0-9]+\] |
| 27 |
** ... |
| 28 |
*/ |
| 29 |
bfloat16x8x2_t |
| 30 |
test_vld2q_bf16 (bfloat16_t * ptr) |
| 31 |
{ |
| 32 |
return vld2q_bf16 (ptr); |
| 33 |
} |
| 34 |
|
| 35 |
/* |
| 36 |
**test_vld2_dup_bf16: |
| 37 |
** ... |
| 38 |
** vld[0-9]+.16 {d[0-9]+\[\], d[0-9]+\[\]}, \[r[0-9]+\] |
| 39 |
** ... |
| 40 |
*/ |
| 41 |
bfloat16x4x2_t |
| 42 |
test_vld2_dup_bf16 (bfloat16_t * ptr) |
| 43 |
{ |
| 44 |
return vld2_dup_bf16 (ptr); |
| 45 |
} |
| 46 |
|
| 47 |
/* |
| 48 |
**test_vld2q_dup_bf16: |
| 49 |
** ... |
| 50 |
** vld[0-9]+.16 {d[0-9]+, d[0-9]+, d[0-9]+, d[0-9]+}, \[r[0-9]+\] |
| 51 |
** ... |
| 52 |
*/ |
| 53 |
bfloat16x8x2_t |
| 54 |
test_vld2q_dup_bf16 (bfloat16_t * ptr) |
| 55 |
{ |
| 56 |
return vld2q_dup_bf16 (ptr); |
| 57 |
} |
| 58 |
|
| 59 |
/* |
| 60 |
**test_vld3_bf16: |
| 61 |
** ... |
| 62 |
** vld[0-9]+.16 {d[0-9]+-d[0-9]+}, \[r[0-9]+\] |
| 63 |
** ... |
| 64 |
*/ |
| 65 |
bfloat16x4x3_t |
| 66 |
test_vld3_bf16 (bfloat16_t * ptr) |
| 67 |
{ |
| 68 |
return vld3_bf16 (ptr); |
| 69 |
} |
| 70 |
|
| 71 |
/* |
| 72 |
**test_vld3q_bf16: |
| 73 |
** ... |
| 74 |
** vld[0-9]+.16 {d[0-9]+, d[0-9]+, d[0-9]+}, \[r[0-9]+\] |
| 75 |
** ... |
| 76 |
*/ |
| 77 |
bfloat16x8x3_t |
| 78 |
test_vld3q_bf16 (bfloat16_t * ptr) |
| 79 |
{ |
| 80 |
return vld3q_bf16 (ptr); |
| 81 |
} |
| 82 |
|
| 83 |
/* |
| 84 |
**test_vld3_dup_bf16: |
| 85 |
** ... |
| 86 |
** vld[0-9]+.16 {d[0-9]+\[\], d[0-9]+\[\], d[0-9]+\[\]}, \[r[0-9]+\] |
| 87 |
** ... |
| 88 |
*/ |
| 89 |
bfloat16x4x3_t |
| 90 |
test_vld3_dup_bf16 (bfloat16_t * ptr) |
| 91 |
{ |
| 92 |
return vld3_dup_bf16 (ptr); |
| 93 |
} |
| 94 |
|
| 95 |
/* |
| 96 |
**test_vld3q_dup_bf16: |
| 97 |
** ... |
| 98 |
** vld[0-9]+.16 {d[0-9]+\[\], d[0-9]+\[\], d[0-9]+\[\]}, \[r[0-9]+\] |
| 99 |
** ... |
| 100 |
*/ |
| 101 |
bfloat16x8x3_t |
| 102 |
test_vld3q_dup_bf16 (bfloat16_t * ptr) |
| 103 |
{ |
| 104 |
return vld3q_dup_bf16 (ptr); |
| 105 |
} |
| 106 |
|
| 107 |
/* |
| 108 |
**test_vld4_bf16: |
| 109 |
** ... |
| 110 |
** vld4.16 {d[0-9]+-d[0-9]+}, \[r[0-9]+\] |
| 111 |
** ... |
| 112 |
*/ |
| 113 |
bfloat16x4x4_t |
| 114 |
test_vld4_bf16 (bfloat16_t * ptr) |
| 115 |
{ |
| 116 |
return vld4_bf16 (ptr); |
| 117 |
} |
| 118 |
|
| 119 |
/* |
| 120 |
**test_vld4q_bf16: |
| 121 |
** ... |
| 122 |
** vld4.16 {d[0-9]+, d[0-9]+, d[0-9]+, d[0-9]+}, \[r[0-9]+\] |
| 123 |
** ... |
| 124 |
*/ |
| 125 |
bfloat16x8x4_t |
| 126 |
test_vld4q_bf16 (bfloat16_t * ptr) |
| 127 |
{ |
| 128 |
return vld4q_bf16 (ptr); |
| 129 |
} |
| 130 |
|
| 131 |
/* |
| 132 |
**test_vld4_dup_bf16: |
| 133 |
** ... |
| 134 |
** vld4.16 {d[0-9]+\[\], d[0-9]+\[\], d[0-9]+\[\], d[0-9]+\[\]}, \[r[0-9]+\] |
| 135 |
** ... |
| 136 |
*/ |
| 137 |
bfloat16x4x4_t |
| 138 |
test_vld4_dup_bf16 (bfloat16_t * ptr) |
| 139 |
{ |
| 140 |
return vld4_dup_bf16 (ptr); |
| 141 |
} |
| 142 |
|
| 143 |
/* |
| 144 |
**test_vld4q_dup_bf16: |
| 145 |
** ... |
| 146 |
** vld4.16 {d[0-9]+\[\], d[0-9]+\[\], d[0-9]+\[\], d[0-9]+\[\]}, \[r[0-9]+\] |
| 147 |
** ... |
| 148 |
*/ |
| 149 |
bfloat16x8x4_t |
| 150 |
test_vld4q_dup_bf16 (bfloat16_t * ptr) |
| 151 |
{ |
| 152 |
return vld4q_dup_bf16 (ptr); |
| 153 |
} |