| 1 |
/* { dg-final { check-function-bodies "**" "" "-DCHECK_ASM" } } */ |
| 2 |
|
| 3 |
#include "test_sme2_acle.h" |
| 4 |
|
| 5 |
/* |
| 6 |
** add_write_0_z0_z0: |
| 7 |
** mov (w8|w9|w10|w11), #?0 |
| 8 |
** add za\.s\[\1, 0, vgx2\], {z0\.s - z1\.s}, {z0\.s - z1\.s} |
| 9 |
** ret |
| 10 |
*/ |
| 11 |
TEST_ZA_XN (add_write_0_z0_z0, svint32x2_t, |
| 12 |
svadd_write_za32_s32_vg1x2 (0, z0, z0), |
| 13 |
svadd_write_za32_vg1x2 (0, z0, z0)) |
| 14 |
|
| 15 |
/* |
| 16 |
** add_write_w0_z0_z0: |
| 17 |
** mov (w8|w9|w10|w11), w0 |
| 18 |
** add za\.s\[\1, 0, vgx2\], {z0\.s - z1\.s}, {z0\.s - z1\.s} |
| 19 |
** ret |
| 20 |
*/ |
| 21 |
TEST_ZA_XN (add_write_w0_z0_z0, svint32x2_t, |
| 22 |
svadd_write_za32_s32_vg1x2 (w0, z0, z0), |
| 23 |
svadd_write_za32_vg1x2 (w0, z0, z0)) |
| 24 |
|
| 25 |
/* |
| 26 |
** add_write_w8_z0_z4: |
| 27 |
** add za\.s\[w8, 0, vgx2\], {z0\.s - z1\.s}, {z4\.s - z5\.s} |
| 28 |
** ret |
| 29 |
*/ |
| 30 |
TEST_ZA_XN (add_write_w8_z0_z4, svint32x2_t, |
| 31 |
svadd_write_za32_s32_vg1x2 (w8, z0, z4), |
| 32 |
svadd_write_za32_vg1x2 (w8, z0, z4)) |
| 33 |
|
| 34 |
/* |
| 35 |
** add_write_w8_z4_z18: |
| 36 |
** add za\.s\[w8, 0, vgx2\], {z4\.s - z5\.s}, {z18\.s - z19\.s} |
| 37 |
** ret |
| 38 |
*/ |
| 39 |
TEST_ZA_XN (add_write_w8_z4_z18, svint32x2_t, |
| 40 |
svadd_write_za32_s32_vg1x2 (w8, z4, z18), |
| 41 |
svadd_write_za32_vg1x2 (w8, z4, z18)) |
| 42 |
|
| 43 |
/* Leave the assembler to check for correctness for misaligned registers. */ |
| 44 |
|
| 45 |
/* |
| 46 |
** add_write_w8_z23_z0: |
| 47 |
** ... |
| 48 |
** add za\.s\[w8, 0, vgx2\], [^\n]+, {z0\.s - z1\.s} |
| 49 |
** ret |
| 50 |
*/ |
| 51 |
TEST_ZA_XN (add_write_w8_z23_z0, svint32x2_t, |
| 52 |
svadd_write_za32_s32_vg1x2 (w8, z23, z0), |
| 53 |
svadd_write_za32_vg1x2 (w8, z23, z0)) |
| 54 |
|
| 55 |
/* |
| 56 |
** add_write_w8_z18_z23: |
| 57 |
** ... |
| 58 |
** add za\.s\[w8, 0, vgx2\], {z18\.s - z19\.s}, [^\n]+ |
| 59 |
** ret |
| 60 |
*/ |
| 61 |
TEST_ZA_XN (add_write_w8_z18_z23, svint32x2_t, |
| 62 |
svadd_write_za32_s32_vg1x2 (w8, z18, z23), |
| 63 |
svadd_write_za32_vg1x2 (w8, z18, z23)) |
| 64 |
|
| 65 |
/* |
| 66 |
** add_write_w8_z4_z28: |
| 67 |
** add za\.s\[w8, 0, vgx2\], {z4\.s - z5\.s}, {z28\.s - z29\.s} |
| 68 |
** ret |
| 69 |
*/ |
| 70 |
TEST_ZA_XN (add_write_w8_z4_z28, svint32x2_t, |
| 71 |
svadd_write_za32_s32_vg1x2 (w8, z4, z28), |
| 72 |
svadd_write_za32_vg1x2 (w8, z4, z28)) |
| 73 |
|
| 74 |
/* |
| 75 |
** add_write_w8p7_z4_z0: |
| 76 |
** add za\.s\[w8, 7, vgx2\], {z4\.s - z5\.s}, {z0\.s - z1\.s} |
| 77 |
** ret |
| 78 |
*/ |
| 79 |
TEST_ZA_XN (add_write_w8p7_z4_z0, svint32x2_t, |
| 80 |
svadd_write_za32_s32_vg1x2 (w8 + 7, z4, z0), |
| 81 |
svadd_write_za32_vg1x2 (w8 + 7, z4, z0)) |
| 82 |
|
| 83 |
/* |
| 84 |
** add_write_w8p8_z4_z4: |
| 85 |
** add (w8|w9|w10|w11), w8, #?8 |
| 86 |
** add za\.s\[\1, 0, vgx2\], {z4\.s - z5\.s}, {z4\.s - z5\.s} |
| 87 |
** ret |
| 88 |
*/ |
| 89 |
TEST_ZA_XN (add_write_w8p8_z4_z4, svint32x2_t, |
| 90 |
svadd_write_za32_s32_vg1x2 (w8 + 8, z4, z4), |
| 91 |
svadd_write_za32_vg1x2 (w8 + 8, z4, z4)) |
| 92 |
|
| 93 |
/* |
| 94 |
** add_write_w8m1_z4_z0: |
| 95 |
** sub (w8|w9|w10|w11), w8, #?1 |
| 96 |
** add za\.s\[\1, 0, vgx2\], {z4\.s - z5\.s}, {z0\.s - z1\.s} |
| 97 |
** ret |
| 98 |
*/ |
| 99 |
TEST_ZA_XN (add_write_w8m1_z4_z0, svint32x2_t, |
| 100 |
svadd_write_za32_s32_vg1x2 (w8 - 1, z4, z0), |
| 101 |
svadd_write_za32_vg1x2 (w8 - 1, z4, z0)) |
| 102 |
|
| 103 |
/* |
| 104 |
** add_write_single_0_z1_z0: |
| 105 |
** mov (w8|w9|w10|w11), #?0 |
| 106 |
** add za\.s\[\1, 0, vgx2\], {z1\.s - z2\.s}, z0\.s |
| 107 |
** ret |
| 108 |
*/ |
| 109 |
TEST_ZA_SINGLE (add_write_single_0_z1_z0, svint32x2_t, svint32_t, |
| 110 |
svadd_write_single_za32_s32_vg1x2 (0, z1, z0), |
| 111 |
svadd_write_za32_vg1x2 (0, z1, z0)) |
| 112 |
|
| 113 |
/* |
| 114 |
** add_write_single_w0_z1_z0: |
| 115 |
** mov (w8|w9|w10|w11), w0 |
| 116 |
** add za\.s\[\1, 0, vgx2\], {z1\.s - z2\.s}, z0\.s |
| 117 |
** ret |
| 118 |
*/ |
| 119 |
TEST_ZA_SINGLE (add_write_single_w0_z1_z0, svint32x2_t, svint32_t, |
| 120 |
svadd_write_single_za32_s32_vg1x2 (w0, z1, z0), |
| 121 |
svadd_write_za32_vg1x2 (w0, z1, z0)) |
| 122 |
|
| 123 |
/* |
| 124 |
** add_write_single_w8_z1_z0: |
| 125 |
** add za\.s\[w8, 0, vgx2\], {z1\.s - z2\.s}, z0\.s |
| 126 |
** ret |
| 127 |
*/ |
| 128 |
TEST_ZA_SINGLE (add_write_single_w8_z1_z0, svint32x2_t, svint32_t, |
| 129 |
svadd_write_single_za32_s32_vg1x2 (w8, z1, z0), |
| 130 |
svadd_write_za32_vg1x2 (w8, z1, z0)) |
| 131 |
|
| 132 |
/* |
| 133 |
** add_write_single_w8p7_z1_z0: |
| 134 |
** add za\.s\[w8, 7, vgx2\], {z1\.s - z2\.s}, z0\.s |
| 135 |
** ret |
| 136 |
*/ |
| 137 |
TEST_ZA_SINGLE (add_write_single_w8p7_z1_z0, svint32x2_t, svint32_t, |
| 138 |
svadd_write_single_za32_s32_vg1x2 (w8 + 7, z1, z0), |
| 139 |
svadd_write_za32_vg1x2 (w8 + 7, z1, z0)) |
| 140 |
|
| 141 |
/* |
| 142 |
** add_write_single_w8p8_z1_z0: |
| 143 |
** add (w8|w9|w10|w11), w8, #?8 |
| 144 |
** add za\.s\[\1, 0, vgx2\], {z1\.s - z2\.s}, z0\.s |
| 145 |
** ret |
| 146 |
*/ |
| 147 |
TEST_ZA_SINGLE (add_write_single_w8p8_z1_z0, svint32x2_t, svint32_t, |
| 148 |
svadd_write_single_za32_s32_vg1x2 (w8 + 8, z1, z0), |
| 149 |
svadd_write_za32_vg1x2 (w8 + 8, z1, z0)) |
| 150 |
|
| 151 |
/* |
| 152 |
** add_write_single_w0m1_z1_z0: |
| 153 |
** sub (w8|w9|w10|w11), w0, #?1 |
| 154 |
** add za\.s\[\1, 0, vgx2\], {z1\.s - z2\.s}, z0\.s |
| 155 |
** ret |
| 156 |
*/ |
| 157 |
TEST_ZA_SINGLE (add_write_single_w0m1_z1_z0, svint32x2_t, svint32_t, |
| 158 |
svadd_write_single_za32_s32_vg1x2 (w0 - 1, z1, z0), |
| 159 |
svadd_write_za32_vg1x2 (w0 - 1, z1, z0)) |
| 160 |
|
| 161 |
/* |
| 162 |
** add_write_single_w8_z0_z15: |
| 163 |
** str d15, \[sp, #?-16\]! |
| 164 |
** add za\.s\[w8, 0, vgx2\], {z0\.s - z1\.s}, z15\.s |
| 165 |
** ldr d15, \[sp\], #?16 |
| 166 |
** ret |
| 167 |
*/ |
| 168 |
TEST_ZA_SINGLE_Z15 (add_write_single_w8_z0_z15, svint32x2_t, svint32_t, |
| 169 |
svadd_write_single_za32_s32_vg1x2 (w8, z0, z15), |
| 170 |
svadd_write_za32_vg1x2 (w8, z0, z15)) |
| 171 |
|
| 172 |
/* |
| 173 |
** add_write_single_w8_z20_z16: |
| 174 |
** mov (z[0-7]).d, z16.d |
| 175 |
** add za\.s\[w8, 0, vgx2\], {z20\.s - z21\.s}, \1\.s |
| 176 |
** ret |
| 177 |
*/ |
| 178 |
TEST_ZA_SINGLE (add_write_single_w8_z20_z16, svint32x2_t, svint32_t, |
| 179 |
svadd_write_single_za32_s32_vg1x2 (w8, z20, z16), |
| 180 |
svadd_write_za32_vg1x2 (w8, z20, z16)) |