| 1 |
/* { dg-do assemble { target aarch64_asm_sme2p1_ok } } */ |
| 2 |
/* { dg-do compile { target { ! aarch64_asm_sme2p1_ok } } } */ |
| 3 |
/* { dg-final { check-function-bodies "**" "" "-DCHECK_ASM" } } */ |
| 4 |
|
| 5 |
#include "test_sme2_acle.h" |
| 6 |
|
| 7 |
#pragma GCC target "+sme2p1" |
| 8 |
|
| 9 |
/* |
| 10 |
** readz_za16_s16_z0_0_0: |
| 11 |
** mov (w1[2-5]), (?:wzr|#?0) |
| 12 |
** movaz {z0\.h - z3\.h}, za0v\.h\[\1, 0:3\] |
| 13 |
** ret |
| 14 |
*/ |
| 15 |
TEST_READ_ZA_XN (readz_za16_s16_z0_0_0, svint16x4_t, |
| 16 |
z0 = svreadz_ver_za16_s16_vg4 (0, 0), |
| 17 |
z0 = svreadz_ver_za16_s16_vg4 (0, 0)) |
| 18 |
|
| 19 |
/* |
| 20 |
** readz_za16_u16_z4_1_1: |
| 21 |
** mov (w1[2-5]), #?1 |
| 22 |
** movaz {z4\.h - z7\.h}, za1v\.h\[\1, 0:3\] |
| 23 |
** ret |
| 24 |
*/ |
| 25 |
TEST_READ_ZA_XN (readz_za16_u16_z4_1_1, svuint16x4_t, |
| 26 |
z4 = svreadz_ver_za16_u16_vg4 (1, 1), |
| 27 |
z4 = svreadz_ver_za16_u16_vg4 (1, 1)) |
| 28 |
|
| 29 |
/* |
| 30 |
** readz_za16_f16_z28_0_w11: |
| 31 |
** mov (w1[2-5]), w11 |
| 32 |
** movaz {z28\.h - z31\.h}, za0v\.h\[\1, 0:3\] |
| 33 |
** ret |
| 34 |
*/ |
| 35 |
TEST_READ_ZA_XN (readz_za16_f16_z28_0_w11, svfloat16x4_t, |
| 36 |
z28 = svreadz_ver_za16_f16_vg4 (0, w11), |
| 37 |
z28 = svreadz_ver_za16_f16_vg4 (0, w11)) |
| 38 |
|
| 39 |
/* |
| 40 |
** readz_za16_s16_z0_1_w12: |
| 41 |
** movaz {z0\.h - z3\.h}, za1v\.h\[w12, 0:3\] |
| 42 |
** ret |
| 43 |
*/ |
| 44 |
TEST_READ_ZA_XN (readz_za16_s16_z0_1_w12, svint16x4_t, |
| 45 |
z0 = svreadz_ver_za16_s16_vg4 (1, w12), |
| 46 |
z0 = svreadz_ver_za16_s16_vg4 (1, w12)) |
| 47 |
|
| 48 |
/* |
| 49 |
** readz_za16_u16_z18_0_w15: |
| 50 |
** movaz {[^\n]+}, za0v\.h\[w15, 0:3\] |
| 51 |
** mov [^\n]+ |
| 52 |
** mov [^\n]+ |
| 53 |
** mov [^\n]+ |
| 54 |
** mov [^\n]+ |
| 55 |
** ret |
| 56 |
*/ |
| 57 |
TEST_READ_ZA_XN (readz_za16_u16_z18_0_w15, svuint16x4_t, |
| 58 |
z18 = svreadz_ver_za16_u16_vg4 (0, w15), |
| 59 |
z18 = svreadz_ver_za16_u16_vg4 (0, w15)) |
| 60 |
|
| 61 |
/* |
| 62 |
** readz_za16_bf16_z23_1_w12p4: |
| 63 |
** movaz {[^\n]+}, za1v\.h\[w12, 4:7\] |
| 64 |
** mov [^\n]+ |
| 65 |
** mov [^\n]+ |
| 66 |
** mov [^\n]+ |
| 67 |
** mov [^\n]+ |
| 68 |
** ret |
| 69 |
*/ |
| 70 |
TEST_READ_ZA_XN (readz_za16_bf16_z23_1_w12p4, svbfloat16x4_t, |
| 71 |
z23 = svreadz_ver_za16_bf16_vg4 (1, w12 + 4), |
| 72 |
z23 = svreadz_ver_za16_bf16_vg4 (1, w12 + 4)) |
| 73 |
|
| 74 |
/* |
| 75 |
** readz_za16_u16_z4_0_w12p1: |
| 76 |
** add (w[0-9]+), w12, #?1 |
| 77 |
** movaz {z4\.h - z7\.h}, za0v\.h\[\1, 0:3\] |
| 78 |
** ret |
| 79 |
*/ |
| 80 |
TEST_READ_ZA_XN (readz_za16_u16_z4_0_w12p1, svuint16x4_t, |
| 81 |
z4 = svreadz_ver_za16_u16_vg4 (0, w12 + 1), |
| 82 |
z4 = svreadz_ver_za16_u16_vg4 (0, w12 + 1)) |
| 83 |
|
| 84 |
/* |
| 85 |
** readz_za16_s16_z28_1_w12p2: |
| 86 |
** add (w[0-9]+), w12, #?2 |
| 87 |
** movaz {z28\.h - z31\.h}, za1v\.h\[\1, 0:3\] |
| 88 |
** ret |
| 89 |
*/ |
| 90 |
TEST_READ_ZA_XN (readz_za16_s16_z28_1_w12p2, svint16x4_t, |
| 91 |
z28 = svreadz_ver_za16_s16_vg4 (1, w12 + 2), |
| 92 |
z28 = svreadz_ver_za16_s16_vg4 (1, w12 + 2)) |
| 93 |
|
| 94 |
/* |
| 95 |
** readz_za16_f16_z0_0_w15p3: |
| 96 |
** add (w[0-9]+), w15, #?3 |
| 97 |
** movaz {z0\.h - z3\.h}, za0v\.h\[\1, 0:3\] |
| 98 |
** ret |
| 99 |
*/ |
| 100 |
TEST_READ_ZA_XN (readz_za16_f16_z0_0_w15p3, svfloat16x4_t, |
| 101 |
z0 = svreadz_ver_za16_f16_vg4 (0, w15 + 3), |
| 102 |
z0 = svreadz_ver_za16_f16_vg4 (0, w15 + 3)) |
| 103 |
|
| 104 |
/* |
| 105 |
** readz_za16_u16_z28_1_w12p6: |
| 106 |
** add (w[0-9]+), w12, #?6 |
| 107 |
** movaz {z28\.h - z31\.h}, za1v\.h\[\1, 0:3\] |
| 108 |
** ret |
| 109 |
*/ |
| 110 |
TEST_READ_ZA_XN (readz_za16_u16_z28_1_w12p6, svuint16x4_t, |
| 111 |
z28 = svreadz_ver_za16_u16_vg4 (1, w12 + 6), |
| 112 |
z28 = svreadz_ver_za16_u16_vg4 (1, w12 + 6)) |
| 113 |
|
| 114 |
/* |
| 115 |
** readz_za16_s16_z0_0_w15p8: |
| 116 |
** add (w[0-9]+), w15, #?8 |
| 117 |
** movaz {z0\.h - z3\.h}, za0v\.h\[\1, 0:3\] |
| 118 |
** ret |
| 119 |
*/ |
| 120 |
TEST_READ_ZA_XN (readz_za16_s16_z0_0_w15p8, svint16x4_t, |
| 121 |
z0 = svreadz_ver_za16_s16_vg4 (0, w15 + 8), |
| 122 |
z0 = svreadz_ver_za16_s16_vg4 (0, w15 + 8)) |
| 123 |
|
| 124 |
/* |
| 125 |
** readz_za16_bf16_z4_1_w12m1: |
| 126 |
** sub (w[0-9]+), w12, #?1 |
| 127 |
** movaz {z4\.h - z7\.h}, za1v\.h\[\1, 0:3\] |
| 128 |
** ret |
| 129 |
*/ |
| 130 |
TEST_READ_ZA_XN (readz_za16_bf16_z4_1_w12m1, svbfloat16x4_t, |
| 131 |
z4 = svreadz_ver_za16_bf16_vg4 (1, w12 - 1), |
| 132 |
z4 = svreadz_ver_za16_bf16_vg4 (1, w12 - 1)) |
| 133 |
|
| 134 |
/* |
| 135 |
** readz_za16_u16_z28_0_w16: |
| 136 |
** mov (w1[2-5]), w16 |
| 137 |
** movaz {z28\.h - z31\.h}, za0v\.h\[\1, 0:3\] |
| 138 |
** ret |
| 139 |
*/ |
| 140 |
TEST_READ_ZA_XN (readz_za16_u16_z28_0_w16, svuint16x4_t, |
| 141 |
z28 = svreadz_ver_za16_u16_vg4 (0, w16), |
| 142 |
z28 = svreadz_ver_za16_u16_vg4 (0, w16)) |