Debian Codegraph @

gcc/testsuite/gcc.target/aarch64/sme2/acle-asm/mla_lane_za16_bf16_vg1x2.c

raw
1 /* { dg-do assemble { target aarch64_asm_sme-b16b16_ok } } */
2 /* { dg-do compile { target { ! aarch64_asm_sme-b16b16_ok } } } */
3 /* { dg-final { check-function-bodies "**" "" "-DCHECK_ASM" } } */
4
5 #include "test_sme2_acle.h"
6
7 #pragma GCC target "+sme-b16b16"
8
9 /*
10 ** mla_lane_0_z0_z4_0:
11 ** mov (w8|w9|w10|w11), #?0
12 ** bfmla za\.h\[\1, 0, vgx2\], {z0\.h - z1\.h}, z4\.h\[0\]
13 ** ret
14 */
15 TEST_ZA_LANE (mla_lane_0_z0_z4_0, svbfloat16x2_t, svbfloat16_t,
16 svmla_lane_za16_bf16_vg1x2 (0, z0, z4, 0),
17 svmla_lane_za16_vg1x2 (0, z0, z4, 0))
18
19 /*
20 ** mla_lane_w0_z0_z7_1:
21 ** mov (w8|w9|w10|w11), w0
22 ** bfmla za\.h\[\1, 0, vgx2\], {z0\.h - z1\.h}, z7\.h\[1\]
23 ** ret
24 */
25 TEST_ZA_LANE (mla_lane_w0_z0_z7_1, svbfloat16x2_t, svbfloat16_t,
26 svmla_lane_za16_bf16_vg1x2 (w0, z0, z7, 1),
27 svmla_lane_za16_vg1x2 (w0, z0, z7, 1))
28
29 /*
30 ** mla_lane_w8_z28_z4_2:
31 ** bfmla za\.h\[w8, 0, vgx2\], {z28\.h - z29\.h}, z4\.h\[2\]
32 ** ret
33 */
34 TEST_ZA_LANE (mla_lane_w8_z28_z4_2, svbfloat16x2_t, svbfloat16_t,
35 svmla_lane_za16_bf16_vg1x2 (w8, z28, z4, 2),
36 svmla_lane_za16_vg1x2 (w8, z28, z4, 2))
37
38 /*
39 ** mla_lane_w8p7_z0_z4_3:
40 ** bfmla za\.h\[w8, 7, vgx2\], {z0\.h - z1\.h}, z4\.h\[3\]
41 ** ret
42 */
43 TEST_ZA_LANE (mla_lane_w8p7_z0_z4_3, svbfloat16x2_t, svbfloat16_t,
44 svmla_lane_za16_bf16_vg1x2 (w8 + 7, z0, z4, 3),
45 svmla_lane_za16_vg1x2 (w8 + 7, z0, z4, 3))
46
47 /*
48 ** mla_lane_w8p8_z0_z4_4:
49 ** add (w8|w9|w10|w11), w8, #?8
50 ** bfmla za\.h\[\1, 0, vgx2\], {z0\.h - z1\.h}, z4\.h\[4\]
51 ** ret
52 */
53 TEST_ZA_LANE (mla_lane_w8p8_z0_z4_4, svbfloat16x2_t, svbfloat16_t,
54 svmla_lane_za16_bf16_vg1x2 (w8 + 8, z0, z4, 4),
55 svmla_lane_za16_vg1x2 (w8 + 8, z0, z4, 4))
56
57 /*
58 ** mla_lane_w0m1_z0_z4_5:
59 ** sub (w8|w9|w10|w11), w0, #?1
60 ** bfmla za\.h\[\1, 0, vgx2\], {z0\.h - z1\.h}, z4\.h\[5\]
61 ** ret
62 */
63 TEST_ZA_LANE (mla_lane_w0m1_z0_z4_5, svbfloat16x2_t, svbfloat16_t,
64 svmla_lane_za16_bf16_vg1x2 (w0 - 1, z0, z4, 5),
65 svmla_lane_za16_vg1x2 (w0 - 1, z0, z4, 5))
66
67 /*
68 ** mla_lane_w8_z4_z15_6:
69 ** str d15, \[sp, #?-16\]!
70 ** bfmla za\.h\[w8, 0, vgx2\], {z4\.h - z5\.h}, z15\.h\[6\]
71 ** ldr d15, \[sp\], #?16
72 ** ret
73 */
74 TEST_ZA_LANE_Z15 (mla_lane_w8_z4_z15_6, svbfloat16x2_t, svbfloat16_t,
75 svmla_lane_za16_bf16_vg1x2 (w8, z4, z15, 6),
76 svmla_lane_za16_vg1x2 (w8, z4, z15, 6))
77
78 /*
79 ** mla_lane_w8_z28_z16_7:
80 ** mov (z[0-7]).d, z16.d
81 ** bfmla za\.h\[w8, 0, vgx2\], {z28\.h - z29\.h}, \1\.h\[7\]
82 ** ret
83 */
84 TEST_ZA_LANE (mla_lane_w8_z28_z16_7, svbfloat16x2_t, svbfloat16_t,
85 svmla_lane_za16_bf16_vg1x2 (w8, z28, z16, 7),
86 svmla_lane_za16_vg1x2 (w8, z28, z16, 7))
87
88 /*
89 ** mla_lane_w8_z17_z7_0:
90 ** mov [^\n]+
91 ** mov [^\n]+
92 ** bfmla za\.h\[w8, 0, vgx2\], [^\n]+, z7\.h\[0\]
93 ** ret
94 */
95 TEST_ZA_LANE (mla_lane_w8_z17_z7_0, svbfloat16x2_t, svbfloat16_t,
96 svmla_lane_za16_bf16_vg1x2 (w8, z17, z7, 0),
97 svmla_lane_za16_vg1x2 (w8, z17, z7, 0))
98
99 /*
100 ** mla_lane_w8_z22_z4_1:
101 ** bfmla za\.h\[w8, 0, vgx2\], {z22\.h - z23\.h}, z4\.h\[1\]
102 ** ret
103 */
104 TEST_ZA_LANE (mla_lane_w8_z22_z4_1, svbfloat16x2_t, svbfloat16_t,
105 svmla_lane_za16_bf16_vg1x2 (w8, z22, z4, 1),
106 svmla_lane_za16_vg1x2 (w8, z22, z4, 1))