Debian Codegraph @

gcc/testsuite/gcc.target/aarch64/sme2/acle-asm/mla_za16_f16_vg1x4.c

raw
1 /* { dg-do assemble { target aarch64_asm_sme-f16f16_ok } } */
2 /* { dg-do compile { target { ! aarch64_asm_sme-f16f16_ok } } } */
3 /* { dg-final { check-function-bodies "**" "" "-DCHECK_ASM" } } */
4
5 #include "test_sme2_acle.h"
6
7 #pragma GCC target "+sme-f16f16"
8
9 /*
10 ** mla_0_z0_z0:
11 ** mov (w8|w9|w10|w11), #?0
12 ** fmla za\.h\[\1, 0, vgx4\], {z0\.h - z3\.h}, {z0\.h - z3\.h}
13 ** ret
14 */
15 TEST_ZA_XN (mla_0_z0_z0, svfloat16x4_t,
16 svmla_za16_f16_vg1x4 (0, z0, z0),
17 svmla_za16_vg1x4 (0, z0, z0))
18
19 /*
20 ** mla_w0_z0_z0:
21 ** mov (w8|w9|w10|w11), w0
22 ** fmla za\.h\[\1, 0, vgx4\], {z0\.h - z3\.h}, {z0\.h - z3\.h}
23 ** ret
24 */
25 TEST_ZA_XN (mla_w0_z0_z0, svfloat16x4_t,
26 svmla_za16_f16_vg1x4 (w0, z0, z0),
27 svmla_za16_vg1x4 (w0, z0, z0))
28
29 /*
30 ** mla_w8_z0_z4:
31 ** fmla za\.h\[w8, 0, vgx4\], {z0\.h - z3\.h}, {z4\.h - z7\.h}
32 ** ret
33 */
34 TEST_ZA_XN (mla_w8_z0_z4, svfloat16x4_t,
35 svmla_za16_f16_vg1x4 (w8, z0, z4),
36 svmla_za16_vg1x4 (w8, z0, z4))
37
38 /* Leave the assembler to check for correctness for misaligned registers. */
39
40 /*
41 ** mla_w8_z0_z18:
42 ** ...
43 ** fmla za\.h\[w8, 0, vgx4\], {z0\.h - z3\.h}, [^\n]+
44 ** ret
45 */
46 TEST_ZA_XN (mla_w8_z0_z18, svfloat16x4_t,
47 svmla_za16_f16_vg1x4 (w8, z0, z18),
48 svmla_za16_vg1x4 (w8, z0, z18))
49
50 /*
51 ** mla_w8_z18_z28:
52 ** ...
53 ** fmla za\.h\[w8, 0, vgx4\], [^\n]+, {z28\.h - z31\.h}
54 ** ret
55 */
56 TEST_ZA_XN (mla_w8_z18_z28, svfloat16x4_t,
57 svmla_za16_f16_vg1x4 (w8, z18, z28),
58 svmla_za16_vg1x4 (w8, z18, z28))
59
60 /*
61 ** mla_w8_z28_z23:
62 ** ...
63 ** fmla za\.h\[w8, 0, vgx4\], {z28\.h - z31\.h}, [^\n]+
64 ** ret
65 */
66 TEST_ZA_XN (mla_w8_z28_z23, svfloat16x4_t,
67 svmla_za16_f16_vg1x4 (w8, z28, z23),
68 svmla_za16_vg1x4 (w8, z28, z23))
69
70 /*
71 ** mla_w8p7_z4_z0:
72 ** fmla za\.h\[w8, 7, vgx4\], {z4\.h - z7\.h}, {z0\.h - z3\.h}
73 ** ret
74 */
75 TEST_ZA_XN (mla_w8p7_z4_z0, svfloat16x4_t,
76 svmla_za16_f16_vg1x4 (w8 + 7, z4, z0),
77 svmla_za16_vg1x4 (w8 + 7, z4, z0))
78
79 /*
80 ** mla_w8p8_z4_z4:
81 ** add (w8|w9|w10|w11), w8, #?8
82 ** fmla za\.h\[\1, 0, vgx4\], {z4\.h - z7\.h}, {z4\.h - z7\.h}
83 ** ret
84 */
85 TEST_ZA_XN (mla_w8p8_z4_z4, svfloat16x4_t,
86 svmla_za16_f16_vg1x4 (w8 + 8, z4, z4),
87 svmla_za16_vg1x4 (w8 + 8, z4, z4))
88
89 /*
90 ** mla_w8m1_z4_z0:
91 ** sub (w8|w9|w10|w11), w8, #?1
92 ** fmla za\.h\[\1, 0, vgx4\], {z4\.h - z7\.h}, {z0\.h - z3\.h}
93 ** ret
94 */
95 TEST_ZA_XN (mla_w8m1_z4_z0, svfloat16x4_t,
96 svmla_za16_f16_vg1x4 (w8 - 1, z4, z0),
97 svmla_za16_vg1x4 (w8 - 1, z4, z0))
98
99 /*
100 ** mla_single_0_z1_z0:
101 ** mov (w8|w9|w10|w11), #?0
102 ** fmla za\.h\[\1, 0, vgx4\], {z1\.h - z4\.h}, z0\.h
103 ** ret
104 */
105 TEST_ZA_SINGLE (mla_single_0_z1_z0, svfloat16x4_t, svfloat16_t,
106 svmla_single_za16_f16_vg1x4 (0, z1, z0),
107 svmla_za16_vg1x4 (0, z1, z0))
108
109 /*
110 ** mla_single_w0_z1_z0:
111 ** mov (w8|w9|w10|w11), w0
112 ** fmla za\.h\[\1, 0, vgx4\], {z1\.h - z4\.h}, z0\.h
113 ** ret
114 */
115 TEST_ZA_SINGLE (mla_single_w0_z1_z0, svfloat16x4_t, svfloat16_t,
116 svmla_single_za16_f16_vg1x4 (w0, z1, z0),
117 svmla_za16_vg1x4 (w0, z1, z0))
118
119 /*
120 ** mla_single_w8_z1_z0:
121 ** fmla za\.h\[w8, 0, vgx4\], {z1\.h - z4\.h}, z0\.h
122 ** ret
123 */
124 TEST_ZA_SINGLE (mla_single_w8_z1_z0, svfloat16x4_t, svfloat16_t,
125 svmla_single_za16_f16_vg1x4 (w8, z1, z0),
126 svmla_za16_vg1x4 (w8, z1, z0))
127
128 /*
129 ** mla_single_w8p7_z1_z0:
130 ** fmla za\.h\[w8, 7, vgx4\], {z1\.h - z4\.h}, z0\.h
131 ** ret
132 */
133 TEST_ZA_SINGLE (mla_single_w8p7_z1_z0, svfloat16x4_t, svfloat16_t,
134 svmla_single_za16_f16_vg1x4 (w8 + 7, z1, z0),
135 svmla_za16_vg1x4 (w8 + 7, z1, z0))
136
137 /*
138 ** mla_single_w8p8_z1_z0:
139 ** add (w8|w9|w10|w11), w8, #?8
140 ** fmla za\.h\[\1, 0, vgx4\], {z1\.h - z4\.h}, z0\.h
141 ** ret
142 */
143 TEST_ZA_SINGLE (mla_single_w8p8_z1_z0, svfloat16x4_t, svfloat16_t,
144 svmla_single_za16_f16_vg1x4 (w8 + 8, z1, z0),
145 svmla_za16_vg1x4 (w8 + 8, z1, z0))
146
147 /*
148 ** mla_single_w0m1_z1_z0:
149 ** sub (w8|w9|w10|w11), w0, #?1
150 ** fmla za\.h\[\1, 0, vgx4\], {z1\.h - z4\.h}, z0\.h
151 ** ret
152 */
153 TEST_ZA_SINGLE (mla_single_w0m1_z1_z0, svfloat16x4_t, svfloat16_t,
154 svmla_single_za16_f16_vg1x4 (w0 - 1, z1, z0),
155 svmla_za16_vg1x4 (w0 - 1, z1, z0))
156
157 /*
158 ** mla_single_w8_z0_z15:
159 ** str d15, \[sp, #?-16\]!
160 ** fmla za\.h\[w8, 0, vgx4\], {z0\.h - z3\.h}, z15\.h
161 ** ldr d15, \[sp\], #?16
162 ** ret
163 */
164 TEST_ZA_SINGLE_Z15 (mla_single_w8_z0_z15, svfloat16x4_t, svfloat16_t,
165 svmla_single_za16_f16_vg1x4 (w8, z0, z15),
166 svmla_za16_vg1x4 (w8, z0, z15))
167
168 /*
169 ** mla_single_w8_z20_z16:
170 ** mov (z[0-7]).d, z16.d
171 ** fmla za\.h\[w8, 0, vgx4\], {z20\.h - z23\.h}, \1\.h
172 ** ret
173 */
174 TEST_ZA_SINGLE (mla_single_w8_z20_z16, svfloat16x4_t, svfloat16_t,
175 svmla_single_za16_f16_vg1x4 (w8, z20, z16),
176 svmla_za16_vg1x4 (w8, z20, z16))