Debian Codegraph @

gcc/testsuite/gcc.target/aarch64/sme2/acle-asm/minnm_bf16_x4.c

raw
1 /* { dg-do assemble { target aarch64_asm_sve-b16b16_ok } } */
2 /* { dg-do compile { target { ! aarch64_asm_sve-b16b16_ok } } } */
3 /* { dg-final { check-function-bodies "**" "" "-DCHECK_ASM" } } */
4
5 #include "test_sme2_acle.h"
6
7 #pragma GCC target "+sve-b16b16"
8
9 /*
10 ** minnm_z0_z0_z4:
11 ** bfminnm {z0\.h - z3\.h}, {z0\.h - z3\.h}, {z4\.h - z7\.h}
12 ** ret
13 */
14 TEST_XN (minnm_z0_z0_z4, svbfloat16x4_t, z0,
15 svminnm_bf16_x4 (z0, z4),
16 svminnm (z0, z4))
17
18 /*
19 ** minnm_z0_z4_z0:
20 ** bfminnm {z0\.h - z3\.h}, {z0\.h - z3\.h}, {z4\.h - z7\.h}
21 ** ret
22 */
23 TEST_XN (minnm_z0_z4_z0, svbfloat16x4_t, z0,
24 svminnm_bf16_x4 (z4, z0),
25 svminnm (z4, z0))
26
27 /*
28 ** minnm_z0_z4_z28:
29 ** (
30 ** mov [^\n]+
31 ** mov [^\n]+
32 ** mov [^\n]+
33 ** mov [^\n]+
34 ** bfminnm [^\n]+, {z28\.h - z31\.h}
35 ** |
36 ** bfminnm [^\n]+, {z28\.h - z31\.h}
37 ** mov [^\n]+
38 ** mov [^\n]+
39 ** mov [^\n]+
40 ** mov [^\n]+
41 ** )
42 ** ret
43 */
44 TEST_XN (minnm_z0_z4_z28, svbfloat16x4_t, z0,
45 svminnm_bf16_x4 (z4, z28),
46 svminnm (z4, z28))
47
48 /*
49 ** minnm_z18_z18_z4:
50 ** mov [^\n]+
51 ** mov [^\n]+
52 ** mov [^\n]+
53 ** mov [^\n]+
54 ** bfminnm [^\n]+, {z4\.h - z7\.h}
55 ** mov [^\n]+
56 ** mov [^\n]+
57 ** mov [^\n]+
58 ** mov [^\n]+
59 ** ret
60 */
61 TEST_XN (minnm_z18_z18_z4, svbfloat16x4_t, z18,
62 svminnm_bf16_x4 (z18, z4),
63 svminnm (z18, z4))
64
65 /*
66 ** minnm_z23_z23_z28:
67 ** mov [^\n]+
68 ** mov [^\n]+
69 ** mov [^\n]+
70 ** mov [^\n]+
71 ** bfminnm [^\n]+, {z28\.h - z31\.h}
72 ** mov [^\n]+
73 ** mov [^\n]+
74 ** mov [^\n]+
75 ** mov [^\n]+
76 ** ret
77 */
78 TEST_XN (minnm_z23_z23_z28, svbfloat16x4_t, z23,
79 svminnm_bf16_x4 (z23, z28),
80 svminnm (z23, z28))
81
82 /*
83 ** minnm_z28_z28_z0:
84 ** bfminnm {z28\.h - z31\.h}, {z28\.h - z31\.h}, {z0\.h - z3\.h}
85 ** ret
86 */
87 TEST_XN (minnm_z28_z28_z0, svbfloat16x4_t, z28,
88 svminnm_bf16_x4 (z28, z0),
89 svminnm (z28, z0))
90
91 /*
92 ** minnm_z0_z0_z18:
93 ** (
94 ** mov [^\n]+
95 ** mov [^\n]+
96 ** mov [^\n]+
97 ** mov [^\n]+
98 ** bfminnm {z0\.h - z3\.h}, {z0\.h - z3\.h}, [^\n]+
99 ** |
100 ** bfminnm {z0\.h - z3\.h}, {z0\.h - z3\.h}, [^\n]+
101 ** mov [^\n]+
102 ** mov [^\n]+
103 ** mov [^\n]+
104 ** mov [^\n]+
105 ** )
106 ** ret
107 */
108 TEST_XN (minnm_z0_z0_z18, svbfloat16x4_t, z0,
109 svminnm_bf16_x4 (z0, z18),
110 svminnm (z0, z18))
111
112 /*
113 ** minnm_z4_z4_z23:
114 ** (
115 ** mov [^\n]+
116 ** mov [^\n]+
117 ** mov [^\n]+
118 ** mov [^\n]+
119 ** bfminnm {z4\.h - z7\.h}, {z4\.h - z7\.h}, [^\n]+
120 ** |
121 ** bfminnm {z4\.h - z7\.h}, {z4\.h - z7\.h}, [^\n]+
122 ** mov [^\n]+
123 ** mov [^\n]+
124 ** mov [^\n]+
125 ** mov [^\n]+
126 ** )
127 ** ret
128 */
129 TEST_XN (minnm_z4_z4_z23, svbfloat16x4_t, z4,
130 svminnm_bf16_x4 (z4, z23),
131 svminnm (z4, z23))
132
133 /*
134 ** minnm_single_z24_z24_z0:
135 ** bfminnm {z24\.h - z27\.h}, {z24\.h - z27\.h}, z0\.h
136 ** ret
137 */
138 TEST_XN_SINGLE (minnm_single_z24_z24_z0, svbfloat16x4_t, svbfloat16_t, z24,
139 svminnm_single_bf16_x4 (z24, z0),
140 svminnm (z24, z0))
141
142 /*
143 ** minnm_single_z24_z28_z0:
144 ** (
145 ** mov [^\n]+
146 ** mov [^\n]+
147 ** mov [^\n]+
148 ** mov [^\n]+
149 ** bfminnm {z24\.h - z27\.h}, {z24\.h - z27\.h}, z0\.h
150 ** |
151 ** bfminnm {z28\.h - z31\.h}, {z28\.h - z31\.h}, z0\.h
152 ** mov [^\n]+
153 ** mov [^\n]+
154 ** mov [^\n]+
155 ** mov [^\n]+
156 ** )
157 ** ret
158 */
159 TEST_XN_SINGLE (minnm_single_z24_z28_z0, svbfloat16x4_t, svbfloat16_t, z24,
160 svminnm_single_bf16_x4 (z28, z0),
161 svminnm (z28, z0))
162
163 /*
164 ** minnm_single_z24_z1_z0:
165 ** mov [^\n]+
166 ** mov [^\n]+
167 ** mov [^\n]+
168 ** mov [^\n]+
169 ** bfminnm {z24\.h - z27\.h}, {z24\.h - z27\.h}, z0\.h
170 ** ret
171 */
172 TEST_XN_SINGLE (minnm_single_z24_z1_z0, svbfloat16x4_t, svbfloat16_t, z24,
173 svminnm_single_bf16_x4 (z1, z0),
174 svminnm (z1, z0))
175
176 /*
177 ** minnm_single_z1_z24_z0:
178 ** bfminnm {z24\.h - z27\.h}, {z24\.h - z27\.h}, z0\.h
179 ** mov [^\n]+
180 ** mov [^\n]+
181 ** mov [^\n]+
182 ** mov [^\n]+
183 ** ret
184 */
185 TEST_XN_SINGLE (minnm_single_z1_z24_z0, svbfloat16x4_t, svbfloat16_t, z1,
186 svminnm_single_bf16_x4 (z24, z0),
187 svminnm (z24, z0))
188
189 /*
190 ** minnm_single_z1_z1_z0:
191 ** mov [^\n]+
192 ** mov [^\n]+
193 ** mov [^\n]+
194 ** mov [^\n]+
195 ** bfminnm ({z[0-9]+\.h - z[0-9]+\.h}), \1, z0\.h
196 ** mov [^\n]+
197 ** mov [^\n]+
198 ** mov [^\n]+
199 ** mov [^\n]+
200 ** ret
201 */
202 TEST_XN_SINGLE (minnm_single_z1_z1_z0, svbfloat16x4_t, svbfloat16_t, z1,
203 svminnm_single_bf16_x4 (z1, z0),
204 svminnm (z1, z0))
205
206 /*
207 ** minnm_single_z18_z18_z0:
208 ** mov [^\n]+
209 ** mov [^\n]+
210 ** mov [^\n]+
211 ** mov [^\n]+
212 ** bfminnm [^\n]+, z0\.h
213 ** mov [^\n]+
214 ** mov [^\n]+
215 ** mov [^\n]+
216 ** mov [^\n]+
217 ** ret
218 */
219 TEST_XN_SINGLE (minnm_single_z18_z18_z0, svbfloat16x4_t, svbfloat16_t, z18,
220 svminnm_single_bf16_x4 (z18, z0),
221 svminnm (z18, z0))
222
223 /*
224 ** minnm_single_awkward:
225 ** ...
226 ** bfminnm ({z[0-9]+\.h - z[0-9]+\.h}), \1, z[0-9]+\.h
227 ** ...
228 ** ret
229 */
230 TEST_XN_SINGLE_AWKWARD (minnm_single_awkward, svbfloat16x4_t, svbfloat16_t,
231 z0_res = svminnm_single_bf16_x4 (z1, z0),
232 z0_res = svminnm (z1, z0))
233
234 /*
235 ** minnm_single_z0_z0_z15:
236 ** ...
237 ** bfminnm {z0\.h - z3\.h}, {z0\.h - z3\.h}, z15\.h
238 ** ...
239 ** ret
240 */
241 TEST_XN_SINGLE_Z15 (minnm_single_z0_z0_z15, svbfloat16x4_t, svbfloat16_t,
242 z0 = svminnm_single_bf16_x4 (z0, z15),
243 z0 = svminnm (z0, z15))
244
245 /*
246 ** minnm_single_z24_z24_z16:
247 ** mov (z[0-7])\.d, z16\.d
248 ** bfminnm {z24\.h - z27\.h}, {z24\.h - z27\.h}, \1\.h
249 ** ret
250 */
251 TEST_XN_SINGLE (minnm_single_z24_z24_z16, svbfloat16x4_t, svbfloat16_t, z24,
252 svminnm_single_bf16_x4 (z24, z16),
253 svminnm (z24, z16))