1 /* { dg-final { check-function-bodies "**" "" "-DCHECK_ASM" } } */
2
3 #include "test_sve_acle.h"
4
5 /*
6 ** mul_f16_m_tied1:
7 ** fmul z0\.h, p0/m, z0\.h, z1\.h
8 ** ret
9 */
10 TEST_UNIFORM_Z (mul_f16_m_tied1, svfloat16_t,
11 z0 = svmul_f16_m (p0, z0, z1),
12 z0 = svmul_m (p0, z0, z1))
13
14 /*
15 ** mul_f16_m_tied2:
16 ** mov (z[0-9]+)\.d, z0\.d
17 ** movprfx z0, z1
18 ** fmul z0\.h, p0/m, z0\.h, \1\.h
19 ** ret
20 */
21 TEST_UNIFORM_Z (mul_f16_m_tied2, svfloat16_t,
22 z0 = svmul_f16_m (p0, z1, z0),
23 z0 = svmul_m (p0, z1, z0))
24
25 /*
26 ** mul_f16_m_untied:
27 ** movprfx z0, z1
28 ** fmul z0\.h, p0/m, z0\.h, z2\.h
29 ** ret
30 */
31 TEST_UNIFORM_Z (mul_f16_m_untied, svfloat16_t,
32 z0 = svmul_f16_m (p0, z1, z2),
33 z0 = svmul_m (p0, z1, z2))
34
35 /*
36 ** mul_h4_f16_m_tied1:
37 ** mov (z[0-9]+\.h), h4
38 ** fmul z0\.h, p0/m, z0\.h, \1
39 ** ret
40 */
41 TEST_UNIFORM_ZD (mul_h4_f16_m_tied1, svfloat16_t, __fp16,
42 z0 = svmul_n_f16_m (p0, z0, d4),
43 z0 = svmul_m (p0, z0, d4))
44
45 /*
46 ** mul_h4_f16_m_untied:
47 ** mov (z[0-9]+\.h), h4
48 ** movprfx z0, z1
49 ** fmul z0\.h, p0/m, z0\.h, \1
50 ** ret
51 */
52 TEST_UNIFORM_ZD (mul_h4_f16_m_untied, svfloat16_t, __fp16,
53 z0 = svmul_n_f16_m (p0, z1, d4),
54 z0 = svmul_m (p0, z1, d4))
55
56 /*
57 ** mul_1_f16_m_tied1:
58 ** fmov (z[0-9]+\.h), #1\.0(?:e\+0)?
59 ** fmul z0\.h, p0/m, z0\.h, \1
60 ** ret
61 */
62 TEST_UNIFORM_Z (mul_1_f16_m_tied1, svfloat16_t,
63 z0 = svmul_n_f16_m (p0, z0, 1),
64 z0 = svmul_m (p0, z0, 1))
65
66 /*
67 ** mul_1_f16_m_untied: { xfail *-*-* }
68 ** fmov (z[0-9]+\.h), #1\.0(?:e\+0)?
69 ** movprfx z0, z1
70 ** fmul z0\.h, p0/m, z0\.h, \1
71 ** ret
72 */
73 TEST_UNIFORM_Z (mul_1_f16_m_untied, svfloat16_t,
74 z0 = svmul_n_f16_m (p0, z1, 1),
75 z0 = svmul_m (p0, z1, 1))
76
77 /*
78 ** mul_0p5_f16_m_tied1:
79 ** fmul z0\.h, p0/m, z0\.h, #0\.5
80 ** ret
81 */
82 TEST_UNIFORM_Z (mul_0p5_f16_m_tied1, svfloat16_t,
83 z0 = svmul_n_f16_m (p0, z0, 0.5),
84 z0 = svmul_m (p0, z0, 0.5))
85
86 /*
87 ** mul_0p5_f16_m_untied:
88 ** movprfx z0, z1
89 ** fmul z0\.h, p0/m, z0\.h, #0\.5
90 ** ret
91 */
92 TEST_UNIFORM_Z (mul_0p5_f16_m_untied, svfloat16_t,
93 z0 = svmul_n_f16_m (p0, z1, 0.5),
94 z0 = svmul_m (p0, z1, 0.5))
95
96 /*
97 ** mul_2_f16_m_tied1:
98 ** fmul z0\.h, p0/m, z0\.h, #2\.0
99 ** ret
100 */
101 TEST_UNIFORM_Z (mul_2_f16_m_tied1, svfloat16_t,
102 z0 = svmul_n_f16_m (p0, z0, 2),
103 z0 = svmul_m (p0, z0, 2))
104
105 /*
106 ** mul_2_f16_m_untied:
107 ** movprfx z0, z1
108 ** fmul z0\.h, p0/m, z0\.h, #2\.0
109 ** ret
110 */
111 TEST_UNIFORM_Z (mul_2_f16_m_untied, svfloat16_t,
112 z0 = svmul_n_f16_m (p0, z1, 2),
113 z0 = svmul_m (p0, z1, 2))
114
115 /*
116 ** mul_f16_z_tied1:
117 ** movprfx z0\.h, p0/z, z0\.h
118 ** fmul z0\.h, p0/m, z0\.h, z1\.h
119 ** ret
120 */
121 TEST_UNIFORM_Z (mul_f16_z_tied1, svfloat16_t,
122 z0 = svmul_f16_z (p0, z0, z1),
123 z0 = svmul_z (p0, z0, z1))
124
125 /*
126 ** mul_f16_z_tied2:
127 ** movprfx z0\.h, p0/z, z0\.h
128 ** fmul z0\.h, p0/m, z0\.h, z1\.h
129 ** ret
130 */
131 TEST_UNIFORM_Z (mul_f16_z_tied2, svfloat16_t,
132 z0 = svmul_f16_z (p0, z1, z0),
133 z0 = svmul_z (p0, z1, z0))
134
135 /*
136 ** mul_f16_z_untied:
137 ** (
138 ** movprfx z0\.h, p0/z, z1\.h
139 ** fmul z0\.h, p0/m, z0\.h, z2\.h
140 ** |
141 ** movprfx z0\.h, p0/z, z2\.h
142 ** fmul z0\.h, p0/m, z0\.h, z1\.h
143 ** )
144 ** ret
145 */
146 TEST_UNIFORM_Z (mul_f16_z_untied, svfloat16_t,
147 z0 = svmul_f16_z (p0, z1, z2),
148 z0 = svmul_z (p0, z1, z2))
149
150 /*
151 ** mul_h4_f16_z_tied1:
152 ** mov (z[0-9]+\.h), h4
153 ** movprfx z0\.h, p0/z, z0\.h
154 ** fmul z0\.h, p0/m, z0\.h, \1
155 ** ret
156 */
157 TEST_UNIFORM_ZD (mul_h4_f16_z_tied1, svfloat16_t, __fp16,
158 z0 = svmul_n_f16_z (p0, z0, d4),
159 z0 = svmul_z (p0, z0, d4))
160
161 /*
162 ** mul_h4_f16_z_untied:
163 ** mov (z[0-9]+\.h), h4
164 ** (
165 ** movprfx z0\.h, p0/z, z1\.h
166 ** fmul z0\.h, p0/m, z0\.h, \1
167 ** |
168 ** movprfx z0\.h, p0/z, \1
169 ** fmul z0\.h, p0/m, z0\.h, z1\.h
170 ** )
171 ** ret
172 */
173 TEST_UNIFORM_ZD (mul_h4_f16_z_untied, svfloat16_t, __fp16,
174 z0 = svmul_n_f16_z (p0, z1, d4),
175 z0 = svmul_z (p0, z1, d4))
176
177 /*
178 ** mul_1_f16_z_tied1:
179 ** fmov (z[0-9]+\.h), #1\.0(?:e\+0)?
180 ** movprfx z0\.h, p0/z, z0\.h
181 ** fmul z0\.h, p0/m, z0\.h, \1
182 ** ret
183 */
184 TEST_UNIFORM_Z (mul_1_f16_z_tied1, svfloat16_t,
185 z0 = svmul_n_f16_z (p0, z0, 1),
186 z0 = svmul_z (p0, z0, 1))
187
188 /*
189 ** mul_1_f16_z_untied:
190 ** fmov (z[0-9]+\.h), #1\.0(?:e\+0)?
191 ** (
192 ** movprfx z0\.h, p0/z, z1\.h
193 ** fmul z0\.h, p0/m, z0\.h, \1
194 ** |
195 ** movprfx z0\.h, p0/z, \1
196 ** fmul z0\.h, p0/m, z0\.h, z1\.h
197 ** )
198 ** ret
199 */
200 TEST_UNIFORM_Z (mul_1_f16_z_untied, svfloat16_t,
201 z0 = svmul_n_f16_z (p0, z1, 1),
202 z0 = svmul_z (p0, z1, 1))
203
204 /*
205 ** mul_0p5_f16_z_tied1:
206 ** movprfx z0\.h, p0/z, z0\.h
207 ** fmul z0\.h, p0/m, z0\.h, #0\.5
208 ** ret
209 */
210 TEST_UNIFORM_Z (mul_0p5_f16_z_tied1, svfloat16_t,
211 z0 = svmul_n_f16_z (p0, z0, 0.5),
212 z0 = svmul_z (p0, z0, 0.5))
213
214 /*
215 ** mul_0p5_f16_z_untied:
216 ** movprfx z0\.h, p0/z, z1\.h
217 ** fmul z0\.h, p0/m, z0\.h, #0\.5
218 ** ret
219 */
220 TEST_UNIFORM_Z (mul_0p5_f16_z_untied, svfloat16_t,
221 z0 = svmul_n_f16_z (p0, z1, 0.5),
222 z0 = svmul_z (p0, z1, 0.5))
223
224 /*
225 ** mul_2_f16_z_tied1:
226 ** movprfx z0\.h, p0/z, z0\.h
227 ** fmul z0\.h, p0/m, z0\.h, #2\.0
228 ** ret
229 */
230 TEST_UNIFORM_Z (mul_2_f16_z_tied1, svfloat16_t,
231 z0 = svmul_n_f16_z (p0, z0, 2),
232 z0 = svmul_z (p0, z0, 2))
233
234 /*
235 ** mul_2_f16_z_untied:
236 ** movprfx z0\.h, p0/z, z1\.h
237 ** fmul z0\.h, p0/m, z0\.h, #2\.0
238 ** ret
239 */
240 TEST_UNIFORM_Z (mul_2_f16_z_untied, svfloat16_t,
241 z0 = svmul_n_f16_z (p0, z1, 2),
242 z0 = svmul_z (p0, z1, 2))
243
244 /*
245 ** mul_f16_x_tied1:
246 ** fmul z0\.h, p0/m, z0\.h, z1\.h
247 ** ret
248 */
249 TEST_UNIFORM_Z (mul_f16_x_tied1, svfloat16_t,
250 z0 = svmul_f16_x (p0, z0, z1),
251 z0 = svmul_x (p0, z0, z1))
252
253 /*
254 ** mul_f16_x_tied2:
255 ** fmul z0\.h, p0/m, z0\.h, z1\.h
256 ** ret
257 */
258 TEST_UNIFORM_Z (mul_f16_x_tied2, svfloat16_t,
259 z0 = svmul_f16_x (p0, z1, z0),
260 z0 = svmul_x (p0, z1, z0))
261
262 /*
263 ** mul_f16_x_untied:
264 ** (
265 ** movprfx z0, z1
266 ** fmul z0\.h, p0/m, z0\.h, z2\.h
267 ** |
268 ** movprfx z0, z2
269 ** fmul z0\.h, p0/m, z0\.h, z1\.h
270 ** )
271 ** ret
272 */
273 TEST_UNIFORM_Z (mul_f16_x_untied, svfloat16_t,
274 z0 = svmul_f16_x (p0, z1, z2),
275 z0 = svmul_x (p0, z1, z2))
276
277 /*
278 ** mul_h4_f16_x_tied1:
279 ** mov (z[0-9]+\.h), h4
280 ** fmul z0\.h, p0/m, z0\.h, \1
281 ** ret
282 */
283 TEST_UNIFORM_ZD (mul_h4_f16_x_tied1, svfloat16_t, __fp16,
284 z0 = svmul_n_f16_x (p0, z0, d4),
285 z0 = svmul_x (p0, z0, d4))
286
287 /*
288 ** mul_h4_f16_x_untied:
289 ** mov z0\.h, h4
290 ** fmul z0\.h, p0/m, z0\.h, z1\.h
291 ** ret
292 */
293 TEST_UNIFORM_ZD (mul_h4_f16_x_untied, svfloat16_t, __fp16,
294 z0 = svmul_n_f16_x (p0, z1, d4),
295 z0 = svmul_x (p0, z1, d4))
296
297 /*
298 ** mul_1_f16_x_tied1:
299 ** fmov (z[0-9]+\.h), #1\.0(?:e\+0)?
300 ** fmul z0\.h, p0/m, z0\.h, \1
301 ** ret
302 */
303 TEST_UNIFORM_Z (mul_1_f16_x_tied1, svfloat16_t,
304 z0 = svmul_n_f16_x (p0, z0, 1),
305 z0 = svmul_x (p0, z0, 1))
306
307 /*
308 ** mul_1_f16_x_untied:
309 ** fmov z0\.h, #1\.0(?:e\+0)?
310 ** fmul z0\.h, p0/m, z0\.h, z1\.h
311 ** ret
312 */
313 TEST_UNIFORM_Z (mul_1_f16_x_untied, svfloat16_t,
314 z0 = svmul_n_f16_x (p0, z1, 1),
315 z0 = svmul_x (p0, z1, 1))
316
317 /*
318 ** mul_0p5_f16_x_tied1:
319 ** fmul z0\.h, p0/m, z0\.h, #0\.5
320 ** ret
321 */
322 TEST_UNIFORM_Z (mul_0p5_f16_x_tied1, svfloat16_t,
323 z0 = svmul_n_f16_x (p0, z0, 0.5),
324 z0 = svmul_x (p0, z0, 0.5))
325
326 /*
327 ** mul_0p5_f16_x_untied:
328 ** movprfx z0, z1
329 ** fmul z0\.h, p0/m, z0\.h, #0\.5
330 ** ret
331 */
332 TEST_UNIFORM_Z (mul_0p5_f16_x_untied, svfloat16_t,
333 z0 = svmul_n_f16_x (p0, z1, 0.5),
334 z0 = svmul_x (p0, z1, 0.5))
335
336 /*
337 ** mul_2_f16_x_tied1:
338 ** fmul z0\.h, p0/m, z0\.h, #2\.0
339 ** ret
340 */
341 TEST_UNIFORM_Z (mul_2_f16_x_tied1, svfloat16_t,
342 z0 = svmul_n_f16_x (p0, z0, 2),
343 z0 = svmul_x (p0, z0, 2))
344
345 /*
346 ** mul_2_f16_x_untied:
347 ** movprfx z0, z1
348 ** fmul z0\.h, p0/m, z0\.h, #2\.0
349 ** ret
350 */
351 TEST_UNIFORM_Z (mul_2_f16_x_untied, svfloat16_t,
352 z0 = svmul_n_f16_x (p0, z1, 2),
353 z0 = svmul_x (p0, z1, 2))
354
355 /*
356 ** ptrue_mul_f16_x_tied1:
357 ** fmul z0\.h, (z0\.h, z1\.h|z1\.h, z0\.h)
358 ** ret
359 */
360 TEST_UNIFORM_Z (ptrue_mul_f16_x_tied1, svfloat16_t,
361 z0 = svmul_f16_x (svptrue_b16 (), z0, z1),
362 z0 = svmul_x (svptrue_b16 (), z0, z1))
363
364 /*
365 ** ptrue_mul_f16_x_tied2:
366 ** fmul z0\.h, (z0\.h, z1\.h|z1\.h, z0\.h)
367 ** ret
368 */
369 TEST_UNIFORM_Z (ptrue_mul_f16_x_tied2, svfloat16_t,
370 z0 = svmul_f16_x (svptrue_b16 (), z1, z0),
371 z0 = svmul_x (svptrue_b16 (), z1, z0))
372
373 /*
374 ** ptrue_mul_f16_x_untied:
375 ** fmul z0\.h, (z1\.h, z2\.h|z2\.h, z1\.h)
376 ** ret
377 */
378 TEST_UNIFORM_Z (ptrue_mul_f16_x_untied, svfloat16_t,
379 z0 = svmul_f16_x (svptrue_b16 (), z1, z2),
380 z0 = svmul_x (svptrue_b16 (), z1, z2))
381
382 /*
383 ** ptrue_mul_1_f16_x_tied1:
384 ** fmov (z[0-9]+\.h), #1\.0(?:e\+0)?
385 ** fmul z0\.h, (z0\.h, \1|\1, z0\.h)
386 ** ret
387 */
388 TEST_UNIFORM_Z (ptrue_mul_1_f16_x_tied1, svfloat16_t,
389 z0 = svmul_n_f16_x (svptrue_b16 (), z0, 1),
390 z0 = svmul_x (svptrue_b16 (), z0, 1))
391
392 /*
393 ** ptrue_mul_1_f16_x_untied:
394 ** fmov (z[0-9]+\.h), #1\.0(?:e\+0)?
395 ** fmul z0\.h, (z1\.h, \1|\1, z1\.h)
396 ** ret
397 */
398 TEST_UNIFORM_Z (ptrue_mul_1_f16_x_untied, svfloat16_t,
399 z0 = svmul_n_f16_x (svptrue_b16 (), z1, 1),
400 z0 = svmul_x (svptrue_b16 (), z1, 1))
401
402 /*
403 ** ptrue_mul_0p5_f16_x_tied1:
404 ** ...
405 ** ptrue p[0-9]+\.b[^\n]*
406 ** ...
407 ** ret
408 */
409 TEST_UNIFORM_Z (ptrue_mul_0p5_f16_x_tied1, svfloat16_t,
410 z0 = svmul_n_f16_x (svptrue_b16 (), z0, 0.5),
411 z0 = svmul_x (svptrue_b16 (), z0, 0.5))
412
413 /*
414 ** ptrue_mul_0p5_f16_x_untied:
415 ** ...
416 ** ptrue p[0-9]+\.b[^\n]*
417 ** ...
418 ** ret
419 */
420 TEST_UNIFORM_Z (ptrue_mul_0p5_f16_x_untied, svfloat16_t,
421 z0 = svmul_n_f16_x (svptrue_b16 (), z1, 0.5),
422 z0 = svmul_x (svptrue_b16 (), z1, 0.5))
423
424 /*
425 ** ptrue_mul_2_f16_x_tied1:
426 ** ...
427 ** ptrue p[0-9]+\.b[^\n]*
428 ** ...
429 ** ret
430 */
431 TEST_UNIFORM_Z (ptrue_mul_2_f16_x_tied1, svfloat16_t,
432 z0 = svmul_n_f16_x (svptrue_b16 (), z0, 2),
433 z0 = svmul_x (svptrue_b16 (), z0, 2))
434
435 /*
436 ** ptrue_mul_2_f16_x_untied:
437 ** ...
438 ** ptrue p[0-9]+\.b[^\n]*
439 ** ...
440 ** ret
441 */
442 TEST_UNIFORM_Z (ptrue_mul_2_f16_x_untied, svfloat16_t,
443 z0 = svmul_n_f16_x (svptrue_b16 (), z1, 2),
444 z0 = svmul_x (svptrue_b16 (), z1, 2))