1 /* { dg-final { check-function-bodies "**" "" "-DCHECK_ASM" } } */
2
3 #include "test_sve_acle.h"
4
5 /*
6 ** cvt_f32_f16_m_tied1:
7 ** fcvt z0\.s, p0/m, z4\.h
8 ** ret
9 */
10 TEST_DUAL_Z (cvt_f32_f16_m_tied1, svfloat32_t, svfloat16_t,
11 z0 = svcvt_f32_f16_m (z0, p0, z4),
12 z0 = svcvt_f32_m (z0, p0, z4))
13
14 /*
15 ** cvt_f32_f16_m_tied2:
16 ** mov (z[0-9]+)\.d, z0\.d
17 ** movprfx z0, z4
18 ** fcvt z0\.s, p0/m, \1\.h
19 ** ret
20 */
21 TEST_DUAL_Z_REV (cvt_f32_f16_m_tied2, svfloat32_t, svfloat16_t,
22 z0_res = svcvt_f32_f16_m (z4, p0, z0),
23 z0_res = svcvt_f32_m (z4, p0, z0))
24
25 /*
26 ** cvt_f32_f16_m_untied:
27 ** movprfx z0, z1
28 ** fcvt z0\.s, p0/m, z4\.h
29 ** ret
30 */
31 TEST_DUAL_Z (cvt_f32_f16_m_untied, svfloat32_t, svfloat16_t,
32 z0 = svcvt_f32_f16_m (z1, p0, z4),
33 z0 = svcvt_f32_m (z1, p0, z4))
34
35 /*
36 ** cvt_f32_f64_m_tied1:
37 ** fcvt z0\.s, p0/m, z4\.d
38 ** ret
39 */
40 TEST_DUAL_Z (cvt_f32_f64_m_tied1, svfloat32_t, svfloat64_t,
41 z0 = svcvt_f32_f64_m (z0, p0, z4),
42 z0 = svcvt_f32_m (z0, p0, z4))
43
44 /*
45 ** cvt_f32_f64_m_tied2:
46 ** mov (z[0-9]+\.d), z0\.d
47 ** movprfx z0, z4
48 ** fcvt z0\.s, p0/m, \1
49 ** ret
50 */
51 TEST_DUAL_Z_REV (cvt_f32_f64_m_tied2, svfloat32_t, svfloat64_t,
52 z0_res = svcvt_f32_f64_m (z4, p0, z0),
53 z0_res = svcvt_f32_m (z4, p0, z0))
54
55 /*
56 ** cvt_f32_f64_m_untied:
57 ** movprfx z0, z1
58 ** fcvt z0\.s, p0/m, z4\.d
59 ** ret
60 */
61 TEST_DUAL_Z (cvt_f32_f64_m_untied, svfloat32_t, svfloat64_t,
62 z0 = svcvt_f32_f64_m (z1, p0, z4),
63 z0 = svcvt_f32_m (z1, p0, z4))
64
65 /*
66 ** cvt_f32_s32_m_tied1:
67 ** scvtf z0\.s, p0/m, z4\.s
68 ** ret
69 */
70 TEST_DUAL_Z (cvt_f32_s32_m_tied1, svfloat32_t, svint32_t,
71 z0 = svcvt_f32_s32_m (z0, p0, z4),
72 z0 = svcvt_f32_m (z0, p0, z4))
73
74 /*
75 ** cvt_f32_s32_m_tied2:
76 ** mov (z[0-9]+)\.d, z0\.d
77 ** movprfx z0, z4
78 ** scvtf z0\.s, p0/m, \1\.s
79 ** ret
80 */
81 TEST_DUAL_Z_REV (cvt_f32_s32_m_tied2, svfloat32_t, svint32_t,
82 z0_res = svcvt_f32_s32_m (z4, p0, z0),
83 z0_res = svcvt_f32_m (z4, p0, z0))
84
85 /*
86 ** cvt_f32_s32_m_untied:
87 ** movprfx z0, z1
88 ** scvtf z0\.s, p0/m, z4\.s
89 ** ret
90 */
91 TEST_DUAL_Z (cvt_f32_s32_m_untied, svfloat32_t, svint32_t,
92 z0 = svcvt_f32_s32_m (z1, p0, z4),
93 z0 = svcvt_f32_m (z1, p0, z4))
94
95 /*
96 ** cvt_f32_s64_m_tied1:
97 ** scvtf z0\.s, p0/m, z4\.d
98 ** ret
99 */
100 TEST_DUAL_Z (cvt_f32_s64_m_tied1, svfloat32_t, svint64_t,
101 z0 = svcvt_f32_s64_m (z0, p0, z4),
102 z0 = svcvt_f32_m (z0, p0, z4))
103
104 /*
105 ** cvt_f32_s64_m_tied2:
106 ** mov (z[0-9]+\.d), z0\.d
107 ** movprfx z0, z4
108 ** scvtf z0\.s, p0/m, \1
109 ** ret
110 */
111 TEST_DUAL_Z_REV (cvt_f32_s64_m_tied2, svfloat32_t, svint64_t,
112 z0_res = svcvt_f32_s64_m (z4, p0, z0),
113 z0_res = svcvt_f32_m (z4, p0, z0))
114
115 /*
116 ** cvt_f32_s64_m_untied:
117 ** movprfx z0, z1
118 ** scvtf z0\.s, p0/m, z4\.d
119 ** ret
120 */
121 TEST_DUAL_Z (cvt_f32_s64_m_untied, svfloat32_t, svint64_t,
122 z0 = svcvt_f32_s64_m (z1, p0, z4),
123 z0 = svcvt_f32_m (z1, p0, z4))
124
125 /*
126 ** cvt_f32_u32_m_tied1:
127 ** ucvtf z0\.s, p0/m, z4\.s
128 ** ret
129 */
130 TEST_DUAL_Z (cvt_f32_u32_m_tied1, svfloat32_t, svuint32_t,
131 z0 = svcvt_f32_u32_m (z0, p0, z4),
132 z0 = svcvt_f32_m (z0, p0, z4))
133
134 /*
135 ** cvt_f32_u32_m_tied2:
136 ** mov (z[0-9]+)\.d, z0\.d
137 ** movprfx z0, z4
138 ** ucvtf z0\.s, p0/m, \1\.s
139 ** ret
140 */
141 TEST_DUAL_Z_REV (cvt_f32_u32_m_tied2, svfloat32_t, svuint32_t,
142 z0_res = svcvt_f32_u32_m (z4, p0, z0),
143 z0_res = svcvt_f32_m (z4, p0, z0))
144
145 /*
146 ** cvt_f32_u32_m_untied:
147 ** movprfx z0, z1
148 ** ucvtf z0\.s, p0/m, z4\.s
149 ** ret
150 */
151 TEST_DUAL_Z (cvt_f32_u32_m_untied, svfloat32_t, svuint32_t,
152 z0 = svcvt_f32_u32_m (z1, p0, z4),
153 z0 = svcvt_f32_m (z1, p0, z4))
154
155 /*
156 ** cvt_f32_u64_m_tied1:
157 ** ucvtf z0\.s, p0/m, z4\.d
158 ** ret
159 */
160 TEST_DUAL_Z (cvt_f32_u64_m_tied1, svfloat32_t, svuint64_t,
161 z0 = svcvt_f32_u64_m (z0, p0, z4),
162 z0 = svcvt_f32_m (z0, p0, z4))
163
164 /*
165 ** cvt_f32_u64_m_tied2:
166 ** mov (z[0-9]+\.d), z0\.d
167 ** movprfx z0, z4
168 ** ucvtf z0\.s, p0/m, \1
169 ** ret
170 */
171 TEST_DUAL_Z_REV (cvt_f32_u64_m_tied2, svfloat32_t, svuint64_t,
172 z0_res = svcvt_f32_u64_m (z4, p0, z0),
173 z0_res = svcvt_f32_m (z4, p0, z0))
174
175 /*
176 ** cvt_f32_u64_m_untied:
177 ** movprfx z0, z1
178 ** ucvtf z0\.s, p0/m, z4\.d
179 ** ret
180 */
181 TEST_DUAL_Z (cvt_f32_u64_m_untied, svfloat32_t, svuint64_t,
182 z0 = svcvt_f32_u64_m (z1, p0, z4),
183 z0 = svcvt_f32_m (z1, p0, z4))
184
185 /*
186 ** cvt_f32_f16_z_tied1:
187 ** mov (z[0-9]+)\.d, z0\.d
188 ** movprfx z0\.s, p0/z, \1\.s
189 ** fcvt z0\.s, p0/m, \1\.h
190 ** ret
191 */
192 TEST_DUAL_Z_REV (cvt_f32_f16_z_tied1, svfloat32_t, svfloat16_t,
193 z0_res = svcvt_f32_f16_z (p0, z0),
194 z0_res = svcvt_f32_z (p0, z0))
195
196 /*
197 ** cvt_f32_f16_z_untied:
198 ** movprfx z0\.s, p0/z, z4\.s
199 ** fcvt z0\.s, p0/m, z4\.h
200 ** ret
201 */
202 TEST_DUAL_Z (cvt_f32_f16_z_untied, svfloat32_t, svfloat16_t,
203 z0 = svcvt_f32_f16_z (p0, z4),
204 z0 = svcvt_f32_z (p0, z4))
205
206 /*
207 ** cvt_f32_f64_z_tied1:
208 ** mov (z[0-9]+\.d), z0\.d
209 ** movprfx z0\.d, p0/z, \1
210 ** fcvt z0\.s, p0/m, \1
211 ** ret
212 */
213 TEST_DUAL_Z_REV (cvt_f32_f64_z_tied1, svfloat32_t, svfloat64_t,
214 z0_res = svcvt_f32_f64_z (p0, z0),
215 z0_res = svcvt_f32_z (p0, z0))
216
217 /*
218 ** cvt_f32_f64_z_untied:
219 ** movprfx z0\.d, p0/z, z4\.d
220 ** fcvt z0\.s, p0/m, z4\.d
221 ** ret
222 */
223 TEST_DUAL_Z (cvt_f32_f64_z_untied, svfloat32_t, svfloat64_t,
224 z0 = svcvt_f32_f64_z (p0, z4),
225 z0 = svcvt_f32_z (p0, z4))
226
227 /*
228 ** cvt_f32_s32_z_tied1:
229 ** mov (z[0-9]+)\.d, z0\.d
230 ** movprfx z0\.s, p0/z, \1\.s
231 ** scvtf z0\.s, p0/m, \1\.s
232 ** ret
233 */
234 TEST_DUAL_Z_REV (cvt_f32_s32_z_tied1, svfloat32_t, svint32_t,
235 z0_res = svcvt_f32_s32_z (p0, z0),
236 z0_res = svcvt_f32_z (p0, z0))
237
238 /*
239 ** cvt_f32_s32_z_untied:
240 ** movprfx z0\.s, p0/z, z4\.s
241 ** scvtf z0\.s, p0/m, z4\.s
242 ** ret
243 */
244 TEST_DUAL_Z (cvt_f32_s32_z_untied, svfloat32_t, svint32_t,
245 z0 = svcvt_f32_s32_z (p0, z4),
246 z0 = svcvt_f32_z (p0, z4))
247
248 /*
249 ** cvt_f32_s64_z_tied1:
250 ** mov (z[0-9]+\.d), z0\.d
251 ** movprfx z0\.d, p0/z, \1
252 ** scvtf z0\.s, p0/m, \1
253 ** ret
254 */
255 TEST_DUAL_Z_REV (cvt_f32_s64_z_tied1, svfloat32_t, svint64_t,
256 z0_res = svcvt_f32_s64_z (p0, z0),
257 z0_res = svcvt_f32_z (p0, z0))
258
259 /*
260 ** cvt_f32_s64_z_untied:
261 ** movprfx z0\.d, p0/z, z4\.d
262 ** scvtf z0\.s, p0/m, z4\.d
263 ** ret
264 */
265 TEST_DUAL_Z (cvt_f32_s64_z_untied, svfloat32_t, svint64_t,
266 z0 = svcvt_f32_s64_z (p0, z4),
267 z0 = svcvt_f32_z (p0, z4))
268
269 /*
270 ** cvt_f32_u32_z_tied1:
271 ** mov (z[0-9]+)\.d, z0\.d
272 ** movprfx z0\.s, p0/z, \1\.s
273 ** ucvtf z0\.s, p0/m, \1\.s
274 ** ret
275 */
276 TEST_DUAL_Z_REV (cvt_f32_u32_z_tied1, svfloat32_t, svuint32_t,
277 z0_res = svcvt_f32_u32_z (p0, z0),
278 z0_res = svcvt_f32_z (p0, z0))
279
280 /*
281 ** cvt_f32_u32_z_untied:
282 ** movprfx z0\.s, p0/z, z4\.s
283 ** ucvtf z0\.s, p0/m, z4\.s
284 ** ret
285 */
286 TEST_DUAL_Z (cvt_f32_u32_z_untied, svfloat32_t, svuint32_t,
287 z0 = svcvt_f32_u32_z (p0, z4),
288 z0 = svcvt_f32_z (p0, z4))
289
290 /*
291 ** cvt_f32_u64_z_tied1:
292 ** mov (z[0-9]+\.d), z0\.d
293 ** movprfx z0\.d, p0/z, \1
294 ** ucvtf z0\.s, p0/m, \1
295 ** ret
296 */
297 TEST_DUAL_Z_REV (cvt_f32_u64_z_tied1, svfloat32_t, svuint64_t,
298 z0_res = svcvt_f32_u64_z (p0, z0),
299 z0_res = svcvt_f32_z (p0, z0))
300
301 /*
302 ** cvt_f32_u64_z_untied:
303 ** movprfx z0\.d, p0/z, z4\.d
304 ** ucvtf z0\.s, p0/m, z4\.d
305 ** ret
306 */
307 TEST_DUAL_Z (cvt_f32_u64_z_untied, svfloat32_t, svuint64_t,
308 z0 = svcvt_f32_u64_z (p0, z4),
309 z0 = svcvt_f32_z (p0, z4))
310
311 /*
312 ** cvt_f32_f16_x_tied1:
313 ** fcvt z0\.s, p0/m, z0\.h
314 ** ret
315 */
316 TEST_DUAL_Z_REV (cvt_f32_f16_x_tied1, svfloat32_t, svfloat16_t,
317 z0_res = svcvt_f32_f16_x (p0, z0),
318 z0_res = svcvt_f32_x (p0, z0))
319
320 /*
321 ** cvt_f32_f16_x_untied:
322 ** movprfx z0, z4
323 ** fcvt z0\.s, p0/m, z4\.h
324 ** ret
325 */
326 TEST_DUAL_Z (cvt_f32_f16_x_untied, svfloat32_t, svfloat16_t,
327 z0 = svcvt_f32_f16_x (p0, z4),
328 z0 = svcvt_f32_x (p0, z4))
329
330 /*
331 ** cvt_f32_f64_x_tied1:
332 ** fcvt z0\.s, p0/m, z0\.d
333 ** ret
334 */
335 TEST_DUAL_Z_REV (cvt_f32_f64_x_tied1, svfloat32_t, svfloat64_t,
336 z0_res = svcvt_f32_f64_x (p0, z0),
337 z0_res = svcvt_f32_x (p0, z0))
338
339 /*
340 ** cvt_f32_f64_x_untied:
341 ** movprfx z0, z4
342 ** fcvt z0\.s, p0/m, z4\.d
343 ** ret
344 */
345 TEST_DUAL_Z (cvt_f32_f64_x_untied, svfloat32_t, svfloat64_t,
346 z0 = svcvt_f32_f64_x (p0, z4),
347 z0 = svcvt_f32_x (p0, z4))
348
349 /*
350 ** cvt_f32_s32_x_tied1:
351 ** scvtf z0\.s, p0/m, z0\.s
352 ** ret
353 */
354 TEST_DUAL_Z_REV (cvt_f32_s32_x_tied1, svfloat32_t, svint32_t,
355 z0_res = svcvt_f32_s32_x (p0, z0),
356 z0_res = svcvt_f32_x (p0, z0))
357
358 /*
359 ** cvt_f32_s32_x_untied:
360 ** movprfx z0, z4
361 ** scvtf z0\.s, p0/m, z4\.s
362 ** ret
363 */
364 TEST_DUAL_Z (cvt_f32_s32_x_untied, svfloat32_t, svint32_t,
365 z0 = svcvt_f32_s32_x (p0, z4),
366 z0 = svcvt_f32_x (p0, z4))
367
368 /*
369 ** cvt_f32_s64_x_tied1:
370 ** scvtf z0\.s, p0/m, z0\.d
371 ** ret
372 */
373 TEST_DUAL_Z_REV (cvt_f32_s64_x_tied1, svfloat32_t, svint64_t,
374 z0_res = svcvt_f32_s64_x (p0, z0),
375 z0_res = svcvt_f32_x (p0, z0))
376
377 /*
378 ** cvt_f32_s64_x_untied:
379 ** movprfx z0, z4
380 ** scvtf z0\.s, p0/m, z4\.d
381 ** ret
382 */
383 TEST_DUAL_Z (cvt_f32_s64_x_untied, svfloat32_t, svint64_t,
384 z0 = svcvt_f32_s64_x (p0, z4),
385 z0 = svcvt_f32_x (p0, z4))
386
387 /*
388 ** cvt_f32_u32_x_tied1:
389 ** ucvtf z0\.s, p0/m, z0\.s
390 ** ret
391 */
392 TEST_DUAL_Z_REV (cvt_f32_u32_x_tied1, svfloat32_t, svuint32_t,
393 z0_res = svcvt_f32_u32_x (p0, z0),
394 z0_res = svcvt_f32_x (p0, z0))
395
396 /*
397 ** cvt_f32_u32_x_untied:
398 ** movprfx z0, z4
399 ** ucvtf z0\.s, p0/m, z4\.s
400 ** ret
401 */
402 TEST_DUAL_Z (cvt_f32_u32_x_untied, svfloat32_t, svuint32_t,
403 z0 = svcvt_f32_u32_x (p0, z4),
404 z0 = svcvt_f32_x (p0, z4))
405
406 /*
407 ** cvt_f32_u64_x_tied1:
408 ** ucvtf z0\.s, p0/m, z0\.d
409 ** ret
410 */
411 TEST_DUAL_Z_REV (cvt_f32_u64_x_tied1, svfloat32_t, svuint64_t,
412 z0_res = svcvt_f32_u64_x (p0, z0),
413 z0_res = svcvt_f32_x (p0, z0))
414
415 /*
416 ** cvt_f32_u64_x_untied:
417 ** movprfx z0, z4
418 ** ucvtf z0\.s, p0/m, z4\.d
419 ** ret
420 */
421 TEST_DUAL_Z (cvt_f32_u64_x_untied, svfloat32_t, svuint64_t,
422 z0 = svcvt_f32_u64_x (p0, z4),
423 z0 = svcvt_f32_x (p0, z4))
424
425 /*
426 ** ptrue_cvt_f32_f16_x_tied1:
427 ** ...
428 ** ptrue p[0-9]+\.b[^\n]*
429 ** ...
430 ** ret
431 */
432 TEST_DUAL_Z_REV (ptrue_cvt_f32_f16_x_tied1, svfloat32_t, svfloat16_t,
433 z0_res = svcvt_f32_f16_x (svptrue_b32 (), z0),
434 z0_res = svcvt_f32_x (svptrue_b32 (), z0))
435
436 /*
437 ** ptrue_cvt_f32_f16_x_untied:
438 ** ...
439 ** ptrue p[0-9]+\.b[^\n]*
440 ** ...
441 ** ret
442 */
443 TEST_DUAL_Z (ptrue_cvt_f32_f16_x_untied, svfloat32_t, svfloat16_t,
444 z0 = svcvt_f32_f16_x (svptrue_b32 (), z4),
445 z0 = svcvt_f32_x (svptrue_b32 (), z4))
446
447 /*
448 ** ptrue_cvt_f32_f64_x_tied1:
449 ** ...
450 ** ptrue p[0-9]+\.b[^\n]*
451 ** ...
452 ** ret
453 */
454 TEST_DUAL_Z_REV (ptrue_cvt_f32_f64_x_tied1, svfloat32_t, svfloat64_t,
455 z0_res = svcvt_f32_f64_x (svptrue_b64 (), z0),
456 z0_res = svcvt_f32_x (svptrue_b64 (), z0))
457
458 /*
459 ** ptrue_cvt_f32_f64_x_untied:
460 ** ...
461 ** ptrue p[0-9]+\.b[^\n]*
462 ** ...
463 ** ret
464 */
465 TEST_DUAL_Z (ptrue_cvt_f32_f64_x_untied, svfloat32_t, svfloat64_t,
466 z0 = svcvt_f32_f64_x (svptrue_b64 (), z4),
467 z0 = svcvt_f32_x (svptrue_b64 (), z4))
468
469 /*
470 ** ptrue_cvt_f32_s32_x_tied1:
471 ** ...
472 ** ptrue p[0-9]+\.b[^\n]*
473 ** ...
474 ** ret
475 */
476 TEST_DUAL_Z_REV (ptrue_cvt_f32_s32_x_tied1, svfloat32_t, svint32_t,
477 z0_res = svcvt_f32_s32_x (svptrue_b32 (), z0),
478 z0_res = svcvt_f32_x (svptrue_b32 (), z0))
479
480 /*
481 ** ptrue_cvt_f32_s32_x_untied:
482 ** ...
483 ** ptrue p[0-9]+\.b[^\n]*
484 ** ...
485 ** ret
486 */
487 TEST_DUAL_Z (ptrue_cvt_f32_s32_x_untied, svfloat32_t, svint32_t,
488 z0 = svcvt_f32_s32_x (svptrue_b32 (), z4),
489 z0 = svcvt_f32_x (svptrue_b32 (), z4))
490
491 /*
492 ** ptrue_cvt_f32_s64_x_tied1:
493 ** ...
494 ** ptrue p[0-9]+\.b[^\n]*
495 ** ...
496 ** ret
497 */
498 TEST_DUAL_Z_REV (ptrue_cvt_f32_s64_x_tied1, svfloat32_t, svint64_t,
499 z0_res = svcvt_f32_s64_x (svptrue_b64 (), z0),
500 z0_res = svcvt_f32_x (svptrue_b64 (), z0))
501
502 /*
503 ** ptrue_cvt_f32_s64_x_untied:
504 ** ...
505 ** ptrue p[0-9]+\.b[^\n]*
506 ** ...
507 ** ret
508 */
509 TEST_DUAL_Z (ptrue_cvt_f32_s64_x_untied, svfloat32_t, svint64_t,
510 z0 = svcvt_f32_s64_x (svptrue_b64 (), z4),
511 z0 = svcvt_f32_x (svptrue_b64 (), z4))
512
513 /*
514 ** ptrue_cvt_f32_u32_x_tied1:
515 ** ...
516 ** ptrue p[0-9]+\.b[^\n]*
517 ** ...
518 ** ret
519 */
520 TEST_DUAL_Z_REV (ptrue_cvt_f32_u32_x_tied1, svfloat32_t, svuint32_t,
521 z0_res = svcvt_f32_u32_x (svptrue_b32 (), z0),
522 z0_res = svcvt_f32_x (svptrue_b32 (), z0))
523
524 /*
525 ** ptrue_cvt_f32_u32_x_untied:
526 ** ...
527 ** ptrue p[0-9]+\.b[^\n]*
528 ** ...
529 ** ret
530 */
531 TEST_DUAL_Z (ptrue_cvt_f32_u32_x_untied, svfloat32_t, svuint32_t,
532 z0 = svcvt_f32_u32_x (svptrue_b32 (), z4),
533 z0 = svcvt_f32_x (svptrue_b32 (), z4))
534
535 /*
536 ** ptrue_cvt_f32_u64_x_tied1:
537 ** ...
538 ** ptrue p[0-9]+\.b[^\n]*
539 ** ...
540 ** ret
541 */
542 TEST_DUAL_Z_REV (ptrue_cvt_f32_u64_x_tied1, svfloat32_t, svuint64_t,
543 z0_res = svcvt_f32_u64_x (svptrue_b64 (), z0),
544 z0_res = svcvt_f32_x (svptrue_b64 (), z0))
545
546 /*
547 ** ptrue_cvt_f32_u64_x_untied:
548 ** ...
549 ** ptrue p[0-9]+\.b[^\n]*
550 ** ...
551 ** ret
552 */
553 TEST_DUAL_Z (ptrue_cvt_f32_u64_x_untied, svfloat32_t, svuint64_t,
554 z0 = svcvt_f32_u64_x (svptrue_b64 (), z4),
555 z0 = svcvt_f32_x (svptrue_b64 (), z4))