1 /* { dg-final { check-function-bodies "**" "" "-DCHECK_ASM" } } */
2
3 #include "test_sve_acle.h"
4
5 /*
6 ** sub_f32_m_tied1:
7 ** fsub z0\.s, p0/m, z0\.s, z1\.s
8 ** ret
9 */
10 TEST_UNIFORM_Z (sub_f32_m_tied1, svfloat32_t,
11 z0 = svsub_f32_m (p0, z0, z1),
12 z0 = svsub_m (p0, z0, z1))
13
14 /*
15 ** sub_f32_m_tied2:
16 ** mov (z[0-9]+)\.d, z0\.d
17 ** movprfx z0, z1
18 ** fsub z0\.s, p0/m, z0\.s, \1\.s
19 ** ret
20 */
21 TEST_UNIFORM_Z (sub_f32_m_tied2, svfloat32_t,
22 z0 = svsub_f32_m (p0, z1, z0),
23 z0 = svsub_m (p0, z1, z0))
24
25 /*
26 ** sub_f32_m_untied:
27 ** movprfx z0, z1
28 ** fsub z0\.s, p0/m, z0\.s, z2\.s
29 ** ret
30 */
31 TEST_UNIFORM_Z (sub_f32_m_untied, svfloat32_t,
32 z0 = svsub_f32_m (p0, z1, z2),
33 z0 = svsub_m (p0, z1, z2))
34
35 /*
36 ** sub_s4_f32_m_tied1:
37 ** mov (z[0-9]+\.s), s4
38 ** fsub z0\.s, p0/m, z0\.s, \1
39 ** ret
40 */
41 TEST_UNIFORM_ZD (sub_s4_f32_m_tied1, svfloat32_t, float,
42 z0 = svsub_n_f32_m (p0, z0, d4),
43 z0 = svsub_m (p0, z0, d4))
44
45 /*
46 ** sub_s4_f32_m_untied:
47 ** mov (z[0-9]+\.s), s4
48 ** movprfx z0, z1
49 ** fsub z0\.s, p0/m, z0\.s, \1
50 ** ret
51 */
52 TEST_UNIFORM_ZD (sub_s4_f32_m_untied, svfloat32_t, float,
53 z0 = svsub_n_f32_m (p0, z1, d4),
54 z0 = svsub_m (p0, z1, d4))
55
56 /*
57 ** sub_1_f32_m_tied1:
58 ** fsub z0\.s, p0/m, z0\.s, #1\.0
59 ** ret
60 */
61 TEST_UNIFORM_Z (sub_1_f32_m_tied1, svfloat32_t,
62 z0 = svsub_n_f32_m (p0, z0, 1),
63 z0 = svsub_m (p0, z0, 1))
64
65 /*
66 ** sub_1_f32_m_untied:
67 ** movprfx z0, z1
68 ** fsub z0\.s, p0/m, z0\.s, #1\.0
69 ** ret
70 */
71 TEST_UNIFORM_Z (sub_1_f32_m_untied, svfloat32_t,
72 z0 = svsub_n_f32_m (p0, z1, 1),
73 z0 = svsub_m (p0, z1, 1))
74
75 /*
76 ** sub_0p5_f32_m_tied1:
77 ** fsub z0\.s, p0/m, z0\.s, #0\.5
78 ** ret
79 */
80 TEST_UNIFORM_Z (sub_0p5_f32_m_tied1, svfloat32_t,
81 z0 = svsub_n_f32_m (p0, z0, 0.5),
82 z0 = svsub_m (p0, z0, 0.5))
83
84 /*
85 ** sub_0p5_f32_m_untied:
86 ** movprfx z0, z1
87 ** fsub z0\.s, p0/m, z0\.s, #0\.5
88 ** ret
89 */
90 TEST_UNIFORM_Z (sub_0p5_f32_m_untied, svfloat32_t,
91 z0 = svsub_n_f32_m (p0, z1, 0.5),
92 z0 = svsub_m (p0, z1, 0.5))
93
94 /*
95 ** sub_m1_f32_m_tied1:
96 ** fadd z0\.s, p0/m, z0\.s, #1\.0
97 ** ret
98 */
99 TEST_UNIFORM_Z (sub_m1_f32_m_tied1, svfloat32_t,
100 z0 = svsub_n_f32_m (p0, z0, -1),
101 z0 = svsub_m (p0, z0, -1))
102
103 /*
104 ** sub_m1_f32_m_untied:
105 ** movprfx z0, z1
106 ** fadd z0\.s, p0/m, z0\.s, #1\.0
107 ** ret
108 */
109 TEST_UNIFORM_Z (sub_m1_f32_m_untied, svfloat32_t,
110 z0 = svsub_n_f32_m (p0, z1, -1),
111 z0 = svsub_m (p0, z1, -1))
112
113 /*
114 ** sub_m0p5_f32_m_tied1:
115 ** fadd z0\.s, p0/m, z0\.s, #0\.5
116 ** ret
117 */
118 TEST_UNIFORM_Z (sub_m0p5_f32_m_tied1, svfloat32_t,
119 z0 = svsub_n_f32_m (p0, z0, -0.5),
120 z0 = svsub_m (p0, z0, -0.5))
121
122 /*
123 ** sub_m0p5_f32_m_untied:
124 ** movprfx z0, z1
125 ** fadd z0\.s, p0/m, z0\.s, #0\.5
126 ** ret
127 */
128 TEST_UNIFORM_Z (sub_m0p5_f32_m_untied, svfloat32_t,
129 z0 = svsub_n_f32_m (p0, z1, -0.5),
130 z0 = svsub_m (p0, z1, -0.5))
131
132 /*
133 ** sub_m2_f32_m:
134 ** fmov (z[0-9]+\.s), #2\.0(?:e\+0)?
135 ** fadd z0\.s, p0/m, z0\.s, \1
136 ** ret
137 */
138 TEST_UNIFORM_Z (sub_m2_f32_m, svfloat32_t,
139 z0 = svsub_n_f32_m (p0, z0, -2),
140 z0 = svsub_m (p0, z0, -2))
141
142 /*
143 ** sub_f32_z_tied1:
144 ** movprfx z0\.s, p0/z, z0\.s
145 ** fsub z0\.s, p0/m, z0\.s, z1\.s
146 ** ret
147 */
148 TEST_UNIFORM_Z (sub_f32_z_tied1, svfloat32_t,
149 z0 = svsub_f32_z (p0, z0, z1),
150 z0 = svsub_z (p0, z0, z1))
151
152 /*
153 ** sub_f32_z_tied2:
154 ** movprfx z0\.s, p0/z, z0\.s
155 ** fsubr z0\.s, p0/m, z0\.s, z1\.s
156 ** ret
157 */
158 TEST_UNIFORM_Z (sub_f32_z_tied2, svfloat32_t,
159 z0 = svsub_f32_z (p0, z1, z0),
160 z0 = svsub_z (p0, z1, z0))
161
162 /*
163 ** sub_f32_z_untied:
164 ** (
165 ** movprfx z0\.s, p0/z, z1\.s
166 ** fsub z0\.s, p0/m, z0\.s, z2\.s
167 ** |
168 ** movprfx z0\.s, p0/z, z2\.s
169 ** fsubr z0\.s, p0/m, z0\.s, z1\.s
170 ** )
171 ** ret
172 */
173 TEST_UNIFORM_Z (sub_f32_z_untied, svfloat32_t,
174 z0 = svsub_f32_z (p0, z1, z2),
175 z0 = svsub_z (p0, z1, z2))
176
177 /*
178 ** sub_s4_f32_z_tied1:
179 ** mov (z[0-9]+\.s), s4
180 ** movprfx z0\.s, p0/z, z0\.s
181 ** fsub z0\.s, p0/m, z0\.s, \1
182 ** ret
183 */
184 TEST_UNIFORM_ZD (sub_s4_f32_z_tied1, svfloat32_t, float,
185 z0 = svsub_n_f32_z (p0, z0, d4),
186 z0 = svsub_z (p0, z0, d4))
187
188 /*
189 ** sub_s4_f32_z_untied:
190 ** mov (z[0-9]+\.s), s4
191 ** (
192 ** movprfx z0\.s, p0/z, z1\.s
193 ** fsub z0\.s, p0/m, z0\.s, \1
194 ** |
195 ** movprfx z0\.s, p0/z, \1
196 ** fsubr z0\.s, p0/m, z0\.s, z1\.s
197 ** )
198 ** ret
199 */
200 TEST_UNIFORM_ZD (sub_s4_f32_z_untied, svfloat32_t, float,
201 z0 = svsub_n_f32_z (p0, z1, d4),
202 z0 = svsub_z (p0, z1, d4))
203
204 /*
205 ** sub_1_f32_z_tied1:
206 ** movprfx z0\.s, p0/z, z0\.s
207 ** fsub z0\.s, p0/m, z0\.s, #1\.0
208 ** ret
209 */
210 TEST_UNIFORM_Z (sub_1_f32_z_tied1, svfloat32_t,
211 z0 = svsub_n_f32_z (p0, z0, 1),
212 z0 = svsub_z (p0, z0, 1))
213
214 /*
215 ** sub_1_f32_z_untied:
216 ** movprfx z0\.s, p0/z, z1\.s
217 ** fsub z0\.s, p0/m, z0\.s, #1\.0
218 ** ret
219 */
220 TEST_UNIFORM_Z (sub_1_f32_z_untied, svfloat32_t,
221 z0 = svsub_n_f32_z (p0, z1, 1),
222 z0 = svsub_z (p0, z1, 1))
223
224 /*
225 ** sub_0p5_f32_z_tied1:
226 ** movprfx z0\.s, p0/z, z0\.s
227 ** fsub z0\.s, p0/m, z0\.s, #0\.5
228 ** ret
229 */
230 TEST_UNIFORM_Z (sub_0p5_f32_z_tied1, svfloat32_t,
231 z0 = svsub_n_f32_z (p0, z0, 0.5),
232 z0 = svsub_z (p0, z0, 0.5))
233
234 /*
235 ** sub_0p5_f32_z_untied:
236 ** movprfx z0\.s, p0/z, z1\.s
237 ** fsub z0\.s, p0/m, z0\.s, #0\.5
238 ** ret
239 */
240 TEST_UNIFORM_Z (sub_0p5_f32_z_untied, svfloat32_t,
241 z0 = svsub_n_f32_z (p0, z1, 0.5),
242 z0 = svsub_z (p0, z1, 0.5))
243
244 /*
245 ** sub_m1_f32_z_tied1:
246 ** movprfx z0\.s, p0/z, z0\.s
247 ** fadd z0\.s, p0/m, z0\.s, #1\.0
248 ** ret
249 */
250 TEST_UNIFORM_Z (sub_m1_f32_z_tied1, svfloat32_t,
251 z0 = svsub_n_f32_z (p0, z0, -1),
252 z0 = svsub_z (p0, z0, -1))
253
254 /*
255 ** sub_m1_f32_z_untied:
256 ** movprfx z0\.s, p0/z, z1\.s
257 ** fadd z0\.s, p0/m, z0\.s, #1\.0
258 ** ret
259 */
260 TEST_UNIFORM_Z (sub_m1_f32_z_untied, svfloat32_t,
261 z0 = svsub_n_f32_z (p0, z1, -1),
262 z0 = svsub_z (p0, z1, -1))
263
264 /*
265 ** sub_m0p5_f32_z_tied1:
266 ** movprfx z0\.s, p0/z, z0\.s
267 ** fadd z0\.s, p0/m, z0\.s, #0\.5
268 ** ret
269 */
270 TEST_UNIFORM_Z (sub_m0p5_f32_z_tied1, svfloat32_t,
271 z0 = svsub_n_f32_z (p0, z0, -0.5),
272 z0 = svsub_z (p0, z0, -0.5))
273
274 /*
275 ** sub_m0p5_f32_z_untied:
276 ** movprfx z0\.s, p0/z, z1\.s
277 ** fadd z0\.s, p0/m, z0\.s, #0\.5
278 ** ret
279 */
280 TEST_UNIFORM_Z (sub_m0p5_f32_z_untied, svfloat32_t,
281 z0 = svsub_n_f32_z (p0, z1, -0.5),
282 z0 = svsub_z (p0, z1, -0.5))
283
284 /*
285 ** sub_m2_f32_z:
286 ** fmov (z[0-9]+\.s), #2\.0(?:e\+0)?
287 ** movprfx z0\.s, p0/z, z0\.s
288 ** fadd z0\.s, p0/m, z0\.s, \1
289 ** ret
290 */
291 TEST_UNIFORM_Z (sub_m2_f32_z, svfloat32_t,
292 z0 = svsub_n_f32_z (p0, z0, -2),
293 z0 = svsub_z (p0, z0, -2))
294
295 /*
296 ** sub_f32_x_tied1:
297 ** fsub z0\.s, p0/m, z0\.s, z1\.s
298 ** ret
299 */
300 TEST_UNIFORM_Z (sub_f32_x_tied1, svfloat32_t,
301 z0 = svsub_f32_x (p0, z0, z1),
302 z0 = svsub_x (p0, z0, z1))
303
304 /*
305 ** sub_f32_x_tied2:
306 ** fsubr z0\.s, p0/m, z0\.s, z1\.s
307 ** ret
308 */
309 TEST_UNIFORM_Z (sub_f32_x_tied2, svfloat32_t,
310 z0 = svsub_f32_x (p0, z1, z0),
311 z0 = svsub_x (p0, z1, z0))
312
313 /*
314 ** sub_f32_x_untied:
315 ** (
316 ** movprfx z0, z1
317 ** fsub z0\.s, p0/m, z0\.s, z2\.s
318 ** |
319 ** movprfx z0, z2
320 ** fsubr z0\.s, p0/m, z0\.s, z1\.s
321 ** )
322 ** ret
323 */
324 TEST_UNIFORM_Z (sub_f32_x_untied, svfloat32_t,
325 z0 = svsub_f32_x (p0, z1, z2),
326 z0 = svsub_x (p0, z1, z2))
327
328 /*
329 ** sub_s4_f32_x_tied1:
330 ** mov (z[0-9]+\.s), s4
331 ** fsub z0\.s, p0/m, z0\.s, \1
332 ** ret
333 */
334 TEST_UNIFORM_ZD (sub_s4_f32_x_tied1, svfloat32_t, float,
335 z0 = svsub_n_f32_x (p0, z0, d4),
336 z0 = svsub_x (p0, z0, d4))
337
338 /*
339 ** sub_s4_f32_x_untied:
340 ** mov z0\.s, s4
341 ** fsubr z0\.s, p0/m, z0\.s, z1\.s
342 ** ret
343 */
344 TEST_UNIFORM_ZD (sub_s4_f32_x_untied, svfloat32_t, float,
345 z0 = svsub_n_f32_x (p0, z1, d4),
346 z0 = svsub_x (p0, z1, d4))
347
348 /*
349 ** sub_1_f32_x_tied1:
350 ** fsub z0\.s, p0/m, z0\.s, #1\.0
351 ** ret
352 */
353 TEST_UNIFORM_Z (sub_1_f32_x_tied1, svfloat32_t,
354 z0 = svsub_n_f32_x (p0, z0, 1),
355 z0 = svsub_x (p0, z0, 1))
356
357 /*
358 ** sub_1_f32_x_untied:
359 ** movprfx z0, z1
360 ** fsub z0\.s, p0/m, z0\.s, #1\.0
361 ** ret
362 */
363 TEST_UNIFORM_Z (sub_1_f32_x_untied, svfloat32_t,
364 z0 = svsub_n_f32_x (p0, z1, 1),
365 z0 = svsub_x (p0, z1, 1))
366
367 /*
368 ** sub_0p5_f32_x_tied1:
369 ** fsub z0\.s, p0/m, z0\.s, #0\.5
370 ** ret
371 */
372 TEST_UNIFORM_Z (sub_0p5_f32_x_tied1, svfloat32_t,
373 z0 = svsub_n_f32_x (p0, z0, 0.5),
374 z0 = svsub_x (p0, z0, 0.5))
375
376 /*
377 ** sub_0p5_f32_x_untied:
378 ** movprfx z0, z1
379 ** fsub z0\.s, p0/m, z0\.s, #0\.5
380 ** ret
381 */
382 TEST_UNIFORM_Z (sub_0p5_f32_x_untied, svfloat32_t,
383 z0 = svsub_n_f32_x (p0, z1, 0.5),
384 z0 = svsub_x (p0, z1, 0.5))
385
386 /*
387 ** sub_m1_f32_x_tied1:
388 ** fadd z0\.s, p0/m, z0\.s, #1\.0
389 ** ret
390 */
391 TEST_UNIFORM_Z (sub_m1_f32_x_tied1, svfloat32_t,
392 z0 = svsub_n_f32_x (p0, z0, -1),
393 z0 = svsub_x (p0, z0, -1))
394
395 /*
396 ** sub_m1_f32_x_untied:
397 ** movprfx z0, z1
398 ** fadd z0\.s, p0/m, z0\.s, #1\.0
399 ** ret
400 */
401 TEST_UNIFORM_Z (sub_m1_f32_x_untied, svfloat32_t,
402 z0 = svsub_n_f32_x (p0, z1, -1),
403 z0 = svsub_x (p0, z1, -1))
404
405 /*
406 ** sub_m0p5_f32_x_tied1:
407 ** fadd z0\.s, p0/m, z0\.s, #0\.5
408 ** ret
409 */
410 TEST_UNIFORM_Z (sub_m0p5_f32_x_tied1, svfloat32_t,
411 z0 = svsub_n_f32_x (p0, z0, -0.5),
412 z0 = svsub_x (p0, z0, -0.5))
413
414 /*
415 ** sub_m0p5_f32_x_untied:
416 ** movprfx z0, z1
417 ** fadd z0\.s, p0/m, z0\.s, #0\.5
418 ** ret
419 */
420 TEST_UNIFORM_Z (sub_m0p5_f32_x_untied, svfloat32_t,
421 z0 = svsub_n_f32_x (p0, z1, -0.5),
422 z0 = svsub_x (p0, z1, -0.5))
423
424 /*
425 ** sub_2_f32_x_tied1:
426 ** fmov (z[0-9]+\.s), #-2\.0(?:e\+0)?
427 ** fadd z0\.s, p0/m, z0\.s, \1
428 ** ret
429 */
430 TEST_UNIFORM_Z (sub_2_f32_x_tied1, svfloat32_t,
431 z0 = svsub_n_f32_x (p0, z0, 2),
432 z0 = svsub_x (p0, z0, 2))
433
434 /*
435 ** sub_2_f32_x_untied:
436 ** fmov z0\.s, #-2\.0(?:e\+0)?
437 ** fadd z0\.s, p0/m, z0\.s, z1\.s
438 ** ret
439 */
440 TEST_UNIFORM_Z (sub_2_f32_x_untied, svfloat32_t,
441 z0 = svsub_n_f32_x (p0, z1, 2),
442 z0 = svsub_x (p0, z1, 2))
443
444 /*
445 ** ptrue_sub_f32_x_tied1:
446 ** fsub z0\.s, z0\.s, z1\.s
447 ** ret
448 */
449 TEST_UNIFORM_Z (ptrue_sub_f32_x_tied1, svfloat32_t,
450 z0 = svsub_f32_x (svptrue_b32 (), z0, z1),
451 z0 = svsub_x (svptrue_b32 (), z0, z1))
452
453 /*
454 ** ptrue_sub_f32_x_tied2:
455 ** fsub z0\.s, z1\.s, z0\.s
456 ** ret
457 */
458 TEST_UNIFORM_Z (ptrue_sub_f32_x_tied2, svfloat32_t,
459 z0 = svsub_f32_x (svptrue_b32 (), z1, z0),
460 z0 = svsub_x (svptrue_b32 (), z1, z0))
461
462 /*
463 ** ptrue_sub_f32_x_untied:
464 ** fsub z0\.s, z1\.s, z2\.s
465 ** ret
466 */
467 TEST_UNIFORM_Z (ptrue_sub_f32_x_untied, svfloat32_t,
468 z0 = svsub_f32_x (svptrue_b32 (), z1, z2),
469 z0 = svsub_x (svptrue_b32 (), z1, z2))
470
471 /*
472 ** ptrue_sub_1_f32_x_tied1:
473 ** ...
474 ** ptrue p[0-9]+\.b[^\n]*
475 ** ...
476 ** ret
477 */
478 TEST_UNIFORM_Z (ptrue_sub_1_f32_x_tied1, svfloat32_t,
479 z0 = svsub_n_f32_x (svptrue_b32 (), z0, 1),
480 z0 = svsub_x (svptrue_b32 (), z0, 1))
481
482 /*
483 ** ptrue_sub_1_f32_x_untied:
484 ** ...
485 ** ptrue p[0-9]+\.b[^\n]*
486 ** ...
487 ** ret
488 */
489 TEST_UNIFORM_Z (ptrue_sub_1_f32_x_untied, svfloat32_t,
490 z0 = svsub_n_f32_x (svptrue_b32 (), z1, 1),
491 z0 = svsub_x (svptrue_b32 (), z1, 1))
492
493 /*
494 ** ptrue_sub_0p5_f32_x_tied1:
495 ** ...
496 ** ptrue p[0-9]+\.b[^\n]*
497 ** ...
498 ** ret
499 */
500 TEST_UNIFORM_Z (ptrue_sub_0p5_f32_x_tied1, svfloat32_t,
501 z0 = svsub_n_f32_x (svptrue_b32 (), z0, 0.5),
502 z0 = svsub_x (svptrue_b32 (), z0, 0.5))
503
504 /*
505 ** ptrue_sub_0p5_f32_x_untied:
506 ** ...
507 ** ptrue p[0-9]+\.b[^\n]*
508 ** ...
509 ** ret
510 */
511 TEST_UNIFORM_Z (ptrue_sub_0p5_f32_x_untied, svfloat32_t,
512 z0 = svsub_n_f32_x (svptrue_b32 (), z1, 0.5),
513 z0 = svsub_x (svptrue_b32 (), z1, 0.5))
514
515 /*
516 ** ptrue_sub_m1_f32_x_tied1:
517 ** ...
518 ** ptrue p[0-9]+\.b[^\n]*
519 ** ...
520 ** ret
521 */
522 TEST_UNIFORM_Z (ptrue_sub_m1_f32_x_tied1, svfloat32_t,
523 z0 = svsub_n_f32_x (svptrue_b32 (), z0, -1),
524 z0 = svsub_x (svptrue_b32 (), z0, -1))
525
526 /*
527 ** ptrue_sub_m1_f32_x_untied:
528 ** ...
529 ** ptrue p[0-9]+\.b[^\n]*
530 ** ...
531 ** ret
532 */
533 TEST_UNIFORM_Z (ptrue_sub_m1_f32_x_untied, svfloat32_t,
534 z0 = svsub_n_f32_x (svptrue_b32 (), z1, -1),
535 z0 = svsub_x (svptrue_b32 (), z1, -1))
536
537 /*
538 ** ptrue_sub_m0p5_f32_x_tied1:
539 ** ...
540 ** ptrue p[0-9]+\.b[^\n]*
541 ** ...
542 ** ret
543 */
544 TEST_UNIFORM_Z (ptrue_sub_m0p5_f32_x_tied1, svfloat32_t,
545 z0 = svsub_n_f32_x (svptrue_b32 (), z0, -0.5),
546 z0 = svsub_x (svptrue_b32 (), z0, -0.5))
547
548 /*
549 ** ptrue_sub_m0p5_f32_x_untied:
550 ** ...
551 ** ptrue p[0-9]+\.b[^\n]*
552 ** ...
553 ** ret
554 */
555 TEST_UNIFORM_Z (ptrue_sub_m0p5_f32_x_untied, svfloat32_t,
556 z0 = svsub_n_f32_x (svptrue_b32 (), z1, -0.5),
557 z0 = svsub_x (svptrue_b32 (), z1, -0.5))
558
559 /*
560 ** ptrue_sub_2_f32_x_tied1:
561 ** fmov (z[0-9]+\.s), #-2\.0(?:e\+0)?
562 ** fadd z0\.s, (z0\.s, \1|\1, z0\.s)
563 ** ret
564 */
565 TEST_UNIFORM_Z (ptrue_sub_2_f32_x_tied1, svfloat32_t,
566 z0 = svsub_n_f32_x (svptrue_b32 (), z0, 2),
567 z0 = svsub_x (svptrue_b32 (), z0, 2))
568
569 /*
570 ** ptrue_sub_2_f32_x_untied:
571 ** fmov (z[0-9]+\.s), #-2\.0(?:e\+0)?
572 ** fadd z0\.s, (z1\.s, \1|\1, z1\.s)
573 ** ret
574 */
575 TEST_UNIFORM_Z (ptrue_sub_2_f32_x_untied, svfloat32_t,
576 z0 = svsub_n_f32_x (svptrue_b32 (), z1, 2),
577 z0 = svsub_x (svptrue_b32 (), z1, 2))