1 /* { dg-final { check-function-bodies "**" "" "-DCHECK_ASM" } } */
2
3 #include "test_sve_acle.h"
4
5 /*
6 ** mulx_f32_m_tied1:
7 ** fmulx z0\.s, p0/m, z0\.s, z1\.s
8 ** ret
9 */
10 TEST_UNIFORM_Z (mulx_f32_m_tied1, svfloat32_t,
11 z0 = svmulx_f32_m (p0, z0, z1),
12 z0 = svmulx_m (p0, z0, z1))
13
14 /*
15 ** mulx_f32_m_tied2:
16 ** mov (z[0-9]+)\.d, z0\.d
17 ** movprfx z0, z1
18 ** fmulx z0\.s, p0/m, z0\.s, \1\.s
19 ** ret
20 */
21 TEST_UNIFORM_Z (mulx_f32_m_tied2, svfloat32_t,
22 z0 = svmulx_f32_m (p0, z1, z0),
23 z0 = svmulx_m (p0, z1, z0))
24
25 /*
26 ** mulx_f32_m_untied:
27 ** movprfx z0, z1
28 ** fmulx z0\.s, p0/m, z0\.s, z2\.s
29 ** ret
30 */
31 TEST_UNIFORM_Z (mulx_f32_m_untied, svfloat32_t,
32 z0 = svmulx_f32_m (p0, z1, z2),
33 z0 = svmulx_m (p0, z1, z2))
34
35 /*
36 ** mulx_s4_f32_m_tied1:
37 ** mov (z[0-9]+\.s), s4
38 ** fmulx z0\.s, p0/m, z0\.s, \1
39 ** ret
40 */
41 TEST_UNIFORM_ZD (mulx_s4_f32_m_tied1, svfloat32_t, float,
42 z0 = svmulx_n_f32_m (p0, z0, d4),
43 z0 = svmulx_m (p0, z0, d4))
44
45 /*
46 ** mulx_s4_f32_m_untied:
47 ** mov (z[0-9]+\.s), s4
48 ** movprfx z0, z1
49 ** fmulx z0\.s, p0/m, z0\.s, \1
50 ** ret
51 */
52 TEST_UNIFORM_ZD (mulx_s4_f32_m_untied, svfloat32_t, float,
53 z0 = svmulx_n_f32_m (p0, z1, d4),
54 z0 = svmulx_m (p0, z1, d4))
55
56 /*
57 ** mulx_1_f32_m_tied1:
58 ** fmov (z[0-9]+\.s), #1\.0(?:e\+0)?
59 ** fmulx z0\.s, p0/m, z0\.s, \1
60 ** ret
61 */
62 TEST_UNIFORM_Z (mulx_1_f32_m_tied1, svfloat32_t,
63 z0 = svmulx_n_f32_m (p0, z0, 1),
64 z0 = svmulx_m (p0, z0, 1))
65
66 /*
67 ** mulx_1_f32_m_untied: { xfail *-*-* }
68 ** fmov (z[0-9]+\.s), #1\.0(?:e\+0)?
69 ** movprfx z0, z1
70 ** fmulx z0\.s, p0/m, z0\.s, \1
71 ** ret
72 */
73 TEST_UNIFORM_Z (mulx_1_f32_m_untied, svfloat32_t,
74 z0 = svmulx_n_f32_m (p0, z1, 1),
75 z0 = svmulx_m (p0, z1, 1))
76
77 /*
78 ** mulx_0p5_f32_m_tied1:
79 ** fmov (z[0-9]+\.s), #(?:0\.5|5\.0e-1)
80 ** fmulx z0\.s, p0/m, z0\.s, \1
81 ** ret
82 */
83 TEST_UNIFORM_Z (mulx_0p5_f32_m_tied1, svfloat32_t,
84 z0 = svmulx_n_f32_m (p0, z0, 0.5),
85 z0 = svmulx_m (p0, z0, 0.5))
86
87 /*
88 ** mulx_0p5_f32_m_untied: { xfail *-*-* }
89 ** fmov (z[0-9]+\.s), #(?:0\.5|5\.0e-1)
90 ** movprfx z0, z1
91 ** fmulx z0\.s, p0/m, z0\.s, \1
92 ** ret
93 */
94 TEST_UNIFORM_Z (mulx_0p5_f32_m_untied, svfloat32_t,
95 z0 = svmulx_n_f32_m (p0, z1, 0.5),
96 z0 = svmulx_m (p0, z1, 0.5))
97
98 /*
99 ** mulx_2_f32_m_tied1:
100 ** fmov (z[0-9]+\.s), #2\.0(?:e\+0)?
101 ** fmulx z0\.s, p0/m, z0\.s, \1
102 ** ret
103 */
104 TEST_UNIFORM_Z (mulx_2_f32_m_tied1, svfloat32_t,
105 z0 = svmulx_n_f32_m (p0, z0, 2),
106 z0 = svmulx_m (p0, z0, 2))
107
108 /*
109 ** mulx_2_f32_m_untied: { xfail *-*-* }
110 ** fmov (z[0-9]+\.s), #2\.0(?:e\+0)?
111 ** movprfx z0, z1
112 ** fmulx z0\.s, p0/m, z0\.s, \1
113 ** ret
114 */
115 TEST_UNIFORM_Z (mulx_2_f32_m_untied, svfloat32_t,
116 z0 = svmulx_n_f32_m (p0, z1, 2),
117 z0 = svmulx_m (p0, z1, 2))
118
119 /*
120 ** mulx_f32_z_tied1:
121 ** movprfx z0\.s, p0/z, z0\.s
122 ** fmulx z0\.s, p0/m, z0\.s, z1\.s
123 ** ret
124 */
125 TEST_UNIFORM_Z (mulx_f32_z_tied1, svfloat32_t,
126 z0 = svmulx_f32_z (p0, z0, z1),
127 z0 = svmulx_z (p0, z0, z1))
128
129 /*
130 ** mulx_f32_z_tied2:
131 ** movprfx z0\.s, p0/z, z0\.s
132 ** fmulx z0\.s, p0/m, z0\.s, z1\.s
133 ** ret
134 */
135 TEST_UNIFORM_Z (mulx_f32_z_tied2, svfloat32_t,
136 z0 = svmulx_f32_z (p0, z1, z0),
137 z0 = svmulx_z (p0, z1, z0))
138
139 /*
140 ** mulx_f32_z_untied:
141 ** (
142 ** movprfx z0\.s, p0/z, z1\.s
143 ** fmulx z0\.s, p0/m, z0\.s, z2\.s
144 ** |
145 ** movprfx z0\.s, p0/z, z2\.s
146 ** fmulx z0\.s, p0/m, z0\.s, z1\.s
147 ** )
148 ** ret
149 */
150 TEST_UNIFORM_Z (mulx_f32_z_untied, svfloat32_t,
151 z0 = svmulx_f32_z (p0, z1, z2),
152 z0 = svmulx_z (p0, z1, z2))
153
154 /*
155 ** mulx_s4_f32_z_tied1:
156 ** mov (z[0-9]+\.s), s4
157 ** movprfx z0\.s, p0/z, z0\.s
158 ** fmulx z0\.s, p0/m, z0\.s, \1
159 ** ret
160 */
161 TEST_UNIFORM_ZD (mulx_s4_f32_z_tied1, svfloat32_t, float,
162 z0 = svmulx_n_f32_z (p0, z0, d4),
163 z0 = svmulx_z (p0, z0, d4))
164
165 /*
166 ** mulx_s4_f32_z_untied:
167 ** mov (z[0-9]+\.s), s4
168 ** (
169 ** movprfx z0\.s, p0/z, z1\.s
170 ** fmulx z0\.s, p0/m, z0\.s, \1
171 ** |
172 ** movprfx z0\.s, p0/z, \1
173 ** fmulx z0\.s, p0/m, z0\.s, z1\.s
174 ** )
175 ** ret
176 */
177 TEST_UNIFORM_ZD (mulx_s4_f32_z_untied, svfloat32_t, float,
178 z0 = svmulx_n_f32_z (p0, z1, d4),
179 z0 = svmulx_z (p0, z1, d4))
180
181 /*
182 ** mulx_1_f32_z_tied1:
183 ** fmov (z[0-9]+\.s), #1\.0(?:e\+0)?
184 ** movprfx z0\.s, p0/z, z0\.s
185 ** fmulx z0\.s, p0/m, z0\.s, \1
186 ** ret
187 */
188 TEST_UNIFORM_Z (mulx_1_f32_z_tied1, svfloat32_t,
189 z0 = svmulx_n_f32_z (p0, z0, 1),
190 z0 = svmulx_z (p0, z0, 1))
191
192 /*
193 ** mulx_1_f32_z_untied:
194 ** fmov (z[0-9]+\.s), #1\.0(?:e\+0)?
195 ** (
196 ** movprfx z0\.s, p0/z, z1\.s
197 ** fmulx z0\.s, p0/m, z0\.s, \1
198 ** |
199 ** movprfx z0\.s, p0/z, \1
200 ** fmulx z0\.s, p0/m, z0\.s, z1\.s
201 ** )
202 ** ret
203 */
204 TEST_UNIFORM_Z (mulx_1_f32_z_untied, svfloat32_t,
205 z0 = svmulx_n_f32_z (p0, z1, 1),
206 z0 = svmulx_z (p0, z1, 1))
207
208 /*
209 ** mulx_0p5_f32_z_tied1:
210 ** fmov (z[0-9]+\.s), #(?:0\.5|5\.0e-1)
211 ** movprfx z0\.s, p0/z, z0\.s
212 ** fmulx z0\.s, p0/m, z0\.s, \1
213 ** ret
214 */
215 TEST_UNIFORM_Z (mulx_0p5_f32_z_tied1, svfloat32_t,
216 z0 = svmulx_n_f32_z (p0, z0, 0.5),
217 z0 = svmulx_z (p0, z0, 0.5))
218
219 /*
220 ** mulx_0p5_f32_z_untied:
221 ** fmov (z[0-9]+\.s), #(?:0\.5|5\.0e-1)
222 ** (
223 ** movprfx z0\.s, p0/z, z1\.s
224 ** fmulx z0\.s, p0/m, z0\.s, \1
225 ** |
226 ** movprfx z0\.s, p0/z, \1
227 ** fmulx z0\.s, p0/m, z0\.s, z1\.s
228 ** )
229 ** ret
230 */
231 TEST_UNIFORM_Z (mulx_0p5_f32_z_untied, svfloat32_t,
232 z0 = svmulx_n_f32_z (p0, z1, 0.5),
233 z0 = svmulx_z (p0, z1, 0.5))
234
235 /*
236 ** mulx_2_f32_z_tied1:
237 ** fmov (z[0-9]+\.s), #2\.0(?:e\+0)?
238 ** movprfx z0\.s, p0/z, z0\.s
239 ** fmulx z0\.s, p0/m, z0\.s, \1
240 ** ret
241 */
242 TEST_UNIFORM_Z (mulx_2_f32_z_tied1, svfloat32_t,
243 z0 = svmulx_n_f32_z (p0, z0, 2),
244 z0 = svmulx_z (p0, z0, 2))
245
246 /*
247 ** mulx_2_f32_z_untied:
248 ** fmov (z[0-9]+\.s), #2\.0(?:e\+0)?
249 ** (
250 ** movprfx z0\.s, p0/z, z1\.s
251 ** fmulx z0\.s, p0/m, z0\.s, \1
252 ** |
253 ** movprfx z0\.s, p0/z, \1
254 ** fmulx z0\.s, p0/m, z0\.s, z1\.s
255 ** )
256 ** ret
257 */
258 TEST_UNIFORM_Z (mulx_2_f32_z_untied, svfloat32_t,
259 z0 = svmulx_n_f32_z (p0, z1, 2),
260 z0 = svmulx_z (p0, z1, 2))
261
262 /*
263 ** mulx_f32_x_tied1:
264 ** fmulx z0\.s, p0/m, z0\.s, z1\.s
265 ** ret
266 */
267 TEST_UNIFORM_Z (mulx_f32_x_tied1, svfloat32_t,
268 z0 = svmulx_f32_x (p0, z0, z1),
269 z0 = svmulx_x (p0, z0, z1))
270
271 /*
272 ** mulx_f32_x_tied2:
273 ** fmulx z0\.s, p0/m, z0\.s, z1\.s
274 ** ret
275 */
276 TEST_UNIFORM_Z (mulx_f32_x_tied2, svfloat32_t,
277 z0 = svmulx_f32_x (p0, z1, z0),
278 z0 = svmulx_x (p0, z1, z0))
279
280 /*
281 ** mulx_f32_x_untied:
282 ** (
283 ** movprfx z0, z1
284 ** fmulx z0\.s, p0/m, z0\.s, z2\.s
285 ** |
286 ** movprfx z0, z2
287 ** fmulx z0\.s, p0/m, z0\.s, z1\.s
288 ** )
289 ** ret
290 */
291 TEST_UNIFORM_Z (mulx_f32_x_untied, svfloat32_t,
292 z0 = svmulx_f32_x (p0, z1, z2),
293 z0 = svmulx_x (p0, z1, z2))
294
295 /*
296 ** mulx_s4_f32_x_tied1:
297 ** mov (z[0-9]+\.s), s4
298 ** fmulx z0\.s, p0/m, z0\.s, \1
299 ** ret
300 */
301 TEST_UNIFORM_ZD (mulx_s4_f32_x_tied1, svfloat32_t, float,
302 z0 = svmulx_n_f32_x (p0, z0, d4),
303 z0 = svmulx_x (p0, z0, d4))
304
305 /*
306 ** mulx_s4_f32_x_untied:
307 ** mov z0\.s, s4
308 ** fmulx z0\.s, p0/m, z0\.s, z1\.s
309 ** ret
310 */
311 TEST_UNIFORM_ZD (mulx_s4_f32_x_untied, svfloat32_t, float,
312 z0 = svmulx_n_f32_x (p0, z1, d4),
313 z0 = svmulx_x (p0, z1, d4))
314
315 /*
316 ** mulx_1_f32_x_tied1:
317 ** fmov (z[0-9]+\.s), #1\.0(?:e\+0)?
318 ** fmulx z0\.s, p0/m, z0\.s, \1
319 ** ret
320 */
321 TEST_UNIFORM_Z (mulx_1_f32_x_tied1, svfloat32_t,
322 z0 = svmulx_n_f32_x (p0, z0, 1),
323 z0 = svmulx_x (p0, z0, 1))
324
325 /*
326 ** mulx_1_f32_x_untied:
327 ** fmov z0\.s, #1\.0(?:e\+0)?
328 ** fmulx z0\.s, p0/m, z0\.s, z1\.s
329 ** ret
330 */
331 TEST_UNIFORM_Z (mulx_1_f32_x_untied, svfloat32_t,
332 z0 = svmulx_n_f32_x (p0, z1, 1),
333 z0 = svmulx_x (p0, z1, 1))
334
335 /*
336 ** mulx_0p5_f32_x_tied1:
337 ** fmov (z[0-9]+\.s), #(?:0\.5|5\.0e-1)
338 ** fmulx z0\.s, p0/m, z0\.s, \1
339 ** ret
340 */
341 TEST_UNIFORM_Z (mulx_0p5_f32_x_tied1, svfloat32_t,
342 z0 = svmulx_n_f32_x (p0, z0, 0.5),
343 z0 = svmulx_x (p0, z0, 0.5))
344
345 /*
346 ** mulx_0p5_f32_x_untied:
347 ** fmov z0\.s, #(?:0\.5|5\.0e-1)
348 ** fmulx z0\.s, p0/m, z0\.s, z1\.s
349 ** ret
350 */
351 TEST_UNIFORM_Z (mulx_0p5_f32_x_untied, svfloat32_t,
352 z0 = svmulx_n_f32_x (p0, z1, 0.5),
353 z0 = svmulx_x (p0, z1, 0.5))
354
355 /*
356 ** mulx_2_f32_x_tied1:
357 ** fmov (z[0-9]+\.s), #2\.0(?:e\+0)?
358 ** fmulx z0\.s, p0/m, z0\.s, \1
359 ** ret
360 */
361 TEST_UNIFORM_Z (mulx_2_f32_x_tied1, svfloat32_t,
362 z0 = svmulx_n_f32_x (p0, z0, 2),
363 z0 = svmulx_x (p0, z0, 2))
364
365 /*
366 ** mulx_2_f32_x_untied:
367 ** fmov z0\.s, #2\.0(?:e\+0)?
368 ** fmulx z0\.s, p0/m, z0\.s, z1\.s
369 ** ret
370 */
371 TEST_UNIFORM_Z (mulx_2_f32_x_untied, svfloat32_t,
372 z0 = svmulx_n_f32_x (p0, z1, 2),
373 z0 = svmulx_x (p0, z1, 2))
374
375 /*
376 ** ptrue_mulx_f32_x_tied1:
377 ** ...
378 ** ptrue p[0-9]+\.b[^\n]*
379 ** ...
380 ** ret
381 */
382 TEST_UNIFORM_Z (ptrue_mulx_f32_x_tied1, svfloat32_t,
383 z0 = svmulx_f32_x (svptrue_b32 (), z0, z1),
384 z0 = svmulx_x (svptrue_b32 (), z0, z1))
385
386 /*
387 ** ptrue_mulx_f32_x_tied2:
388 ** ...
389 ** ptrue p[0-9]+\.b[^\n]*
390 ** ...
391 ** ret
392 */
393 TEST_UNIFORM_Z (ptrue_mulx_f32_x_tied2, svfloat32_t,
394 z0 = svmulx_f32_x (svptrue_b32 (), z1, z0),
395 z0 = svmulx_x (svptrue_b32 (), z1, z0))
396
397 /*
398 ** ptrue_mulx_f32_x_untied:
399 ** ...
400 ** ptrue p[0-9]+\.b[^\n]*
401 ** ...
402 ** ret
403 */
404 TEST_UNIFORM_Z (ptrue_mulx_f32_x_untied, svfloat32_t,
405 z0 = svmulx_f32_x (svptrue_b32 (), z1, z2),
406 z0 = svmulx_x (svptrue_b32 (), z1, z2))
407
408 /*
409 ** ptrue_mulx_1_f32_x_tied1:
410 ** ...
411 ** ptrue p[0-9]+\.b[^\n]*
412 ** ...
413 ** ret
414 */
415 TEST_UNIFORM_Z (ptrue_mulx_1_f32_x_tied1, svfloat32_t,
416 z0 = svmulx_n_f32_x (svptrue_b32 (), z0, 1),
417 z0 = svmulx_x (svptrue_b32 (), z0, 1))
418
419 /*
420 ** ptrue_mulx_1_f32_x_untied:
421 ** ...
422 ** ptrue p[0-9]+\.b[^\n]*
423 ** ...
424 ** ret
425 */
426 TEST_UNIFORM_Z (ptrue_mulx_1_f32_x_untied, svfloat32_t,
427 z0 = svmulx_n_f32_x (svptrue_b32 (), z1, 1),
428 z0 = svmulx_x (svptrue_b32 (), z1, 1))
429
430 /*
431 ** ptrue_mulx_0p5_f32_x_tied1:
432 ** ...
433 ** ptrue p[0-9]+\.b[^\n]*
434 ** ...
435 ** ret
436 */
437 TEST_UNIFORM_Z (ptrue_mulx_0p5_f32_x_tied1, svfloat32_t,
438 z0 = svmulx_n_f32_x (svptrue_b32 (), z0, 0.5),
439 z0 = svmulx_x (svptrue_b32 (), z0, 0.5))
440
441 /*
442 ** ptrue_mulx_0p5_f32_x_untied:
443 ** ...
444 ** ptrue p[0-9]+\.b[^\n]*
445 ** ...
446 ** ret
447 */
448 TEST_UNIFORM_Z (ptrue_mulx_0p5_f32_x_untied, svfloat32_t,
449 z0 = svmulx_n_f32_x (svptrue_b32 (), z1, 0.5),
450 z0 = svmulx_x (svptrue_b32 (), z1, 0.5))
451
452 /*
453 ** ptrue_mulx_2_f32_x_tied1:
454 ** ...
455 ** ptrue p[0-9]+\.b[^\n]*
456 ** ...
457 ** ret
458 */
459 TEST_UNIFORM_Z (ptrue_mulx_2_f32_x_tied1, svfloat32_t,
460 z0 = svmulx_n_f32_x (svptrue_b32 (), z0, 2),
461 z0 = svmulx_x (svptrue_b32 (), z0, 2))
462
463 /*
464 ** ptrue_mulx_2_f32_x_untied:
465 ** ...
466 ** ptrue p[0-9]+\.b[^\n]*
467 ** ...
468 ** ret
469 */
470 TEST_UNIFORM_Z (ptrue_mulx_2_f32_x_untied, svfloat32_t,
471 z0 = svmulx_n_f32_x (svptrue_b32 (), z1, 2),
472 z0 = svmulx_x (svptrue_b32 (), z1, 2))