1 /* { dg-additional-options "-fno-trapping-math" } */
2 /* { dg-final { check-function-bodies "**" "" "-DCHECK_ASM" } } */
3
4 #include "test_sve_acle.h"
5
6 /*
7 ** mul_f32_m_tied1:
8 ** fmul z0\.s, p0/m, z0\.s, z1\.s
9 ** ret
10 */
11 TEST_UNIFORM_Z (mul_f32_m_tied1, svfloat32_t,
12 z0 = svmul_f32_m (p0, z0, z1),
13 z0 = svmul_m (p0, z0, z1))
14
15 /*
16 ** mul_f32_m_tied2:
17 ** mov (z[0-9]+)\.d, z0\.d
18 ** movprfx z0, z1
19 ** fmul z0\.s, p0/m, z0\.s, \1\.s
20 ** ret
21 */
22 TEST_UNIFORM_Z (mul_f32_m_tied2, svfloat32_t,
23 z0 = svmul_f32_m (p0, z1, z0),
24 z0 = svmul_m (p0, z1, z0))
25
26 /*
27 ** mul_f32_m_untied:
28 ** movprfx z0, z1
29 ** fmul z0\.s, p0/m, z0\.s, z2\.s
30 ** ret
31 */
32 TEST_UNIFORM_Z (mul_f32_m_untied, svfloat32_t,
33 z0 = svmul_f32_m (p0, z1, z2),
34 z0 = svmul_m (p0, z1, z2))
35
36 /*
37 ** mul_s4_f32_m_tied1:
38 ** mov (z[0-9]+\.s), s4
39 ** fmul z0\.s, p0/m, z0\.s, \1
40 ** ret
41 */
42 TEST_UNIFORM_ZD (mul_s4_f32_m_tied1, svfloat32_t, float,
43 z0 = svmul_n_f32_m (p0, z0, d4),
44 z0 = svmul_m (p0, z0, d4))
45
46 /*
47 ** mul_s4_f32_m_untied:
48 ** mov (z[0-9]+\.s), s4
49 ** movprfx z0, z1
50 ** fmul z0\.s, p0/m, z0\.s, \1
51 ** ret
52 */
53 TEST_UNIFORM_ZD (mul_s4_f32_m_untied, svfloat32_t, float,
54 z0 = svmul_n_f32_m (p0, z1, d4),
55 z0 = svmul_m (p0, z1, d4))
56
57 /*
58 ** mul_1_f32_m_tied1:
59 ** fmov (z[0-9]+\.s), #1\.0(?:e\+0)?
60 ** fmul z0\.s, p0/m, z0\.s, \1
61 ** ret
62 */
63 TEST_UNIFORM_Z (mul_1_f32_m_tied1, svfloat32_t,
64 z0 = svmul_n_f32_m (p0, z0, 1),
65 z0 = svmul_m (p0, z0, 1))
66
67 /*
68 ** mul_1_f32_m_untied: { xfail *-*-* }
69 ** fmov (z[0-9]+\.s), #1\.0(?:e\+0)?
70 ** movprfx z0, z1
71 ** fmul z0\.s, p0/m, z0\.s, \1
72 ** ret
73 */
74 TEST_UNIFORM_Z (mul_1_f32_m_untied, svfloat32_t,
75 z0 = svmul_n_f32_m (p0, z1, 1),
76 z0 = svmul_m (p0, z1, 1))
77
78 /*
79 ** mul_0p5_f32_m_tied1:
80 ** fmul z0\.s, p0/m, z0\.s, #0\.5
81 ** ret
82 */
83 TEST_UNIFORM_Z (mul_0p5_f32_m_tied1, svfloat32_t,
84 z0 = svmul_n_f32_m (p0, z0, 0.5),
85 z0 = svmul_m (p0, z0, 0.5))
86
87 /*
88 ** mul_0p5_f32_m_untied:
89 ** movprfx z0, z1
90 ** fmul z0\.s, p0/m, z0\.s, #0\.5
91 ** ret
92 */
93 TEST_UNIFORM_Z (mul_0p5_f32_m_untied, svfloat32_t,
94 z0 = svmul_n_f32_m (p0, z1, 0.5),
95 z0 = svmul_m (p0, z1, 0.5))
96
97 /*
98 ** mul_2_f32_m_tied1:
99 ** fmul z0\.s, p0/m, z0\.s, #2\.0
100 ** ret
101 */
102 TEST_UNIFORM_Z (mul_2_f32_m_tied1, svfloat32_t,
103 z0 = svmul_n_f32_m (p0, z0, 2),
104 z0 = svmul_m (p0, z0, 2))
105
106 /*
107 ** mul_2_f32_m_untied:
108 ** movprfx z0, z1
109 ** fmul z0\.s, p0/m, z0\.s, #2\.0
110 ** ret
111 */
112 TEST_UNIFORM_Z (mul_2_f32_m_untied, svfloat32_t,
113 z0 = svmul_n_f32_m (p0, z1, 2),
114 z0 = svmul_m (p0, z1, 2))
115
116 /*
117 ** mul_f32_z_tied1:
118 ** movprfx z0\.s, p0/z, z0\.s
119 ** fmul z0\.s, p0/m, z0\.s, z1\.s
120 ** ret
121 */
122 TEST_UNIFORM_Z (mul_f32_z_tied1, svfloat32_t,
123 z0 = svmul_f32_z (p0, z0, z1),
124 z0 = svmul_z (p0, z0, z1))
125
126 /*
127 ** mul_f32_z_tied2:
128 ** movprfx z0\.s, p0/z, z0\.s
129 ** fmul z0\.s, p0/m, z0\.s, z1\.s
130 ** ret
131 */
132 TEST_UNIFORM_Z (mul_f32_z_tied2, svfloat32_t,
133 z0 = svmul_f32_z (p0, z1, z0),
134 z0 = svmul_z (p0, z1, z0))
135
136 /*
137 ** mul_f32_z_untied:
138 ** (
139 ** movprfx z0\.s, p0/z, z1\.s
140 ** fmul z0\.s, p0/m, z0\.s, z2\.s
141 ** |
142 ** movprfx z0\.s, p0/z, z2\.s
143 ** fmul z0\.s, p0/m, z0\.s, z1\.s
144 ** )
145 ** ret
146 */
147 TEST_UNIFORM_Z (mul_f32_z_untied, svfloat32_t,
148 z0 = svmul_f32_z (p0, z1, z2),
149 z0 = svmul_z (p0, z1, z2))
150
151 /*
152 ** mul_s4_f32_z_tied1:
153 ** mov (z[0-9]+\.s), s4
154 ** movprfx z0\.s, p0/z, z0\.s
155 ** fmul z0\.s, p0/m, z0\.s, \1
156 ** ret
157 */
158 TEST_UNIFORM_ZD (mul_s4_f32_z_tied1, svfloat32_t, float,
159 z0 = svmul_n_f32_z (p0, z0, d4),
160 z0 = svmul_z (p0, z0, d4))
161
162 /*
163 ** mul_s4_f32_z_untied:
164 ** mov (z[0-9]+\.s), s4
165 ** (
166 ** movprfx z0\.s, p0/z, z1\.s
167 ** fmul z0\.s, p0/m, z0\.s, \1
168 ** |
169 ** movprfx z0\.s, p0/z, \1
170 ** fmul z0\.s, p0/m, z0\.s, z1\.s
171 ** )
172 ** ret
173 */
174 TEST_UNIFORM_ZD (mul_s4_f32_z_untied, svfloat32_t, float,
175 z0 = svmul_n_f32_z (p0, z1, d4),
176 z0 = svmul_z (p0, z1, d4))
177
178 /*
179 ** mul_1_f32_z_tied1:
180 ** fmov (z[0-9]+\.s), #1\.0(?:e\+0)?
181 ** movprfx z0\.s, p0/z, z0\.s
182 ** fmul z0\.s, p0/m, z0\.s, \1
183 ** ret
184 */
185 TEST_UNIFORM_Z (mul_1_f32_z_tied1, svfloat32_t,
186 z0 = svmul_n_f32_z (p0, z0, 1),
187 z0 = svmul_z (p0, z0, 1))
188
189 /*
190 ** mul_1_f32_z_untied:
191 ** fmov (z[0-9]+\.s), #1\.0(?:e\+0)?
192 ** (
193 ** movprfx z0\.s, p0/z, z1\.s
194 ** fmul z0\.s, p0/m, z0\.s, \1
195 ** |
196 ** movprfx z0\.s, p0/z, \1
197 ** fmul z0\.s, p0/m, z0\.s, z1\.s
198 ** )
199 ** ret
200 */
201 TEST_UNIFORM_Z (mul_1_f32_z_untied, svfloat32_t,
202 z0 = svmul_n_f32_z (p0, z1, 1),
203 z0 = svmul_z (p0, z1, 1))
204
205 /*
206 ** mul_0p5_f32_z_tied1:
207 ** movprfx z0\.s, p0/z, z0\.s
208 ** fmul z0\.s, p0/m, z0\.s, #0\.5
209 ** ret
210 */
211 TEST_UNIFORM_Z (mul_0p5_f32_z_tied1, svfloat32_t,
212 z0 = svmul_n_f32_z (p0, z0, 0.5),
213 z0 = svmul_z (p0, z0, 0.5))
214
215 /*
216 ** mul_0p5_f32_z_untied:
217 ** movprfx z0\.s, p0/z, z1\.s
218 ** fmul z0\.s, p0/m, z0\.s, #0\.5
219 ** ret
220 */
221 TEST_UNIFORM_Z (mul_0p5_f32_z_untied, svfloat32_t,
222 z0 = svmul_n_f32_z (p0, z1, 0.5),
223 z0 = svmul_z (p0, z1, 0.5))
224
225 /*
226 ** mul_2_f32_z_tied1:
227 ** movprfx z0\.s, p0/z, z0\.s
228 ** fmul z0\.s, p0/m, z0\.s, #2\.0
229 ** ret
230 */
231 TEST_UNIFORM_Z (mul_2_f32_z_tied1, svfloat32_t,
232 z0 = svmul_n_f32_z (p0, z0, 2),
233 z0 = svmul_z (p0, z0, 2))
234
235 /*
236 ** mul_2_f32_z_untied:
237 ** movprfx z0\.s, p0/z, z1\.s
238 ** fmul z0\.s, p0/m, z0\.s, #2\.0
239 ** ret
240 */
241 TEST_UNIFORM_Z (mul_2_f32_z_untied, svfloat32_t,
242 z0 = svmul_n_f32_z (p0, z1, 2),
243 z0 = svmul_z (p0, z1, 2))
244
245 /*
246 ** mul_f32_x_tied1:
247 ** fmul z0\.s, (z0\.s, z1\.s|z1\.s, z0\.s)
248 ** ret
249 */
250 TEST_UNIFORM_Z (mul_f32_x_tied1, svfloat32_t,
251 z0 = svmul_f32_x (p0, z0, z1),
252 z0 = svmul_x (p0, z0, z1))
253
254 /*
255 ** mul_f32_x_tied2:
256 ** fmul z0\.s, (z0\.s, z1\.s|z1\.s, z0\.s)
257 ** ret
258 */
259 TEST_UNIFORM_Z (mul_f32_x_tied2, svfloat32_t,
260 z0 = svmul_f32_x (p0, z1, z0),
261 z0 = svmul_x (p0, z1, z0))
262
263 /*
264 ** mul_f32_x_untied:
265 ** fmul z0\.s, (z1\.s, z2\.s|z2\.s, z1\.s)
266 ** ret
267 */
268 TEST_UNIFORM_Z (mul_f32_x_untied, svfloat32_t,
269 z0 = svmul_f32_x (p0, z1, z2),
270 z0 = svmul_x (p0, z1, z2))
271
272 /*
273 ** mul_s4_f32_x_tied1:
274 ** mov (z[0-9]+\.s), s4
275 ** fmul z0\.s, (z0\.s, \1|\1, z0\.s)
276 ** ret
277 */
278 TEST_UNIFORM_ZD (mul_s4_f32_x_tied1, svfloat32_t, float,
279 z0 = svmul_n_f32_x (p0, z0, d4),
280 z0 = svmul_x (p0, z0, d4))
281
282 /*
283 ** mul_s4_f32_x_untied:
284 ** mov (z[0-9]+\.s), s4
285 ** fmul z0\.s, (z1\.s, \1|\1, z1\.s)
286 ** ret
287 */
288 TEST_UNIFORM_ZD (mul_s4_f32_x_untied, svfloat32_t, float,
289 z0 = svmul_n_f32_x (p0, z1, d4),
290 z0 = svmul_x (p0, z1, d4))
291
292 /*
293 ** mul_1_f32_x_tied1:
294 ** fmov (z[0-9]+\.s), #1\.0(?:e\+0)?
295 ** fmul z0\.s, (z0\.s, \1|\1, z0\.s)
296 ** ret
297 */
298 TEST_UNIFORM_Z (mul_1_f32_x_tied1, svfloat32_t,
299 z0 = svmul_n_f32_x (p0, z0, 1),
300 z0 = svmul_x (p0, z0, 1))
301
302 /*
303 ** mul_1_f32_x_untied:
304 ** fmov (z[0-9]+\.s), #1\.0(?:e\+0)?
305 ** fmul z0\.s, (z1\.s, \1|\1, z1\.s)
306 ** ret
307 */
308 TEST_UNIFORM_Z (mul_1_f32_x_untied, svfloat32_t,
309 z0 = svmul_n_f32_x (p0, z1, 1),
310 z0 = svmul_x (p0, z1, 1))
311
312 /*
313 ** mul_0p5_f32_x_tied1:
314 ** fmul z0\.s, p0/m, z0\.s, #0\.5
315 ** ret
316 */
317 TEST_UNIFORM_Z (mul_0p5_f32_x_tied1, svfloat32_t,
318 z0 = svmul_n_f32_x (p0, z0, 0.5),
319 z0 = svmul_x (p0, z0, 0.5))
320
321 /*
322 ** mul_0p5_f32_x_untied:
323 ** movprfx z0, z1
324 ** fmul z0\.s, p0/m, z0\.s, #0\.5
325 ** ret
326 */
327 TEST_UNIFORM_Z (mul_0p5_f32_x_untied, svfloat32_t,
328 z0 = svmul_n_f32_x (p0, z1, 0.5),
329 z0 = svmul_x (p0, z1, 0.5))
330
331 /*
332 ** mul_2_f32_x_tied1:
333 ** fmul z0\.s, p0/m, z0\.s, #2\.0
334 ** ret
335 */
336 TEST_UNIFORM_Z (mul_2_f32_x_tied1, svfloat32_t,
337 z0 = svmul_n_f32_x (p0, z0, 2),
338 z0 = svmul_x (p0, z0, 2))
339
340 /*
341 ** mul_2_f32_x_untied:
342 ** movprfx z0, z1
343 ** fmul z0\.s, p0/m, z0\.s, #2\.0
344 ** ret
345 */
346 TEST_UNIFORM_Z (mul_2_f32_x_untied, svfloat32_t,
347 z0 = svmul_n_f32_x (p0, z1, 2),
348 z0 = svmul_x (p0, z1, 2))
349
350 /*
351 ** ptrue_mul_f32_x_tied1:
352 ** fmul z0\.s, (z0\.s, z1\.s|z1\.s, z0\.s)
353 ** ret
354 */
355 TEST_UNIFORM_Z (ptrue_mul_f32_x_tied1, svfloat32_t,
356 z0 = svmul_f32_x (svptrue_b32 (), z0, z1),
357 z0 = svmul_x (svptrue_b32 (), z0, z1))
358
359 /*
360 ** ptrue_mul_f32_x_tied2:
361 ** fmul z0\.s, (z0\.s, z1\.s|z1\.s, z0\.s)
362 ** ret
363 */
364 TEST_UNIFORM_Z (ptrue_mul_f32_x_tied2, svfloat32_t,
365 z0 = svmul_f32_x (svptrue_b32 (), z1, z0),
366 z0 = svmul_x (svptrue_b32 (), z1, z0))
367
368 /*
369 ** ptrue_mul_f32_x_untied:
370 ** fmul z0\.s, (z1\.s, z2\.s|z2\.s, z1\.s)
371 ** ret
372 */
373 TEST_UNIFORM_Z (ptrue_mul_f32_x_untied, svfloat32_t,
374 z0 = svmul_f32_x (svptrue_b32 (), z1, z2),
375 z0 = svmul_x (svptrue_b32 (), z1, z2))
376
377 /*
378 ** ptrue_mul_1_f32_x_tied1:
379 ** fmov (z[0-9]+\.s), #1\.0(?:e\+0)?
380 ** fmul z0\.s, (z0\.s, \1|\1, z0\.s)
381 ** ret
382 */
383 TEST_UNIFORM_Z (ptrue_mul_1_f32_x_tied1, svfloat32_t,
384 z0 = svmul_n_f32_x (svptrue_b32 (), z0, 1),
385 z0 = svmul_x (svptrue_b32 (), z0, 1))
386
387 /*
388 ** ptrue_mul_1_f32_x_untied:
389 ** fmov (z[0-9]+\.s), #1\.0(?:e\+0)?
390 ** fmul z0\.s, (z1\.s, \1|\1, z1\.s)
391 ** ret
392 */
393 TEST_UNIFORM_Z (ptrue_mul_1_f32_x_untied, svfloat32_t,
394 z0 = svmul_n_f32_x (svptrue_b32 (), z1, 1),
395 z0 = svmul_x (svptrue_b32 (), z1, 1))
396
397 /*
398 ** ptrue_mul_0p5_f32_x_tied1:
399 ** ...
400 ** ptrue p[0-9]+\.b[^\n]*
401 ** ...
402 ** ret
403 */
404 TEST_UNIFORM_Z (ptrue_mul_0p5_f32_x_tied1, svfloat32_t,
405 z0 = svmul_n_f32_x (svptrue_b32 (), z0, 0.5),
406 z0 = svmul_x (svptrue_b32 (), z0, 0.5))
407
408 /*
409 ** ptrue_mul_0p5_f32_x_untied:
410 ** ...
411 ** ptrue p[0-9]+\.b[^\n]*
412 ** ...
413 ** ret
414 */
415 TEST_UNIFORM_Z (ptrue_mul_0p5_f32_x_untied, svfloat32_t,
416 z0 = svmul_n_f32_x (svptrue_b32 (), z1, 0.5),
417 z0 = svmul_x (svptrue_b32 (), z1, 0.5))
418
419 /*
420 ** ptrue_mul_2_f32_x_tied1:
421 ** ...
422 ** ptrue p[0-9]+\.b[^\n]*
423 ** ...
424 ** ret
425 */
426 TEST_UNIFORM_Z (ptrue_mul_2_f32_x_tied1, svfloat32_t,
427 z0 = svmul_n_f32_x (svptrue_b32 (), z0, 2),
428 z0 = svmul_x (svptrue_b32 (), z0, 2))
429
430 /*
431 ** ptrue_mul_2_f32_x_untied:
432 ** ...
433 ** ptrue p[0-9]+\.b[^\n]*
434 ** ...
435 ** ret
436 */
437 TEST_UNIFORM_Z (ptrue_mul_2_f32_x_untied, svfloat32_t,
438 z0 = svmul_n_f32_x (svptrue_b32 (), z1, 2),
439 z0 = svmul_x (svptrue_b32 (), z1, 2))