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 ** subr_f64_m_tied1:
8 ** fsubr z0\.d, p0/m, z0\.d, z1\.d
9 ** ret
10 */
11 TEST_UNIFORM_Z (subr_f64_m_tied1, svfloat64_t,
12 z0 = svsubr_f64_m (p0, z0, z1),
13 z0 = svsubr_m (p0, z0, z1))
14
15 /*
16 ** subr_f64_m_tied2:
17 ** mov (z[0-9]+\.d), z0\.d
18 ** movprfx z0, z1
19 ** fsubr z0\.d, p0/m, z0\.d, \1
20 ** ret
21 */
22 TEST_UNIFORM_Z (subr_f64_m_tied2, svfloat64_t,
23 z0 = svsubr_f64_m (p0, z1, z0),
24 z0 = svsubr_m (p0, z1, z0))
25
26 /*
27 ** subr_f64_m_untied:
28 ** movprfx z0, z1
29 ** fsubr z0\.d, p0/m, z0\.d, z2\.d
30 ** ret
31 */
32 TEST_UNIFORM_Z (subr_f64_m_untied, svfloat64_t,
33 z0 = svsubr_f64_m (p0, z1, z2),
34 z0 = svsubr_m (p0, z1, z2))
35
36 /*
37 ** subr_d4_f64_m_tied1:
38 ** mov (z[0-9]+\.d), d4
39 ** fsubr z0\.d, p0/m, z0\.d, \1
40 ** ret
41 */
42 TEST_UNIFORM_ZD (subr_d4_f64_m_tied1, svfloat64_t, double,
43 z0 = svsubr_n_f64_m (p0, z0, d4),
44 z0 = svsubr_m (p0, z0, d4))
45
46 /*
47 ** subr_d4_f64_m_untied:
48 ** mov (z[0-9]+\.d), d4
49 ** movprfx z0, z1
50 ** fsubr z0\.d, p0/m, z0\.d, \1
51 ** ret
52 */
53 TEST_UNIFORM_ZD (subr_d4_f64_m_untied, svfloat64_t, double,
54 z0 = svsubr_n_f64_m (p0, z1, d4),
55 z0 = svsubr_m (p0, z1, d4))
56
57 /*
58 ** subr_1_f64_m_tied1:
59 ** fsubr z0\.d, p0/m, z0\.d, #1\.0
60 ** ret
61 */
62 TEST_UNIFORM_Z (subr_1_f64_m_tied1, svfloat64_t,
63 z0 = svsubr_n_f64_m (p0, z0, 1),
64 z0 = svsubr_m (p0, z0, 1))
65
66 /*
67 ** subr_1_f64_m_untied:
68 ** movprfx z0, z1
69 ** fsubr z0\.d, p0/m, z0\.d, #1\.0
70 ** ret
71 */
72 TEST_UNIFORM_Z (subr_1_f64_m_untied, svfloat64_t,
73 z0 = svsubr_n_f64_m (p0, z1, 1),
74 z0 = svsubr_m (p0, z1, 1))
75
76 /*
77 ** subr_0p5_f64_m_tied1:
78 ** fsubr z0\.d, p0/m, z0\.d, #0\.5
79 ** ret
80 */
81 TEST_UNIFORM_Z (subr_0p5_f64_m_tied1, svfloat64_t,
82 z0 = svsubr_n_f64_m (p0, z0, 0.5),
83 z0 = svsubr_m (p0, z0, 0.5))
84
85 /*
86 ** subr_0p5_f64_m_untied:
87 ** movprfx z0, z1
88 ** fsubr z0\.d, p0/m, z0\.d, #0\.5
89 ** ret
90 */
91 TEST_UNIFORM_Z (subr_0p5_f64_m_untied, svfloat64_t,
92 z0 = svsubr_n_f64_m (p0, z1, 0.5),
93 z0 = svsubr_m (p0, z1, 0.5))
94
95 /*
96 ** subr_m1_f64_m_tied1:
97 ** fmov (z[0-9]+\.d), #-1\.0(?:e\+0)?
98 ** fsubr z0\.d, p0/m, z0\.d, \1
99 ** ret
100 */
101 TEST_UNIFORM_Z (subr_m1_f64_m_tied1, svfloat64_t,
102 z0 = svsubr_n_f64_m (p0, z0, -1),
103 z0 = svsubr_m (p0, z0, -1))
104
105 /*
106 ** subr_m1_f64_m_untied: { xfail *-*-* }
107 ** fmov (z[0-9]+\.d), #-1\.0(?:e\+0)?
108 ** movprfx z0, z1
109 ** fsubr z0\.d, p0/m, z0\.d, \1
110 ** ret
111 */
112 TEST_UNIFORM_Z (subr_m1_f64_m_untied, svfloat64_t,
113 z0 = svsubr_n_f64_m (p0, z1, -1),
114 z0 = svsubr_m (p0, z1, -1))
115
116 /*
117 ** subr_f64_z_tied1:
118 ** movprfx z0\.d, p0/z, z0\.d
119 ** fsubr z0\.d, p0/m, z0\.d, z1\.d
120 ** ret
121 */
122 TEST_UNIFORM_Z (subr_f64_z_tied1, svfloat64_t,
123 z0 = svsubr_f64_z (p0, z0, z1),
124 z0 = svsubr_z (p0, z0, z1))
125
126 /*
127 ** subr_f64_z_tied2:
128 ** movprfx z0\.d, p0/z, z0\.d
129 ** fsub z0\.d, p0/m, z0\.d, z1\.d
130 ** ret
131 */
132 TEST_UNIFORM_Z (subr_f64_z_tied2, svfloat64_t,
133 z0 = svsubr_f64_z (p0, z1, z0),
134 z0 = svsubr_z (p0, z1, z0))
135
136 /*
137 ** subr_f64_z_untied:
138 ** (
139 ** movprfx z0\.d, p0/z, z1\.d
140 ** fsubr z0\.d, p0/m, z0\.d, z2\.d
141 ** |
142 ** movprfx z0\.d, p0/z, z2\.d
143 ** fsub z0\.d, p0/m, z0\.d, z1\.d
144 ** )
145 ** ret
146 */
147 TEST_UNIFORM_Z (subr_f64_z_untied, svfloat64_t,
148 z0 = svsubr_f64_z (p0, z1, z2),
149 z0 = svsubr_z (p0, z1, z2))
150
151 /*
152 ** subr_d4_f64_z_tied1:
153 ** mov (z[0-9]+\.d), d4
154 ** movprfx z0\.d, p0/z, z0\.d
155 ** fsubr z0\.d, p0/m, z0\.d, \1
156 ** ret
157 */
158 TEST_UNIFORM_ZD (subr_d4_f64_z_tied1, svfloat64_t, double,
159 z0 = svsubr_n_f64_z (p0, z0, d4),
160 z0 = svsubr_z (p0, z0, d4))
161
162 /*
163 ** subr_d4_f64_z_untied:
164 ** mov (z[0-9]+\.d), d4
165 ** (
166 ** movprfx z0\.d, p0/z, z1\.d
167 ** fsubr z0\.d, p0/m, z0\.d, \1
168 ** |
169 ** movprfx z0\.d, p0/z, \1
170 ** fsub z0\.d, p0/m, z0\.d, z1\.d
171 ** )
172 ** ret
173 */
174 TEST_UNIFORM_ZD (subr_d4_f64_z_untied, svfloat64_t, double,
175 z0 = svsubr_n_f64_z (p0, z1, d4),
176 z0 = svsubr_z (p0, z1, d4))
177
178 /*
179 ** subr_1_f64_z_tied1:
180 ** movprfx z0\.d, p0/z, z0\.d
181 ** fsubr z0\.d, p0/m, z0\.d, #1\.0
182 ** ret
183 */
184 TEST_UNIFORM_Z (subr_1_f64_z_tied1, svfloat64_t,
185 z0 = svsubr_n_f64_z (p0, z0, 1),
186 z0 = svsubr_z (p0, z0, 1))
187
188 /*
189 ** subr_1_f64_z_untied:
190 ** movprfx z0\.d, p0/z, z1\.d
191 ** fsubr z0\.d, p0/m, z0\.d, #1\.0
192 ** ret
193 */
194 TEST_UNIFORM_Z (subr_1_f64_z_untied, svfloat64_t,
195 z0 = svsubr_n_f64_z (p0, z1, 1),
196 z0 = svsubr_z (p0, z1, 1))
197
198 /*
199 ** subr_0p5_f64_z_tied1:
200 ** movprfx z0\.d, p0/z, z0\.d
201 ** fsubr z0\.d, p0/m, z0\.d, #0\.5
202 ** ret
203 */
204 TEST_UNIFORM_Z (subr_0p5_f64_z_tied1, svfloat64_t,
205 z0 = svsubr_n_f64_z (p0, z0, 0.5),
206 z0 = svsubr_z (p0, z0, 0.5))
207
208 /*
209 ** subr_0p5_f64_z_untied:
210 ** movprfx z0\.d, p0/z, z1\.d
211 ** fsubr z0\.d, p0/m, z0\.d, #0\.5
212 ** ret
213 */
214 TEST_UNIFORM_Z (subr_0p5_f64_z_untied, svfloat64_t,
215 z0 = svsubr_n_f64_z (p0, z1, 0.5),
216 z0 = svsubr_z (p0, z1, 0.5))
217
218 /*
219 ** subr_m1_f64_z_tied1:
220 ** fmov (z[0-9]+\.d), #-1\.0(?:e\+0)?
221 ** movprfx z0\.d, p0/z, z0\.d
222 ** fsubr z0\.d, p0/m, z0\.d, \1
223 ** ret
224 */
225 TEST_UNIFORM_Z (subr_m1_f64_z_tied1, svfloat64_t,
226 z0 = svsubr_n_f64_z (p0, z0, -1),
227 z0 = svsubr_z (p0, z0, -1))
228
229 /*
230 ** subr_m1_f64_z_untied:
231 ** fmov (z[0-9]+\.d), #-1\.0(?:e\+0)?
232 ** (
233 ** movprfx z0\.d, p0/z, z1\.d
234 ** fsubr z0\.d, p0/m, z0\.d, \1
235 ** |
236 ** movprfx z0\.d, p0/z, \1
237 ** fsub z0\.d, p0/m, z0\.d, z1\.d
238 ** )
239 ** ret
240 */
241 TEST_UNIFORM_Z (subr_m1_f64_z_untied, svfloat64_t,
242 z0 = svsubr_n_f64_z (p0, z1, -1),
243 z0 = svsubr_z (p0, z1, -1))
244
245 /*
246 ** subr_f64_x_tied1:
247 ** fsub z0\.d, z1\.d, z0\.d
248 ** ret
249 */
250 TEST_UNIFORM_Z (subr_f64_x_tied1, svfloat64_t,
251 z0 = svsubr_f64_x (p0, z0, z1),
252 z0 = svsubr_x (p0, z0, z1))
253
254 /*
255 ** subr_f64_x_tied2:
256 ** fsub z0\.d, z0\.d, z1\.d
257 ** ret
258 */
259 TEST_UNIFORM_Z (subr_f64_x_tied2, svfloat64_t,
260 z0 = svsubr_f64_x (p0, z1, z0),
261 z0 = svsubr_x (p0, z1, z0))
262
263 /*
264 ** subr_f64_x_untied:
265 ** fsub z0\.d, z2\.d, z1\.d
266 ** ret
267 */
268 TEST_UNIFORM_Z (subr_f64_x_untied, svfloat64_t,
269 z0 = svsubr_f64_x (p0, z1, z2),
270 z0 = svsubr_x (p0, z1, z2))
271
272 /*
273 ** subr_d4_f64_x_tied1:
274 ** mov (z[0-9]+\.d), d4
275 ** fsub z0\.d, \1, z0\.d
276 ** ret
277 */
278 TEST_UNIFORM_ZD (subr_d4_f64_x_tied1, svfloat64_t, double,
279 z0 = svsubr_n_f64_x (p0, z0, d4),
280 z0 = svsubr_x (p0, z0, d4))
281
282 /*
283 ** subr_d4_f64_x_untied:
284 ** mov (z[0-9]+\.d), d4
285 ** fsub z0\.d, \1, z1\.d
286 ** ret
287 */
288 TEST_UNIFORM_ZD (subr_d4_f64_x_untied, svfloat64_t, double,
289 z0 = svsubr_n_f64_x (p0, z1, d4),
290 z0 = svsubr_x (p0, z1, d4))
291
292 /*
293 ** subr_1_f64_x_tied1:
294 ** fsubr z0\.d, p0/m, z0\.d, #1\.0
295 ** ret
296 */
297 TEST_UNIFORM_Z (subr_1_f64_x_tied1, svfloat64_t,
298 z0 = svsubr_n_f64_x (p0, z0, 1),
299 z0 = svsubr_x (p0, z0, 1))
300
301 /*
302 ** subr_1_f64_x_untied:
303 ** movprfx z0, z1
304 ** fsubr z0\.d, p0/m, z0\.d, #1\.0
305 ** ret
306 */
307 TEST_UNIFORM_Z (subr_1_f64_x_untied, svfloat64_t,
308 z0 = svsubr_n_f64_x (p0, z1, 1),
309 z0 = svsubr_x (p0, z1, 1))
310
311 /*
312 ** subr_0p5_f64_x_tied1:
313 ** fsubr z0\.d, p0/m, z0\.d, #0\.5
314 ** ret
315 */
316 TEST_UNIFORM_Z (subr_0p5_f64_x_tied1, svfloat64_t,
317 z0 = svsubr_n_f64_x (p0, z0, 0.5),
318 z0 = svsubr_x (p0, z0, 0.5))
319
320 /*
321 ** subr_0p5_f64_x_untied:
322 ** movprfx z0, z1
323 ** fsubr z0\.d, p0/m, z0\.d, #0\.5
324 ** ret
325 */
326 TEST_UNIFORM_Z (subr_0p5_f64_x_untied, svfloat64_t,
327 z0 = svsubr_n_f64_x (p0, z1, 0.5),
328 z0 = svsubr_x (p0, z1, 0.5))
329
330 /*
331 ** subr_m1_f64_x_tied1:
332 ** fmov (z[0-9]+\.d), #-1\.0(?:e\+0)?
333 ** fsub z0\.d, \1, z0\.d
334 ** ret
335 */
336 TEST_UNIFORM_Z (subr_m1_f64_x_tied1, svfloat64_t,
337 z0 = svsubr_n_f64_x (p0, z0, -1),
338 z0 = svsubr_x (p0, z0, -1))
339
340 /*
341 ** subr_m1_f64_x_untied:
342 ** fmov (z[0-9]+\.d), #-1\.0(?:e\+0)?
343 ** fsub z0\.d, \1, z1\.d
344 ** ret
345 */
346 TEST_UNIFORM_Z (subr_m1_f64_x_untied, svfloat64_t,
347 z0 = svsubr_n_f64_x (p0, z1, -1),
348 z0 = svsubr_x (p0, z1, -1))
349
350 /*
351 ** ptrue_subr_f64_x_tied1:
352 ** fsub z0\.d, z1\.d, z0\.d
353 ** ret
354 */
355 TEST_UNIFORM_Z (ptrue_subr_f64_x_tied1, svfloat64_t,
356 z0 = svsubr_f64_x (svptrue_b64 (), z0, z1),
357 z0 = svsubr_x (svptrue_b64 (), z0, z1))
358
359 /*
360 ** ptrue_subr_f64_x_tied2:
361 ** fsub z0\.d, z0\.d, z1\.d
362 ** ret
363 */
364 TEST_UNIFORM_Z (ptrue_subr_f64_x_tied2, svfloat64_t,
365 z0 = svsubr_f64_x (svptrue_b64 (), z1, z0),
366 z0 = svsubr_x (svptrue_b64 (), z1, z0))
367
368 /*
369 ** ptrue_subr_f64_x_untied:
370 ** fsub z0\.d, z2\.d, z1\.d
371 ** ret
372 */
373 TEST_UNIFORM_Z (ptrue_subr_f64_x_untied, svfloat64_t,
374 z0 = svsubr_f64_x (svptrue_b64 (), z1, z2),
375 z0 = svsubr_x (svptrue_b64 (), z1, z2))
376
377 /*
378 ** ptrue_subr_1_f64_x_tied1:
379 ** ...
380 ** ptrue p[0-9]+\.b[^\n]*
381 ** ...
382 ** ret
383 */
384 TEST_UNIFORM_Z (ptrue_subr_1_f64_x_tied1, svfloat64_t,
385 z0 = svsubr_n_f64_x (svptrue_b64 (), z0, 1),
386 z0 = svsubr_x (svptrue_b64 (), z0, 1))
387
388 /*
389 ** ptrue_subr_1_f64_x_untied:
390 ** ...
391 ** ptrue p[0-9]+\.b[^\n]*
392 ** ...
393 ** ret
394 */
395 TEST_UNIFORM_Z (ptrue_subr_1_f64_x_untied, svfloat64_t,
396 z0 = svsubr_n_f64_x (svptrue_b64 (), z1, 1),
397 z0 = svsubr_x (svptrue_b64 (), z1, 1))
398
399 /*
400 ** ptrue_subr_0p5_f64_x_tied1:
401 ** ...
402 ** ptrue p[0-9]+\.b[^\n]*
403 ** ...
404 ** ret
405 */
406 TEST_UNIFORM_Z (ptrue_subr_0p5_f64_x_tied1, svfloat64_t,
407 z0 = svsubr_n_f64_x (svptrue_b64 (), z0, 0.5),
408 z0 = svsubr_x (svptrue_b64 (), z0, 0.5))
409
410 /*
411 ** ptrue_subr_0p5_f64_x_untied:
412 ** ...
413 ** ptrue p[0-9]+\.b[^\n]*
414 ** ...
415 ** ret
416 */
417 TEST_UNIFORM_Z (ptrue_subr_0p5_f64_x_untied, svfloat64_t,
418 z0 = svsubr_n_f64_x (svptrue_b64 (), z1, 0.5),
419 z0 = svsubr_x (svptrue_b64 (), z1, 0.5))
420
421 /*
422 ** ptrue_subr_m1_f64_x_tied1:
423 ** fmov (z[0-9]+\.d), #-1\.0(?:e\+0)?
424 ** fsub z0\.d, \1, z0\.d
425 ** ret
426 */
427 TEST_UNIFORM_Z (ptrue_subr_m1_f64_x_tied1, svfloat64_t,
428 z0 = svsubr_n_f64_x (svptrue_b64 (), z0, -1),
429 z0 = svsubr_x (svptrue_b64 (), z0, -1))
430
431 /*
432 ** ptrue_subr_m1_f64_x_untied:
433 ** fmov (z[0-9]+\.d), #-1\.0(?:e\+0)?
434 ** fsub z0\.d, \1, z1\.d
435 ** ret
436 */
437 TEST_UNIFORM_Z (ptrue_subr_m1_f64_x_untied, svfloat64_t,
438 z0 = svsubr_n_f64_x (svptrue_b64 (), z1, -1),
439 z0 = svsubr_x (svptrue_b64 (), z1, -1))