2 Copyright 1988-2022 Free Software Foundation, Inc.
3 This is part of the GCC manual.
4 For copying conditions, see the copyright.rst file.
6 .. _mips-simd-architecture-(msa)-support:
8 MIPS SIMD Architecture (MSA) Support
9 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
11 GCC provides intrinsics to access the SIMD instructions provided by the
12 MSA MIPS SIMD Architecture. The interface is made available by including
13 ``<msa.h>`` and using :option:`-mmsa -mhard-float -mfp64 -mnan=2008`.
14 For each ``__builtin_msa_*``, there is a shortened name of the intrinsic,
17 MSA implements 128-bit wide vector registers, operating on 8-, 16-, 32- and
18 64-bit integer, 16- and 32-bit fixed-point, or 32- and 64-bit floating point
19 data elements. The following vectors typedefs are included in ``msa.h`` :
21 * ``v16i8``, a vector of sixteen signed 8-bit integers;
23 * ``v16u8``, a vector of sixteen unsigned 8-bit integers;
25 * ``v8i16``, a vector of eight signed 16-bit integers;
27 * ``v8u16``, a vector of eight unsigned 16-bit integers;
29 * ``v4i32``, a vector of four signed 32-bit integers;
31 * ``v4u32``, a vector of four unsigned 32-bit integers;
33 * ``v2i64``, a vector of two signed 64-bit integers;
35 * ``v2u64``, a vector of two unsigned 64-bit integers;
37 * ``v4f32``, a vector of four 32-bit floats;
39 * ``v2f64``, a vector of two 64-bit doubles.
41 Instructions and corresponding built-ins may have additional restrictions and/or
42 input/output values manipulated:
44 * ``imm0_1``, an integer literal in range 0 to 1;
46 * ``imm0_3``, an integer literal in range 0 to 3;
48 * ``imm0_7``, an integer literal in range 0 to 7;
50 * ``imm0_15``, an integer literal in range 0 to 15;
52 * ``imm0_31``, an integer literal in range 0 to 31;
54 * ``imm0_63``, an integer literal in range 0 to 63;
56 * ``imm0_255``, an integer literal in range 0 to 255;
58 * ``imm_n16_15``, an integer literal in range -16 to 15;
60 * ``imm_n512_511``, an integer literal in range -512 to 511;
62 * ``imm_n1024_1022``, an integer literal in range -512 to 511 left
63 shifted by 1 bit, i.e., -1024, -1022, ..., 1020, 1022;
65 * ``imm_n2048_2044``, an integer literal in range -512 to 511 left
66 shifted by 2 bits, i.e., -2048, -2044, ..., 2040, 2044;
68 * ``imm_n4096_4088``, an integer literal in range -512 to 511 left
69 shifted by 3 bits, i.e., -4096, -4088, ..., 4080, 4088;
71 * ``imm1_4``, an integer literal in range 1 to 4;
73 * ``i32, i64, u32, u64, f32, f64``, defined as follows:
79 #if __LONG_MAX__ == __LONG_LONG_MAX__
82 typedef long long i64;
85 typedef unsigned int u32;
86 #if __LONG_MAX__ == __LONG_LONG_MAX__
87 typedef unsigned long u64;
89 typedef unsigned long long u64;
96 .. _mips-simd-architecture-built-in-functions:
98 MIPS SIMD Architecture Built-in Functions
99 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
101 The intrinsics provided are listed below; each is named after the
106 v16i8 __builtin_msa_add_a_b (v16i8, v16i8);
107 v8i16 __builtin_msa_add_a_h (v8i16, v8i16);
108 v4i32 __builtin_msa_add_a_w (v4i32, v4i32);
109 v2i64 __builtin_msa_add_a_d (v2i64, v2i64);
111 v16i8 __builtin_msa_adds_a_b (v16i8, v16i8);
112 v8i16 __builtin_msa_adds_a_h (v8i16, v8i16);
113 v4i32 __builtin_msa_adds_a_w (v4i32, v4i32);
114 v2i64 __builtin_msa_adds_a_d (v2i64, v2i64);
116 v16i8 __builtin_msa_adds_s_b (v16i8, v16i8);
117 v8i16 __builtin_msa_adds_s_h (v8i16, v8i16);
118 v4i32 __builtin_msa_adds_s_w (v4i32, v4i32);
119 v2i64 __builtin_msa_adds_s_d (v2i64, v2i64);
121 v16u8 __builtin_msa_adds_u_b (v16u8, v16u8);
122 v8u16 __builtin_msa_adds_u_h (v8u16, v8u16);
123 v4u32 __builtin_msa_adds_u_w (v4u32, v4u32);
124 v2u64 __builtin_msa_adds_u_d (v2u64, v2u64);
126 v16i8 __builtin_msa_addv_b (v16i8, v16i8);
127 v8i16 __builtin_msa_addv_h (v8i16, v8i16);
128 v4i32 __builtin_msa_addv_w (v4i32, v4i32);
129 v2i64 __builtin_msa_addv_d (v2i64, v2i64);
131 v16i8 __builtin_msa_addvi_b (v16i8, imm0_31);
132 v8i16 __builtin_msa_addvi_h (v8i16, imm0_31);
133 v4i32 __builtin_msa_addvi_w (v4i32, imm0_31);
134 v2i64 __builtin_msa_addvi_d (v2i64, imm0_31);
136 v16u8 __builtin_msa_and_v (v16u8, v16u8);
138 v16u8 __builtin_msa_andi_b (v16u8, imm0_255);
140 v16i8 __builtin_msa_asub_s_b (v16i8, v16i8);
141 v8i16 __builtin_msa_asub_s_h (v8i16, v8i16);
142 v4i32 __builtin_msa_asub_s_w (v4i32, v4i32);
143 v2i64 __builtin_msa_asub_s_d (v2i64, v2i64);
145 v16u8 __builtin_msa_asub_u_b (v16u8, v16u8);
146 v8u16 __builtin_msa_asub_u_h (v8u16, v8u16);
147 v4u32 __builtin_msa_asub_u_w (v4u32, v4u32);
148 v2u64 __builtin_msa_asub_u_d (v2u64, v2u64);
150 v16i8 __builtin_msa_ave_s_b (v16i8, v16i8);
151 v8i16 __builtin_msa_ave_s_h (v8i16, v8i16);
152 v4i32 __builtin_msa_ave_s_w (v4i32, v4i32);
153 v2i64 __builtin_msa_ave_s_d (v2i64, v2i64);
155 v16u8 __builtin_msa_ave_u_b (v16u8, v16u8);
156 v8u16 __builtin_msa_ave_u_h (v8u16, v8u16);
157 v4u32 __builtin_msa_ave_u_w (v4u32, v4u32);
158 v2u64 __builtin_msa_ave_u_d (v2u64, v2u64);
160 v16i8 __builtin_msa_aver_s_b (v16i8, v16i8);
161 v8i16 __builtin_msa_aver_s_h (v8i16, v8i16);
162 v4i32 __builtin_msa_aver_s_w (v4i32, v4i32);
163 v2i64 __builtin_msa_aver_s_d (v2i64, v2i64);
165 v16u8 __builtin_msa_aver_u_b (v16u8, v16u8);
166 v8u16 __builtin_msa_aver_u_h (v8u16, v8u16);
167 v4u32 __builtin_msa_aver_u_w (v4u32, v4u32);
168 v2u64 __builtin_msa_aver_u_d (v2u64, v2u64);
170 v16u8 __builtin_msa_bclr_b (v16u8, v16u8);
171 v8u16 __builtin_msa_bclr_h (v8u16, v8u16);
172 v4u32 __builtin_msa_bclr_w (v4u32, v4u32);
173 v2u64 __builtin_msa_bclr_d (v2u64, v2u64);
175 v16u8 __builtin_msa_bclri_b (v16u8, imm0_7);
176 v8u16 __builtin_msa_bclri_h (v8u16, imm0_15);
177 v4u32 __builtin_msa_bclri_w (v4u32, imm0_31);
178 v2u64 __builtin_msa_bclri_d (v2u64, imm0_63);
180 v16u8 __builtin_msa_binsl_b (v16u8, v16u8, v16u8);
181 v8u16 __builtin_msa_binsl_h (v8u16, v8u16, v8u16);
182 v4u32 __builtin_msa_binsl_w (v4u32, v4u32, v4u32);
183 v2u64 __builtin_msa_binsl_d (v2u64, v2u64, v2u64);
185 v16u8 __builtin_msa_binsli_b (v16u8, v16u8, imm0_7);
186 v8u16 __builtin_msa_binsli_h (v8u16, v8u16, imm0_15);
187 v4u32 __builtin_msa_binsli_w (v4u32, v4u32, imm0_31);
188 v2u64 __builtin_msa_binsli_d (v2u64, v2u64, imm0_63);
190 v16u8 __builtin_msa_binsr_b (v16u8, v16u8, v16u8);
191 v8u16 __builtin_msa_binsr_h (v8u16, v8u16, v8u16);
192 v4u32 __builtin_msa_binsr_w (v4u32, v4u32, v4u32);
193 v2u64 __builtin_msa_binsr_d (v2u64, v2u64, v2u64);
195 v16u8 __builtin_msa_binsri_b (v16u8, v16u8, imm0_7);
196 v8u16 __builtin_msa_binsri_h (v8u16, v8u16, imm0_15);
197 v4u32 __builtin_msa_binsri_w (v4u32, v4u32, imm0_31);
198 v2u64 __builtin_msa_binsri_d (v2u64, v2u64, imm0_63);
200 v16u8 __builtin_msa_bmnz_v (v16u8, v16u8, v16u8);
202 v16u8 __builtin_msa_bmnzi_b (v16u8, v16u8, imm0_255);
204 v16u8 __builtin_msa_bmz_v (v16u8, v16u8, v16u8);
206 v16u8 __builtin_msa_bmzi_b (v16u8, v16u8, imm0_255);
208 v16u8 __builtin_msa_bneg_b (v16u8, v16u8);
209 v8u16 __builtin_msa_bneg_h (v8u16, v8u16);
210 v4u32 __builtin_msa_bneg_w (v4u32, v4u32);
211 v2u64 __builtin_msa_bneg_d (v2u64, v2u64);
213 v16u8 __builtin_msa_bnegi_b (v16u8, imm0_7);
214 v8u16 __builtin_msa_bnegi_h (v8u16, imm0_15);
215 v4u32 __builtin_msa_bnegi_w (v4u32, imm0_31);
216 v2u64 __builtin_msa_bnegi_d (v2u64, imm0_63);
218 i32 __builtin_msa_bnz_b (v16u8);
219 i32 __builtin_msa_bnz_h (v8u16);
220 i32 __builtin_msa_bnz_w (v4u32);
221 i32 __builtin_msa_bnz_d (v2u64);
223 i32 __builtin_msa_bnz_v (v16u8);
225 v16u8 __builtin_msa_bsel_v (v16u8, v16u8, v16u8);
227 v16u8 __builtin_msa_bseli_b (v16u8, v16u8, imm0_255);
229 v16u8 __builtin_msa_bset_b (v16u8, v16u8);
230 v8u16 __builtin_msa_bset_h (v8u16, v8u16);
231 v4u32 __builtin_msa_bset_w (v4u32, v4u32);
232 v2u64 __builtin_msa_bset_d (v2u64, v2u64);
234 v16u8 __builtin_msa_bseti_b (v16u8, imm0_7);
235 v8u16 __builtin_msa_bseti_h (v8u16, imm0_15);
236 v4u32 __builtin_msa_bseti_w (v4u32, imm0_31);
237 v2u64 __builtin_msa_bseti_d (v2u64, imm0_63);
239 i32 __builtin_msa_bz_b (v16u8);
240 i32 __builtin_msa_bz_h (v8u16);
241 i32 __builtin_msa_bz_w (v4u32);
242 i32 __builtin_msa_bz_d (v2u64);
244 i32 __builtin_msa_bz_v (v16u8);
246 v16i8 __builtin_msa_ceq_b (v16i8, v16i8);
247 v8i16 __builtin_msa_ceq_h (v8i16, v8i16);
248 v4i32 __builtin_msa_ceq_w (v4i32, v4i32);
249 v2i64 __builtin_msa_ceq_d (v2i64, v2i64);
251 v16i8 __builtin_msa_ceqi_b (v16i8, imm_n16_15);
252 v8i16 __builtin_msa_ceqi_h (v8i16, imm_n16_15);
253 v4i32 __builtin_msa_ceqi_w (v4i32, imm_n16_15);
254 v2i64 __builtin_msa_ceqi_d (v2i64, imm_n16_15);
256 i32 __builtin_msa_cfcmsa (imm0_31);
258 v16i8 __builtin_msa_cle_s_b (v16i8, v16i8);
259 v8i16 __builtin_msa_cle_s_h (v8i16, v8i16);
260 v4i32 __builtin_msa_cle_s_w (v4i32, v4i32);
261 v2i64 __builtin_msa_cle_s_d (v2i64, v2i64);
263 v16i8 __builtin_msa_cle_u_b (v16u8, v16u8);
264 v8i16 __builtin_msa_cle_u_h (v8u16, v8u16);
265 v4i32 __builtin_msa_cle_u_w (v4u32, v4u32);
266 v2i64 __builtin_msa_cle_u_d (v2u64, v2u64);
268 v16i8 __builtin_msa_clei_s_b (v16i8, imm_n16_15);
269 v8i16 __builtin_msa_clei_s_h (v8i16, imm_n16_15);
270 v4i32 __builtin_msa_clei_s_w (v4i32, imm_n16_15);
271 v2i64 __builtin_msa_clei_s_d (v2i64, imm_n16_15);
273 v16i8 __builtin_msa_clei_u_b (v16u8, imm0_31);
274 v8i16 __builtin_msa_clei_u_h (v8u16, imm0_31);
275 v4i32 __builtin_msa_clei_u_w (v4u32, imm0_31);
276 v2i64 __builtin_msa_clei_u_d (v2u64, imm0_31);
278 v16i8 __builtin_msa_clt_s_b (v16i8, v16i8);
279 v8i16 __builtin_msa_clt_s_h (v8i16, v8i16);
280 v4i32 __builtin_msa_clt_s_w (v4i32, v4i32);
281 v2i64 __builtin_msa_clt_s_d (v2i64, v2i64);
283 v16i8 __builtin_msa_clt_u_b (v16u8, v16u8);
284 v8i16 __builtin_msa_clt_u_h (v8u16, v8u16);
285 v4i32 __builtin_msa_clt_u_w (v4u32, v4u32);
286 v2i64 __builtin_msa_clt_u_d (v2u64, v2u64);
288 v16i8 __builtin_msa_clti_s_b (v16i8, imm_n16_15);
289 v8i16 __builtin_msa_clti_s_h (v8i16, imm_n16_15);
290 v4i32 __builtin_msa_clti_s_w (v4i32, imm_n16_15);
291 v2i64 __builtin_msa_clti_s_d (v2i64, imm_n16_15);
293 v16i8 __builtin_msa_clti_u_b (v16u8, imm0_31);
294 v8i16 __builtin_msa_clti_u_h (v8u16, imm0_31);
295 v4i32 __builtin_msa_clti_u_w (v4u32, imm0_31);
296 v2i64 __builtin_msa_clti_u_d (v2u64, imm0_31);
298 i32 __builtin_msa_copy_s_b (v16i8, imm0_15);
299 i32 __builtin_msa_copy_s_h (v8i16, imm0_7);
300 i32 __builtin_msa_copy_s_w (v4i32, imm0_3);
301 i64 __builtin_msa_copy_s_d (v2i64, imm0_1);
303 u32 __builtin_msa_copy_u_b (v16i8, imm0_15);
304 u32 __builtin_msa_copy_u_h (v8i16, imm0_7);
305 u32 __builtin_msa_copy_u_w (v4i32, imm0_3);
306 u64 __builtin_msa_copy_u_d (v2i64, imm0_1);
308 void __builtin_msa_ctcmsa (imm0_31, i32);
310 v16i8 __builtin_msa_div_s_b (v16i8, v16i8);
311 v8i16 __builtin_msa_div_s_h (v8i16, v8i16);
312 v4i32 __builtin_msa_div_s_w (v4i32, v4i32);
313 v2i64 __builtin_msa_div_s_d (v2i64, v2i64);
315 v16u8 __builtin_msa_div_u_b (v16u8, v16u8);
316 v8u16 __builtin_msa_div_u_h (v8u16, v8u16);
317 v4u32 __builtin_msa_div_u_w (v4u32, v4u32);
318 v2u64 __builtin_msa_div_u_d (v2u64, v2u64);
320 v8i16 __builtin_msa_dotp_s_h (v16i8, v16i8);
321 v4i32 __builtin_msa_dotp_s_w (v8i16, v8i16);
322 v2i64 __builtin_msa_dotp_s_d (v4i32, v4i32);
324 v8u16 __builtin_msa_dotp_u_h (v16u8, v16u8);
325 v4u32 __builtin_msa_dotp_u_w (v8u16, v8u16);
326 v2u64 __builtin_msa_dotp_u_d (v4u32, v4u32);
328 v8i16 __builtin_msa_dpadd_s_h (v8i16, v16i8, v16i8);
329 v4i32 __builtin_msa_dpadd_s_w (v4i32, v8i16, v8i16);
330 v2i64 __builtin_msa_dpadd_s_d (v2i64, v4i32, v4i32);
332 v8u16 __builtin_msa_dpadd_u_h (v8u16, v16u8, v16u8);
333 v4u32 __builtin_msa_dpadd_u_w (v4u32, v8u16, v8u16);
334 v2u64 __builtin_msa_dpadd_u_d (v2u64, v4u32, v4u32);
336 v8i16 __builtin_msa_dpsub_s_h (v8i16, v16i8, v16i8);
337 v4i32 __builtin_msa_dpsub_s_w (v4i32, v8i16, v8i16);
338 v2i64 __builtin_msa_dpsub_s_d (v2i64, v4i32, v4i32);
340 v8i16 __builtin_msa_dpsub_u_h (v8i16, v16u8, v16u8);
341 v4i32 __builtin_msa_dpsub_u_w (v4i32, v8u16, v8u16);
342 v2i64 __builtin_msa_dpsub_u_d (v2i64, v4u32, v4u32);
344 v4f32 __builtin_msa_fadd_w (v4f32, v4f32);
345 v2f64 __builtin_msa_fadd_d (v2f64, v2f64);
347 v4i32 __builtin_msa_fcaf_w (v4f32, v4f32);
348 v2i64 __builtin_msa_fcaf_d (v2f64, v2f64);
350 v4i32 __builtin_msa_fceq_w (v4f32, v4f32);
351 v2i64 __builtin_msa_fceq_d (v2f64, v2f64);
353 v4i32 __builtin_msa_fclass_w (v4f32);
354 v2i64 __builtin_msa_fclass_d (v2f64);
356 v4i32 __builtin_msa_fcle_w (v4f32, v4f32);
357 v2i64 __builtin_msa_fcle_d (v2f64, v2f64);
359 v4i32 __builtin_msa_fclt_w (v4f32, v4f32);
360 v2i64 __builtin_msa_fclt_d (v2f64, v2f64);
362 v4i32 __builtin_msa_fcne_w (v4f32, v4f32);
363 v2i64 __builtin_msa_fcne_d (v2f64, v2f64);
365 v4i32 __builtin_msa_fcor_w (v4f32, v4f32);
366 v2i64 __builtin_msa_fcor_d (v2f64, v2f64);
368 v4i32 __builtin_msa_fcueq_w (v4f32, v4f32);
369 v2i64 __builtin_msa_fcueq_d (v2f64, v2f64);
371 v4i32 __builtin_msa_fcule_w (v4f32, v4f32);
372 v2i64 __builtin_msa_fcule_d (v2f64, v2f64);
374 v4i32 __builtin_msa_fcult_w (v4f32, v4f32);
375 v2i64 __builtin_msa_fcult_d (v2f64, v2f64);
377 v4i32 __builtin_msa_fcun_w (v4f32, v4f32);
378 v2i64 __builtin_msa_fcun_d (v2f64, v2f64);
380 v4i32 __builtin_msa_fcune_w (v4f32, v4f32);
381 v2i64 __builtin_msa_fcune_d (v2f64, v2f64);
383 v4f32 __builtin_msa_fdiv_w (v4f32, v4f32);
384 v2f64 __builtin_msa_fdiv_d (v2f64, v2f64);
386 v8i16 __builtin_msa_fexdo_h (v4f32, v4f32);
387 v4f32 __builtin_msa_fexdo_w (v2f64, v2f64);
389 v4f32 __builtin_msa_fexp2_w (v4f32, v4i32);
390 v2f64 __builtin_msa_fexp2_d (v2f64, v2i64);
392 v4f32 __builtin_msa_fexupl_w (v8i16);
393 v2f64 __builtin_msa_fexupl_d (v4f32);
395 v4f32 __builtin_msa_fexupr_w (v8i16);
396 v2f64 __builtin_msa_fexupr_d (v4f32);
398 v4f32 __builtin_msa_ffint_s_w (v4i32);
399 v2f64 __builtin_msa_ffint_s_d (v2i64);
401 v4f32 __builtin_msa_ffint_u_w (v4u32);
402 v2f64 __builtin_msa_ffint_u_d (v2u64);
404 v4f32 __builtin_msa_ffql_w (v8i16);
405 v2f64 __builtin_msa_ffql_d (v4i32);
407 v4f32 __builtin_msa_ffqr_w (v8i16);
408 v2f64 __builtin_msa_ffqr_d (v4i32);
410 v16i8 __builtin_msa_fill_b (i32);
411 v8i16 __builtin_msa_fill_h (i32);
412 v4i32 __builtin_msa_fill_w (i32);
413 v2i64 __builtin_msa_fill_d (i64);
415 v4f32 __builtin_msa_flog2_w (v4f32);
416 v2f64 __builtin_msa_flog2_d (v2f64);
418 v4f32 __builtin_msa_fmadd_w (v4f32, v4f32, v4f32);
419 v2f64 __builtin_msa_fmadd_d (v2f64, v2f64, v2f64);
421 v4f32 __builtin_msa_fmax_w (v4f32, v4f32);
422 v2f64 __builtin_msa_fmax_d (v2f64, v2f64);
424 v4f32 __builtin_msa_fmax_a_w (v4f32, v4f32);
425 v2f64 __builtin_msa_fmax_a_d (v2f64, v2f64);
427 v4f32 __builtin_msa_fmin_w (v4f32, v4f32);
428 v2f64 __builtin_msa_fmin_d (v2f64, v2f64);
430 v4f32 __builtin_msa_fmin_a_w (v4f32, v4f32);
431 v2f64 __builtin_msa_fmin_a_d (v2f64, v2f64);
433 v4f32 __builtin_msa_fmsub_w (v4f32, v4f32, v4f32);
434 v2f64 __builtin_msa_fmsub_d (v2f64, v2f64, v2f64);
436 v4f32 __builtin_msa_fmul_w (v4f32, v4f32);
437 v2f64 __builtin_msa_fmul_d (v2f64, v2f64);
439 v4f32 __builtin_msa_frint_w (v4f32);
440 v2f64 __builtin_msa_frint_d (v2f64);
442 v4f32 __builtin_msa_frcp_w (v4f32);
443 v2f64 __builtin_msa_frcp_d (v2f64);
445 v4f32 __builtin_msa_frsqrt_w (v4f32);
446 v2f64 __builtin_msa_frsqrt_d (v2f64);
448 v4i32 __builtin_msa_fsaf_w (v4f32, v4f32);
449 v2i64 __builtin_msa_fsaf_d (v2f64, v2f64);
451 v4i32 __builtin_msa_fseq_w (v4f32, v4f32);
452 v2i64 __builtin_msa_fseq_d (v2f64, v2f64);
454 v4i32 __builtin_msa_fsle_w (v4f32, v4f32);
455 v2i64 __builtin_msa_fsle_d (v2f64, v2f64);
457 v4i32 __builtin_msa_fslt_w (v4f32, v4f32);
458 v2i64 __builtin_msa_fslt_d (v2f64, v2f64);
460 v4i32 __builtin_msa_fsne_w (v4f32, v4f32);
461 v2i64 __builtin_msa_fsne_d (v2f64, v2f64);
463 v4i32 __builtin_msa_fsor_w (v4f32, v4f32);
464 v2i64 __builtin_msa_fsor_d (v2f64, v2f64);
466 v4f32 __builtin_msa_fsqrt_w (v4f32);
467 v2f64 __builtin_msa_fsqrt_d (v2f64);
469 v4f32 __builtin_msa_fsub_w (v4f32, v4f32);
470 v2f64 __builtin_msa_fsub_d (v2f64, v2f64);
472 v4i32 __builtin_msa_fsueq_w (v4f32, v4f32);
473 v2i64 __builtin_msa_fsueq_d (v2f64, v2f64);
475 v4i32 __builtin_msa_fsule_w (v4f32, v4f32);
476 v2i64 __builtin_msa_fsule_d (v2f64, v2f64);
478 v4i32 __builtin_msa_fsult_w (v4f32, v4f32);
479 v2i64 __builtin_msa_fsult_d (v2f64, v2f64);
481 v4i32 __builtin_msa_fsun_w (v4f32, v4f32);
482 v2i64 __builtin_msa_fsun_d (v2f64, v2f64);
484 v4i32 __builtin_msa_fsune_w (v4f32, v4f32);
485 v2i64 __builtin_msa_fsune_d (v2f64, v2f64);
487 v4i32 __builtin_msa_ftint_s_w (v4f32);
488 v2i64 __builtin_msa_ftint_s_d (v2f64);
490 v4u32 __builtin_msa_ftint_u_w (v4f32);
491 v2u64 __builtin_msa_ftint_u_d (v2f64);
493 v8i16 __builtin_msa_ftq_h (v4f32, v4f32);
494 v4i32 __builtin_msa_ftq_w (v2f64, v2f64);
496 v4i32 __builtin_msa_ftrunc_s_w (v4f32);
497 v2i64 __builtin_msa_ftrunc_s_d (v2f64);
499 v4u32 __builtin_msa_ftrunc_u_w (v4f32);
500 v2u64 __builtin_msa_ftrunc_u_d (v2f64);
502 v8i16 __builtin_msa_hadd_s_h (v16i8, v16i8);
503 v4i32 __builtin_msa_hadd_s_w (v8i16, v8i16);
504 v2i64 __builtin_msa_hadd_s_d (v4i32, v4i32);
506 v8u16 __builtin_msa_hadd_u_h (v16u8, v16u8);
507 v4u32 __builtin_msa_hadd_u_w (v8u16, v8u16);
508 v2u64 __builtin_msa_hadd_u_d (v4u32, v4u32);
510 v8i16 __builtin_msa_hsub_s_h (v16i8, v16i8);
511 v4i32 __builtin_msa_hsub_s_w (v8i16, v8i16);
512 v2i64 __builtin_msa_hsub_s_d (v4i32, v4i32);
514 v8i16 __builtin_msa_hsub_u_h (v16u8, v16u8);
515 v4i32 __builtin_msa_hsub_u_w (v8u16, v8u16);
516 v2i64 __builtin_msa_hsub_u_d (v4u32, v4u32);
518 v16i8 __builtin_msa_ilvev_b (v16i8, v16i8);
519 v8i16 __builtin_msa_ilvev_h (v8i16, v8i16);
520 v4i32 __builtin_msa_ilvev_w (v4i32, v4i32);
521 v2i64 __builtin_msa_ilvev_d (v2i64, v2i64);
523 v16i8 __builtin_msa_ilvl_b (v16i8, v16i8);
524 v8i16 __builtin_msa_ilvl_h (v8i16, v8i16);
525 v4i32 __builtin_msa_ilvl_w (v4i32, v4i32);
526 v2i64 __builtin_msa_ilvl_d (v2i64, v2i64);
528 v16i8 __builtin_msa_ilvod_b (v16i8, v16i8);
529 v8i16 __builtin_msa_ilvod_h (v8i16, v8i16);
530 v4i32 __builtin_msa_ilvod_w (v4i32, v4i32);
531 v2i64 __builtin_msa_ilvod_d (v2i64, v2i64);
533 v16i8 __builtin_msa_ilvr_b (v16i8, v16i8);
534 v8i16 __builtin_msa_ilvr_h (v8i16, v8i16);
535 v4i32 __builtin_msa_ilvr_w (v4i32, v4i32);
536 v2i64 __builtin_msa_ilvr_d (v2i64, v2i64);
538 v16i8 __builtin_msa_insert_b (v16i8, imm0_15, i32);
539 v8i16 __builtin_msa_insert_h (v8i16, imm0_7, i32);
540 v4i32 __builtin_msa_insert_w (v4i32, imm0_3, i32);
541 v2i64 __builtin_msa_insert_d (v2i64, imm0_1, i64);
543 v16i8 __builtin_msa_insve_b (v16i8, imm0_15, v16i8);
544 v8i16 __builtin_msa_insve_h (v8i16, imm0_7, v8i16);
545 v4i32 __builtin_msa_insve_w (v4i32, imm0_3, v4i32);
546 v2i64 __builtin_msa_insve_d (v2i64, imm0_1, v2i64);
548 v16i8 __builtin_msa_ld_b (const void *, imm_n512_511);
549 v8i16 __builtin_msa_ld_h (const void *, imm_n1024_1022);
550 v4i32 __builtin_msa_ld_w (const void *, imm_n2048_2044);
551 v2i64 __builtin_msa_ld_d (const void *, imm_n4096_4088);
553 v16i8 __builtin_msa_ldi_b (imm_n512_511);
554 v8i16 __builtin_msa_ldi_h (imm_n512_511);
555 v4i32 __builtin_msa_ldi_w (imm_n512_511);
556 v2i64 __builtin_msa_ldi_d (imm_n512_511);
558 v8i16 __builtin_msa_madd_q_h (v8i16, v8i16, v8i16);
559 v4i32 __builtin_msa_madd_q_w (v4i32, v4i32, v4i32);
561 v8i16 __builtin_msa_maddr_q_h (v8i16, v8i16, v8i16);
562 v4i32 __builtin_msa_maddr_q_w (v4i32, v4i32, v4i32);
564 v16i8 __builtin_msa_maddv_b (v16i8, v16i8, v16i8);
565 v8i16 __builtin_msa_maddv_h (v8i16, v8i16, v8i16);
566 v4i32 __builtin_msa_maddv_w (v4i32, v4i32, v4i32);
567 v2i64 __builtin_msa_maddv_d (v2i64, v2i64, v2i64);
569 v16i8 __builtin_msa_max_a_b (v16i8, v16i8);
570 v8i16 __builtin_msa_max_a_h (v8i16, v8i16);
571 v4i32 __builtin_msa_max_a_w (v4i32, v4i32);
572 v2i64 __builtin_msa_max_a_d (v2i64, v2i64);
574 v16i8 __builtin_msa_max_s_b (v16i8, v16i8);
575 v8i16 __builtin_msa_max_s_h (v8i16, v8i16);
576 v4i32 __builtin_msa_max_s_w (v4i32, v4i32);
577 v2i64 __builtin_msa_max_s_d (v2i64, v2i64);
579 v16u8 __builtin_msa_max_u_b (v16u8, v16u8);
580 v8u16 __builtin_msa_max_u_h (v8u16, v8u16);
581 v4u32 __builtin_msa_max_u_w (v4u32, v4u32);
582 v2u64 __builtin_msa_max_u_d (v2u64, v2u64);
584 v16i8 __builtin_msa_maxi_s_b (v16i8, imm_n16_15);
585 v8i16 __builtin_msa_maxi_s_h (v8i16, imm_n16_15);
586 v4i32 __builtin_msa_maxi_s_w (v4i32, imm_n16_15);
587 v2i64 __builtin_msa_maxi_s_d (v2i64, imm_n16_15);
589 v16u8 __builtin_msa_maxi_u_b (v16u8, imm0_31);
590 v8u16 __builtin_msa_maxi_u_h (v8u16, imm0_31);
591 v4u32 __builtin_msa_maxi_u_w (v4u32, imm0_31);
592 v2u64 __builtin_msa_maxi_u_d (v2u64, imm0_31);
594 v16i8 __builtin_msa_min_a_b (v16i8, v16i8);
595 v8i16 __builtin_msa_min_a_h (v8i16, v8i16);
596 v4i32 __builtin_msa_min_a_w (v4i32, v4i32);
597 v2i64 __builtin_msa_min_a_d (v2i64, v2i64);
599 v16i8 __builtin_msa_min_s_b (v16i8, v16i8);
600 v8i16 __builtin_msa_min_s_h (v8i16, v8i16);
601 v4i32 __builtin_msa_min_s_w (v4i32, v4i32);
602 v2i64 __builtin_msa_min_s_d (v2i64, v2i64);
604 v16u8 __builtin_msa_min_u_b (v16u8, v16u8);
605 v8u16 __builtin_msa_min_u_h (v8u16, v8u16);
606 v4u32 __builtin_msa_min_u_w (v4u32, v4u32);
607 v2u64 __builtin_msa_min_u_d (v2u64, v2u64);
609 v16i8 __builtin_msa_mini_s_b (v16i8, imm_n16_15);
610 v8i16 __builtin_msa_mini_s_h (v8i16, imm_n16_15);
611 v4i32 __builtin_msa_mini_s_w (v4i32, imm_n16_15);
612 v2i64 __builtin_msa_mini_s_d (v2i64, imm_n16_15);
614 v16u8 __builtin_msa_mini_u_b (v16u8, imm0_31);
615 v8u16 __builtin_msa_mini_u_h (v8u16, imm0_31);
616 v4u32 __builtin_msa_mini_u_w (v4u32, imm0_31);
617 v2u64 __builtin_msa_mini_u_d (v2u64, imm0_31);
619 v16i8 __builtin_msa_mod_s_b (v16i8, v16i8);
620 v8i16 __builtin_msa_mod_s_h (v8i16, v8i16);
621 v4i32 __builtin_msa_mod_s_w (v4i32, v4i32);
622 v2i64 __builtin_msa_mod_s_d (v2i64, v2i64);
624 v16u8 __builtin_msa_mod_u_b (v16u8, v16u8);
625 v8u16 __builtin_msa_mod_u_h (v8u16, v8u16);
626 v4u32 __builtin_msa_mod_u_w (v4u32, v4u32);
627 v2u64 __builtin_msa_mod_u_d (v2u64, v2u64);
629 v16i8 __builtin_msa_move_v (v16i8);
631 v8i16 __builtin_msa_msub_q_h (v8i16, v8i16, v8i16);
632 v4i32 __builtin_msa_msub_q_w (v4i32, v4i32, v4i32);
634 v8i16 __builtin_msa_msubr_q_h (v8i16, v8i16, v8i16);
635 v4i32 __builtin_msa_msubr_q_w (v4i32, v4i32, v4i32);
637 v16i8 __builtin_msa_msubv_b (v16i8, v16i8, v16i8);
638 v8i16 __builtin_msa_msubv_h (v8i16, v8i16, v8i16);
639 v4i32 __builtin_msa_msubv_w (v4i32, v4i32, v4i32);
640 v2i64 __builtin_msa_msubv_d (v2i64, v2i64, v2i64);
642 v8i16 __builtin_msa_mul_q_h (v8i16, v8i16);
643 v4i32 __builtin_msa_mul_q_w (v4i32, v4i32);
645 v8i16 __builtin_msa_mulr_q_h (v8i16, v8i16);
646 v4i32 __builtin_msa_mulr_q_w (v4i32, v4i32);
648 v16i8 __builtin_msa_mulv_b (v16i8, v16i8);
649 v8i16 __builtin_msa_mulv_h (v8i16, v8i16);
650 v4i32 __builtin_msa_mulv_w (v4i32, v4i32);
651 v2i64 __builtin_msa_mulv_d (v2i64, v2i64);
653 v16i8 __builtin_msa_nloc_b (v16i8);
654 v8i16 __builtin_msa_nloc_h (v8i16);
655 v4i32 __builtin_msa_nloc_w (v4i32);
656 v2i64 __builtin_msa_nloc_d (v2i64);
658 v16i8 __builtin_msa_nlzc_b (v16i8);
659 v8i16 __builtin_msa_nlzc_h (v8i16);
660 v4i32 __builtin_msa_nlzc_w (v4i32);
661 v2i64 __builtin_msa_nlzc_d (v2i64);
663 v16u8 __builtin_msa_nor_v (v16u8, v16u8);
665 v16u8 __builtin_msa_nori_b (v16u8, imm0_255);
667 v16u8 __builtin_msa_or_v (v16u8, v16u8);
669 v16u8 __builtin_msa_ori_b (v16u8, imm0_255);
671 v16i8 __builtin_msa_pckev_b (v16i8, v16i8);
672 v8i16 __builtin_msa_pckev_h (v8i16, v8i16);
673 v4i32 __builtin_msa_pckev_w (v4i32, v4i32);
674 v2i64 __builtin_msa_pckev_d (v2i64, v2i64);
676 v16i8 __builtin_msa_pckod_b (v16i8, v16i8);
677 v8i16 __builtin_msa_pckod_h (v8i16, v8i16);
678 v4i32 __builtin_msa_pckod_w (v4i32, v4i32);
679 v2i64 __builtin_msa_pckod_d (v2i64, v2i64);
681 v16i8 __builtin_msa_pcnt_b (v16i8);
682 v8i16 __builtin_msa_pcnt_h (v8i16);
683 v4i32 __builtin_msa_pcnt_w (v4i32);
684 v2i64 __builtin_msa_pcnt_d (v2i64);
686 v16i8 __builtin_msa_sat_s_b (v16i8, imm0_7);
687 v8i16 __builtin_msa_sat_s_h (v8i16, imm0_15);
688 v4i32 __builtin_msa_sat_s_w (v4i32, imm0_31);
689 v2i64 __builtin_msa_sat_s_d (v2i64, imm0_63);
691 v16u8 __builtin_msa_sat_u_b (v16u8, imm0_7);
692 v8u16 __builtin_msa_sat_u_h (v8u16, imm0_15);
693 v4u32 __builtin_msa_sat_u_w (v4u32, imm0_31);
694 v2u64 __builtin_msa_sat_u_d (v2u64, imm0_63);
696 v16i8 __builtin_msa_shf_b (v16i8, imm0_255);
697 v8i16 __builtin_msa_shf_h (v8i16, imm0_255);
698 v4i32 __builtin_msa_shf_w (v4i32, imm0_255);
700 v16i8 __builtin_msa_sld_b (v16i8, v16i8, i32);
701 v8i16 __builtin_msa_sld_h (v8i16, v8i16, i32);
702 v4i32 __builtin_msa_sld_w (v4i32, v4i32, i32);
703 v2i64 __builtin_msa_sld_d (v2i64, v2i64, i32);
705 v16i8 __builtin_msa_sldi_b (v16i8, v16i8, imm0_15);
706 v8i16 __builtin_msa_sldi_h (v8i16, v8i16, imm0_7);
707 v4i32 __builtin_msa_sldi_w (v4i32, v4i32, imm0_3);
708 v2i64 __builtin_msa_sldi_d (v2i64, v2i64, imm0_1);
710 v16i8 __builtin_msa_sll_b (v16i8, v16i8);
711 v8i16 __builtin_msa_sll_h (v8i16, v8i16);
712 v4i32 __builtin_msa_sll_w (v4i32, v4i32);
713 v2i64 __builtin_msa_sll_d (v2i64, v2i64);
715 v16i8 __builtin_msa_slli_b (v16i8, imm0_7);
716 v8i16 __builtin_msa_slli_h (v8i16, imm0_15);
717 v4i32 __builtin_msa_slli_w (v4i32, imm0_31);
718 v2i64 __builtin_msa_slli_d (v2i64, imm0_63);
720 v16i8 __builtin_msa_splat_b (v16i8, i32);
721 v8i16 __builtin_msa_splat_h (v8i16, i32);
722 v4i32 __builtin_msa_splat_w (v4i32, i32);
723 v2i64 __builtin_msa_splat_d (v2i64, i32);
725 v16i8 __builtin_msa_splati_b (v16i8, imm0_15);
726 v8i16 __builtin_msa_splati_h (v8i16, imm0_7);
727 v4i32 __builtin_msa_splati_w (v4i32, imm0_3);
728 v2i64 __builtin_msa_splati_d (v2i64, imm0_1);
730 v16i8 __builtin_msa_sra_b (v16i8, v16i8);
731 v8i16 __builtin_msa_sra_h (v8i16, v8i16);
732 v4i32 __builtin_msa_sra_w (v4i32, v4i32);
733 v2i64 __builtin_msa_sra_d (v2i64, v2i64);
735 v16i8 __builtin_msa_srai_b (v16i8, imm0_7);
736 v8i16 __builtin_msa_srai_h (v8i16, imm0_15);
737 v4i32 __builtin_msa_srai_w (v4i32, imm0_31);
738 v2i64 __builtin_msa_srai_d (v2i64, imm0_63);
740 v16i8 __builtin_msa_srar_b (v16i8, v16i8);
741 v8i16 __builtin_msa_srar_h (v8i16, v8i16);
742 v4i32 __builtin_msa_srar_w (v4i32, v4i32);
743 v2i64 __builtin_msa_srar_d (v2i64, v2i64);
745 v16i8 __builtin_msa_srari_b (v16i8, imm0_7);
746 v8i16 __builtin_msa_srari_h (v8i16, imm0_15);
747 v4i32 __builtin_msa_srari_w (v4i32, imm0_31);
748 v2i64 __builtin_msa_srari_d (v2i64, imm0_63);
750 v16i8 __builtin_msa_srl_b (v16i8, v16i8);
751 v8i16 __builtin_msa_srl_h (v8i16, v8i16);
752 v4i32 __builtin_msa_srl_w (v4i32, v4i32);
753 v2i64 __builtin_msa_srl_d (v2i64, v2i64);
755 v16i8 __builtin_msa_srli_b (v16i8, imm0_7);
756 v8i16 __builtin_msa_srli_h (v8i16, imm0_15);
757 v4i32 __builtin_msa_srli_w (v4i32, imm0_31);
758 v2i64 __builtin_msa_srli_d (v2i64, imm0_63);
760 v16i8 __builtin_msa_srlr_b (v16i8, v16i8);
761 v8i16 __builtin_msa_srlr_h (v8i16, v8i16);
762 v4i32 __builtin_msa_srlr_w (v4i32, v4i32);
763 v2i64 __builtin_msa_srlr_d (v2i64, v2i64);
765 v16i8 __builtin_msa_srlri_b (v16i8, imm0_7);
766 v8i16 __builtin_msa_srlri_h (v8i16, imm0_15);
767 v4i32 __builtin_msa_srlri_w (v4i32, imm0_31);
768 v2i64 __builtin_msa_srlri_d (v2i64, imm0_63);
770 void __builtin_msa_st_b (v16i8, void *, imm_n512_511);
771 void __builtin_msa_st_h (v8i16, void *, imm_n1024_1022);
772 void __builtin_msa_st_w (v4i32, void *, imm_n2048_2044);
773 void __builtin_msa_st_d (v2i64, void *, imm_n4096_4088);
775 v16i8 __builtin_msa_subs_s_b (v16i8, v16i8);
776 v8i16 __builtin_msa_subs_s_h (v8i16, v8i16);
777 v4i32 __builtin_msa_subs_s_w (v4i32, v4i32);
778 v2i64 __builtin_msa_subs_s_d (v2i64, v2i64);
780 v16u8 __builtin_msa_subs_u_b (v16u8, v16u8);
781 v8u16 __builtin_msa_subs_u_h (v8u16, v8u16);
782 v4u32 __builtin_msa_subs_u_w (v4u32, v4u32);
783 v2u64 __builtin_msa_subs_u_d (v2u64, v2u64);
785 v16u8 __builtin_msa_subsus_u_b (v16u8, v16i8);
786 v8u16 __builtin_msa_subsus_u_h (v8u16, v8i16);
787 v4u32 __builtin_msa_subsus_u_w (v4u32, v4i32);
788 v2u64 __builtin_msa_subsus_u_d (v2u64, v2i64);
790 v16i8 __builtin_msa_subsuu_s_b (v16u8, v16u8);
791 v8i16 __builtin_msa_subsuu_s_h (v8u16, v8u16);
792 v4i32 __builtin_msa_subsuu_s_w (v4u32, v4u32);
793 v2i64 __builtin_msa_subsuu_s_d (v2u64, v2u64);
795 v16i8 __builtin_msa_subv_b (v16i8, v16i8);
796 v8i16 __builtin_msa_subv_h (v8i16, v8i16);
797 v4i32 __builtin_msa_subv_w (v4i32, v4i32);
798 v2i64 __builtin_msa_subv_d (v2i64, v2i64);
800 v16i8 __builtin_msa_subvi_b (v16i8, imm0_31);
801 v8i16 __builtin_msa_subvi_h (v8i16, imm0_31);
802 v4i32 __builtin_msa_subvi_w (v4i32, imm0_31);
803 v2i64 __builtin_msa_subvi_d (v2i64, imm0_31);
805 v16i8 __builtin_msa_vshf_b (v16i8, v16i8, v16i8);
806 v8i16 __builtin_msa_vshf_h (v8i16, v8i16, v8i16);
807 v4i32 __builtin_msa_vshf_w (v4i32, v4i32, v4i32);
808 v2i64 __builtin_msa_vshf_d (v2i64, v2i64, v2i64);
810 v16u8 __builtin_msa_xor_v (v16u8, v16u8);
812 v16u8 __builtin_msa_xori_b (v16u8, imm0_255);