Lines Matching +full:k +full:- +full:block

3 # This source code is licensed under the BSD-style license found in the
7 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8__aarch32_neon_mlal_lane_cortex_a7
9 k-block: 8
11 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8__aarch32_neon_mlal_lane_prfm_cortex_a7
13 k-block: 8
15 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8__aarch32_neon_mlal_lane_cortex_a7
17 k-block: 8
19 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8__aarch32_neon_mlal_lane_cortex_a53
21 k-block: 8
23 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8__aarch32_neon_mlal_lane_ld64
25 k-block: 8
27 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8__aarch32_neon_mlal_lane_prfm_cortex_a7
29 k-block: 8
31 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8__aarch32_neon_mlal_lane_prfm_cortex_a53
33 k-block: 8
35 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8__aarch32_neon_mlal_lane_prfm_ld64
37 k-block: 8
39 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c4__aarch32_neondot_cortex_a55
41 k-block: 8
42 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c4__aarch32_neondot_ld64
44 k-block: 8
47 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8__aarch64_neon_mlal_lane_ld64
49 k-block: 8
51 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8__aarch64_neon_mlal_lane_prfm_ld64
53 k-block: 8
56 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c2__neon_mlal_dup
58 k-block: 16
59 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c2__neon_mlal_ld1r
61 k-block: 16
62 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c2__neon_mlal_ld2r
64 k-block: 16
65 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c2__neon_mlal_ld4r
67 k-block: 16
68 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c2__neon_mull_dup
70 k-block: 8
71 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c2__neon_mull_ld1r
73 k-block: 8
74 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c2__neon_mull_ld2r
76 k-block: 8
77 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c2__neon_mull_ld4r
79 k-block: 8
80 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c2s4__neon_mlal
82 k-block: 16
83 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c2s4__neon_mull
85 k-block: 8
86 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c4__neon_mlal_dup
88 k-block: 16
89 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c4__neon_mlal_ld1r
91 k-block: 16
92 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c4__neon_mlal_ld2r
94 k-block: 16
95 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c4__neon_mull_dup
97 k-block: 8
98 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c4__neon_mull_ld1r
100 k-block: 8
101 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c4__neon_mull_ld2r
103 k-block: 8
104 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c4s2__neon_mlal
106 k-block: 16
107 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c4s2__neon_mull
109 k-block: 8
110 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c2__neon_mlal_dup
112 k-block: 16
113 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c2__neon_mlal_ld1r
115 k-block: 16
116 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c2__neon_mlal_ld2r
118 k-block: 16
119 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c2__neon_mlal_ld4r
121 k-block: 16
122 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c2__neon_mull_dup
124 k-block: 8
125 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c2__neon_mull_ld1r
127 k-block: 8
128 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c2__neon_mull_ld2r
130 k-block: 8
131 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c2__neon_mull_ld4r
133 k-block: 8
134 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c2s4__neon_mlal
136 k-block: 16
137 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c2s4__neon_mull
139 k-block: 8
140 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c4__neon_mlal_dup
142 k-block: 16
143 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c4__neon_mlal_ld1r
145 k-block: 16
146 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c4__neon_mlal_ld2r
148 k-block: 16
149 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c4__neon_mull_dup
151 k-block: 8
152 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c4__neon_mull_ld1r
154 k-block: 8
155 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c4__neon_mull_ld2r
157 k-block: 8
158 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c4s2__neon_mlal
160 k-block: 16
161 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c4s2__neon_mull
163 k-block: 8
164 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c2__neon_mlal_dup
166 k-block: 16
167 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c2__neon_mlal_ld1r
169 k-block: 16
170 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c2__neon_mlal_ld2r
172 k-block: 16
173 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c2__neon_mlal_ld4r
175 k-block: 16
176 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c2__neon_mull_dup
178 k-block: 8
179 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c2__neon_mull_ld1r
181 k-block: 8
182 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c2__neon_mull_ld2r
184 k-block: 8
185 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c2__neon_mull_ld4r
187 k-block: 8
188 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c2s4__neon_mlal
190 k-block: 16
191 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c2s4__neon_mull
193 k-block: 8
194 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c4__neon_mlal_dup
196 k-block: 16
197 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c4__neon_mlal_ld1r
199 k-block: 16
200 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c4__neon_mlal_ld2r
202 k-block: 16
203 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c4__neon_mull_dup
205 k-block: 8
206 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c4__neon_mull_ld1r
208 k-block: 8
209 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c4__neon_mull_ld2r
211 k-block: 8
212 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c4s2__neon_mlal
214 k-block: 16
215 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c4s2__neon_mull
217 k-block: 8
218 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c2__neon_mlal_dup
220 k-block: 16
221 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c2__neon_mlal_ld1r
223 k-block: 16
224 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c2__neon_mlal_ld2r
226 k-block: 16
227 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c2__neon_mlal_ld4r
229 k-block: 16
230 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c2__neon_mull_dup
232 k-block: 8
233 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c2__neon_mull_ld1r
235 k-block: 8
236 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c2__neon_mull_ld2r
238 k-block: 8
239 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c2__neon_mull_ld4r
241 k-block: 8
242 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c2s4__neon_mlal
244 k-block: 16
245 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c2s4__neon_mull
247 k-block: 8
248 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c4__neon_mlal_dup
250 k-block: 16
251 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c4__neon_mlal_ld1r
253 k-block: 16
254 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c4__neon_mlal_ld2r
256 k-block: 16
257 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c4__neon_mull_dup
259 k-block: 8
260 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c4__neon_mull_ld1r
262 k-block: 8
263 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c4__neon_mull_ld2r
265 k-block: 8
266 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c4s2__neon_mlal
268 k-block: 16
269 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c4s2__neon_mull
271 k-block: 8
272 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c2__neon_mlal_dup
274 k-block: 16
275 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c2__neon_mlal_ld1r
277 k-block: 16
278 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c2__neon_mlal_ld2r
280 k-block: 16
281 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c2__neon_mlal_ld4r
283 k-block: 16
284 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c2__neon_mull_dup
286 k-block: 8
287 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c2__neon_mull_ld1r
289 k-block: 8
290 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c2__neon_mull_ld2r
292 k-block: 8
293 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c2__neon_mull_ld4r
295 k-block: 8
296 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c2s4__neon_mlal
298 k-block: 16
299 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c2s4__neon_mull
301 k-block: 8
302 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c4__neon_mlal_dup
304 k-block: 16
305 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c4__neon_mlal_ld1r
307 k-block: 16
308 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c4__neon_mlal_ld2r
310 k-block: 16
311 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c4__neon_mull_dup
313 k-block: 8
314 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c4__neon_mull_ld1r
316 k-block: 8
317 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c4__neon_mull_ld2r
319 k-block: 8
320 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c4s2__neon_mlal
322 k-block: 16
323 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c4s2__neon_mull
325 k-block: 8
326 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c2__neon_mlal_dup
328 k-block: 16
329 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c2__neon_mlal_ld1r
331 k-block: 16
332 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c2__neon_mlal_ld2r
334 k-block: 16
335 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c2__neon_mlal_ld4r
337 k-block: 16
338 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c2__neon_mull_dup
340 k-block: 8
341 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c2__neon_mull_ld1r
343 k-block: 8
344 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c2__neon_mull_ld2r
346 k-block: 8
347 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c2__neon_mull_ld4r
349 k-block: 8
350 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c2s4__neon_mlal
352 k-block: 16
353 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c2s4__neon_mull
355 k-block: 8
356 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c4__neon_mlal_dup
358 k-block: 16
359 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c4__neon_mlal_ld1r
361 k-block: 16
362 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c4__neon_mlal_ld2r
364 k-block: 16
365 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c4__neon_mull_dup
367 k-block: 8
368 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c4__neon_mull_ld1r
370 k-block: 8
371 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c4__neon_mull_ld2r
373 k-block: 8
374 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c4s2__neon_mlal
376 k-block: 16
377 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c4s2__neon_mull
379 k-block: 8
380 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c2__neon_mlal_dup
382 k-block: 16
383 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c2__neon_mlal_ld1r
385 k-block: 16
386 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c2__neon_mlal_ld2r
388 k-block: 16
389 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c2__neon_mlal_ld4r
391 k-block: 16
392 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c2__neon_mull_dup
394 k-block: 8
395 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c2__neon_mull_ld1r
397 k-block: 8
398 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c2__neon_mull_ld2r
400 k-block: 8
401 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c2__neon_mull_ld4r
403 k-block: 8
404 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c2s4__neon_mlal
406 k-block: 16
407 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c2s4__neon_mull
409 k-block: 8
410 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c4__neon_mlal_dup
412 k-block: 16
413 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c4__neon_mlal_ld1r
415 k-block: 16
416 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c4__neon_mlal_ld2r
418 k-block: 16
419 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c4__neon_mull_dup
421 k-block: 8
422 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c4__neon_mull_ld1r
424 k-block: 8
425 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c4__neon_mull_ld2r
427 k-block: 8
428 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c4s2__neon_mlal
430 k-block: 16
431 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c4s2__neon_mull
433 k-block: 8
434 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c2__neon_mlal_dup
436 k-block: 16
437 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c2__neon_mlal_ld1r
439 k-block: 16
440 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c2__neon_mlal_ld2r
442 k-block: 16
443 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c2__neon_mlal_ld4r
445 k-block: 16
446 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c2__neon_mull_dup
448 k-block: 8
449 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c2__neon_mull_ld1r
451 k-block: 8
452 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c2__neon_mull_ld2r
454 k-block: 8
455 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c2__neon_mull_ld4r
457 k-block: 8
458 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c2s4__neon_mlal
460 k-block: 16
461 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c2s4__neon_mull
463 k-block: 8
464 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c4__neon_mlal_dup
466 k-block: 16
467 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c4__neon_mlal_ld1r
469 k-block: 16
470 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c4__neon_mlal_ld2r
472 k-block: 16
473 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c4__neon_mull_dup
475 k-block: 8
476 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c4__neon_mull_ld1r
478 k-block: 8
479 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c4__neon_mull_ld2r
481 k-block: 8
482 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c4s2__neon_mlal
484 k-block: 16
485 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c4s2__neon_mull
487 k-block: 8
489 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c8__aarch64_neon_mlal
491 k-block: 16
493 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c8__aarch64_neon_mlal_cortex_a53
495 k-block: 16
497 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c8__aarch64_neon_mlal_prfm
499 k-block: 16
501 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c8__aarch64_neon_mlal_prfm_cortex_a53
503 k-block: 16
505 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c4__aarch64_neondot_ld32
507 k-block: 4
509 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c4__aarch64_neondot_ld64
511 k-block: 8
513 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c8__aarch64_neon_mlal
515 k-block: 16
517 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c8__aarch64_neon_mlal_cortex_a53
519 k-block: 16
521 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c8__aarch64_neon_mlal_prfm
523 k-block: 16
525 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c8__aarch64_neon_mlal_prfm_cortex_a53
527 k-block: 16
529 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c8__aarch64_neon_mull
531 k-block: 8
533 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c16__aarch64_neon_mlal
535 k-block: 16
537 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16__aarch64_neon_mlal_lane_cortex_a53
539 k-block: 8
541 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16__aarch64_neon_mlal_lane_ld64
543 k-block: 8
545 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16__aarch64_neon_mlal_lane_prfm_cortex_a53
547 k-block: 8
549 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16__aarch64_neon_mlal_lane_prfm_ld64
551 k-block: 8
553 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c4__aarch64_neondot_cortex_a55
555 k-block: 16
557 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c4__aarch64_neondot_ld32
559 k-block: 4
561 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c4__aarch64_neondot_ld64
563 k-block: 8
565 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c4__aarch64_neondot_ld128
567 k-block: 16
570 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8__neon_mlal_lane
572 k-block: 8
573 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8__neon_mlal_lane_prfm
575 k-block: 8
576 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8__neon_mull_addw_dup
578 k-block: 8
579 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c4__neondot
581 k-block: 8
582 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c8__neon_mlal
584 k-block: 16
585 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c8__neon_mull
587 k-block: 8
588 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c16__neon_mlal
590 k-block: 16
591 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x16__neon_mlal_lane
593 k-block: 8
594 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x16__neon_mlal_lane_prfm
596 k-block: 8
597 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x16__neon_mull_addw_dup
599 k-block: 8
600 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c4__neondot
602 k-block: 8
603 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c8__neon_mlal
605 k-block: 16
606 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c8__neon_mull
608 k-block: 8
609 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c16__neon_mlal
611 k-block: 16
612 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8__neon_mlal_lane
614 k-block: 8
615 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8__neon_mlal_lane_prfm
617 k-block: 8
618 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8__neon_mull_addw_dup
620 k-block: 8
621 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c8__neon_mlal
623 k-block: 16
624 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c8__neon_mull
626 k-block: 8
627 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c16__neon_mlal
629 k-block: 16
630 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x16__neon_mlal_lane
632 k-block: 8
633 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x16__neon_mlal_lane_prfm
635 k-block: 8
636 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x16__neon_mull_addw_dup
638 k-block: 8
639 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c8__neon_mlal
641 k-block: 16
642 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c8__neon_mull
644 k-block: 8
645 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c16__neon_mlal
647 k-block: 16
648 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x8__neon_mlal_lane
650 k-block: 8
651 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x8__neon_mlal_lane_prfm
653 k-block: 8
654 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x8__neon_mull_addw_dup
656 k-block: 8
657 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c8__neon_mlal
659 k-block: 16
660 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c8__neon_mull
662 k-block: 8
663 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c16__neon_mlal
665 k-block: 16
666 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x16__neon_mlal_lane
668 k-block: 8
669 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x16__neon_mlal_lane_prfm
671 k-block: 8
672 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x16__neon_mull_addw_dup
674 k-block: 8
675 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c8__neon_mlal
677 k-block: 16
678 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c8__neon_mull
680 k-block: 8
681 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c16__neon_mlal
683 k-block: 16
684 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8__neon_mlal_lane
686 k-block: 8
687 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8__neon_mlal_lane_prfm
689 k-block: 8
690 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8__neon_mull_addw_dup
692 k-block: 8
693 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c4__neondot
695 k-block: 8
696 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c8__neon_mlal
698 k-block: 16
699 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c8__neon_mull
701 k-block: 8
702 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c16__neon_mlal
704 k-block: 16
705 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16__neon_mlal_lane
707 k-block: 8
708 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16__neon_mlal_lane_prfm
710 k-block: 8
711 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16__neon_mull_addw_dup
713 k-block: 8
714 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c4__neondot
716 k-block: 8
717 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c8__neon_mlal
719 k-block: 16
720 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c8__neon_mull
722 k-block: 8
723 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c16__neon_mlal
725 k-block: 16
726 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_6x8__neon_mlal_lane
728 k-block: 8
729 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_6x8__neon_mlal_lane_prfm
731 k-block: 8
732 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_6x8c4__neondot
734 k-block: 8
735 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_6x16__neon_mlal_lane
737 k-block: 8
738 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_6x16__neon_mlal_lane_prfm
740 k-block: 8
741 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_6x16c4__neondot
743 k-block: 8
744 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_8x8c4__neondot
746 k-block: 8
747 - name: xnn_qs8_gemm_minmax_rndnu_ukernel_8x16c4__neondot
749 k-block: 8
751 - name: xnn_generate_qs8_gemm_rndnu_ukernel_4x8__aarch32_neon_mlal_lane_ld64
753 k-block: 8
755 - name: xnn_generate_qs8_gemm_rndnu_ukernel_4x8__aarch32_neon_mlal_lane_prfm_ld64
757 k-block: 8
759 - name: xnn_generate_qs8_gemm_rndnu_ukernel_4x8c4__aarch32_neondot_ld64
761 k-block: 8