Lines Matching full:state
31 static void GEMMBenchmark(benchmark::State& state, in GEMMBenchmark() argument
37 state.SkipWithError("cpuinfo initialization failed"); in GEMMBenchmark()
40 if (isa_check && !isa_check(state)) { in GEMMBenchmark()
44 const size_t mc = state.range(0); in GEMMBenchmark()
45 const size_t nc = state.range(1); in GEMMBenchmark()
46 const size_t kc = state.range(2); in GEMMBenchmark()
79 for (auto _ : state) { in GEMMBenchmark()
80 // Use circular buffers (exceeding cache size) and prefetch to control cache state: in GEMMBenchmark()
84 state.PauseTiming(); in GEMMBenchmark()
87 state.ResumeTiming(); in GEMMBenchmark()
105 state.counters["cpufreq"] = cpu_frequency; in GEMMBenchmark()
108 state.counters["OPS"] = benchmark::Counter( in GEMMBenchmark()
109 uint64_t(state.iterations()) * 2 * mc * nc * kc, benchmark::Counter::kIsRate); in GEMMBenchmark()
112 static void GEMMBenchmark(benchmark::State& state, in GEMMBenchmark() argument
118 state.SkipWithError("cpuinfo initialization failed"); in GEMMBenchmark()
121 if (isa_check && !isa_check(state)) { in GEMMBenchmark()
125 const size_t mc = state.range(0); in GEMMBenchmark()
126 const size_t nc = state.range(1); in GEMMBenchmark()
127 const size_t kc = state.range(2); in GEMMBenchmark()
160 for (auto _ : state) { in GEMMBenchmark()
161 // Use circular buffers (exceeding cache size) and prefetch to control cache state: in GEMMBenchmark()
165 state.PauseTiming(); in GEMMBenchmark()
168 state.ResumeTiming(); in GEMMBenchmark()
186 state.counters["cpufreq"] = cpu_frequency; in GEMMBenchmark()
189 state.counters["OPS"] = benchmark::Counter( in GEMMBenchmark()
190 uint64_t(state.iterations()) * 2 * mc * nc * kc, benchmark::Counter::kIsRate); in GEMMBenchmark()
194 static void RuyBenchmark(benchmark::State& state, size_t threads) in RuyBenchmark() argument
196 const size_t mc = state.range(0); in RuyBenchmark()
197 const size_t nc = state.range(1); in RuyBenchmark()
198 const size_t kc = state.range(2); in RuyBenchmark()
253 for (auto _ : state) { in RuyBenchmark()
254 // Use circular buffers (exceeding cache size) and prefetch to control cache state: in RuyBenchmark()
259 state.PauseTiming(); in RuyBenchmark()
262 state.ResumeTiming(); in RuyBenchmark()
273 state.counters["cpufreq"] = cpu_frequency; in RuyBenchmark()
276 state.counters["OPS"] = benchmark::Counter( in RuyBenchmark()
277 uint64_t(state.iterations()) * 2 * mc * nc * kc, benchmark::Counter::kIsRate); in RuyBenchmark()
280 static void ruy_st(benchmark::State& state, const char* net) in ruy_st() argument
282 RuyBenchmark(state, 1); in ruy_st()
287 static void qs8_gemm_1x8__neon_mlal_lane(benchmark::State& state, const char* net) { in qs8_gemm_1x8__neon_mlal_lane() argument
288 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_1x8__neon_mlal_lane, 1, 8, 1, 1, benchmark::utils… in qs8_gemm_1x8__neon_mlal_lane()
290 static void qs8_gemm_2x8__neon_mlal_lane(benchmark::State& state, const char* net) { in qs8_gemm_2x8__neon_mlal_lane() argument
291 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x8__neon_mlal_lane, 2, 8, 1, 1, benchmark::utils… in qs8_gemm_2x8__neon_mlal_lane()
293 static void qs8_gemm_3x8__neon_mlal_lane(benchmark::State& state, const char* net) { in qs8_gemm_3x8__neon_mlal_lane() argument
294 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_3x8__neon_mlal_lane, 3, 8, 1, 1, benchmark::utils… in qs8_gemm_3x8__neon_mlal_lane()
296 static void qs8_gemm_4x8__neon_mlal_lane(benchmark::State& state, const char* net) { in qs8_gemm_4x8__neon_mlal_lane() argument
297 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x8__neon_mlal_lane, 4, 8, 1, 1, benchmark::utils… in qs8_gemm_4x8__neon_mlal_lane()
299 static void qs8_gemm_1x16__neon_mlal_lane(benchmark::State& state, const char* net) { in qs8_gemm_1x16__neon_mlal_lane() argument
300 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_1x16__neon_mlal_lane, 1, 16, 1, 1, benchmark::uti… in qs8_gemm_1x16__neon_mlal_lane()
302 static void qs8_gemm_2x16__neon_mlal_lane(benchmark::State& state, const char* net) { in qs8_gemm_2x16__neon_mlal_lane() argument
303 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x16__neon_mlal_lane, 2, 16, 1, 1, benchmark::uti… in qs8_gemm_2x16__neon_mlal_lane()
305 static void qs8_gemm_3x16__neon_mlal_lane(benchmark::State& state, const char* net) { in qs8_gemm_3x16__neon_mlal_lane() argument
306 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_3x16__neon_mlal_lane, 3, 16, 1, 1, benchmark::uti… in qs8_gemm_3x16__neon_mlal_lane()
308 static void qs8_gemm_4x16__neon_mlal_lane(benchmark::State& state, const char* net) { in qs8_gemm_4x16__neon_mlal_lane() argument
309 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x16__neon_mlal_lane, 4, 16, 1, 1, benchmark::uti… in qs8_gemm_4x16__neon_mlal_lane()
311 static void qs8_gemm_1x8__neon_mull_addw_dup(benchmark::State& state, const char* net) { in qs8_gemm_1x8__neon_mull_addw_dup() argument
312 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_1x8__neon_mull_addw_dup, 1, 8, 1, 1, benchmark::u… in qs8_gemm_1x8__neon_mull_addw_dup()
314 static void qs8_gemm_2x8__neon_mull_addw_dup(benchmark::State& state, const char* net) { in qs8_gemm_2x8__neon_mull_addw_dup() argument
315 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x8__neon_mull_addw_dup, 2, 8, 1, 1, benchmark::u… in qs8_gemm_2x8__neon_mull_addw_dup()
317 static void qs8_gemm_3x8__neon_mull_addw_dup(benchmark::State& state, const char* net) { in qs8_gemm_3x8__neon_mull_addw_dup() argument
318 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_3x8__neon_mull_addw_dup, 3, 8, 1, 1, benchmark::u… in qs8_gemm_3x8__neon_mull_addw_dup()
320 static void qs8_gemm_4x8__neon_mull_addw_dup(benchmark::State& state, const char* net) { in qs8_gemm_4x8__neon_mull_addw_dup() argument
321 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x8__neon_mull_addw_dup, 4, 8, 1, 1, benchmark::u… in qs8_gemm_4x8__neon_mull_addw_dup()
323 static void qs8_gemm_1x16__neon_mull_addw_dup(benchmark::State& state, const char* net) { in qs8_gemm_1x16__neon_mull_addw_dup() argument
324 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_1x16__neon_mull_addw_dup, 1, 16, 1, 1, benchmark:… in qs8_gemm_1x16__neon_mull_addw_dup()
326 static void qs8_gemm_2x16__neon_mull_addw_dup(benchmark::State& state, const char* net) { in qs8_gemm_2x16__neon_mull_addw_dup() argument
327 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x16__neon_mull_addw_dup, 2, 16, 1, 1, benchmark:… in qs8_gemm_2x16__neon_mull_addw_dup()
329 static void qs8_gemm_3x16__neon_mull_addw_dup(benchmark::State& state, const char* net) { in qs8_gemm_3x16__neon_mull_addw_dup() argument
330 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_3x16__neon_mull_addw_dup, 3, 16, 1, 1, benchmark:… in qs8_gemm_3x16__neon_mull_addw_dup()
332 static void qs8_gemm_4x16__neon_mull_addw_dup(benchmark::State& state, const char* net) { in qs8_gemm_4x16__neon_mull_addw_dup() argument
333 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x16__neon_mull_addw_dup, 4, 16, 1, 1, benchmark:… in qs8_gemm_4x16__neon_mull_addw_dup()
335 static void qs8_gemm_1x8c2__neon_mull_padal_dup(benchmark::State& state, const char* net) { in qs8_gemm_1x8c2__neon_mull_padal_dup() argument
336 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_1x8c2__neon_mull_padal_dup, 1, 8, 2, 1, benchmark… in qs8_gemm_1x8c2__neon_mull_padal_dup()
338 static void qs8_gemm_2x8c2__neon_mull_padal_dup(benchmark::State& state, const char* net) { in qs8_gemm_2x8c2__neon_mull_padal_dup() argument
339 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x8c2__neon_mull_padal_dup, 2, 8, 2, 1, benchmark… in qs8_gemm_2x8c2__neon_mull_padal_dup()
341 static void qs8_gemm_3x8c2__neon_mull_padal_dup(benchmark::State& state, const char* net) { in qs8_gemm_3x8c2__neon_mull_padal_dup() argument
342 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_3x8c2__neon_mull_padal_dup, 3, 8, 2, 1, benchmark… in qs8_gemm_3x8c2__neon_mull_padal_dup()
344 static void qs8_gemm_4x8c2__neon_mull_padal_dup(benchmark::State& state, const char* net) { in qs8_gemm_4x8c2__neon_mull_padal_dup() argument
345 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x8c2__neon_mull_padal_dup, 4, 8, 2, 1, benchmark… in qs8_gemm_4x8c2__neon_mull_padal_dup()
347 static void qs8_gemm_1x16c2__neon_mull_padal_dup(benchmark::State& state, const char* net) { in qs8_gemm_1x16c2__neon_mull_padal_dup() argument
348 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_1x16c2__neon_mull_padal_dup, 1, 16, 2, 1, benchma… in qs8_gemm_1x16c2__neon_mull_padal_dup()
350 static void qs8_gemm_2x16c2__neon_mull_padal_dup(benchmark::State& state, const char* net) { in qs8_gemm_2x16c2__neon_mull_padal_dup() argument
351 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x16c2__neon_mull_padal_dup, 2, 16, 2, 1, benchma… in qs8_gemm_2x16c2__neon_mull_padal_dup()
353 static void qs8_gemm_3x16c2__neon_mull_padal_dup(benchmark::State& state, const char* net) { in qs8_gemm_3x16c2__neon_mull_padal_dup() argument
354 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_3x16c2__neon_mull_padal_dup, 3, 16, 2, 1, benchma… in qs8_gemm_3x16c2__neon_mull_padal_dup()
356 static void qs8_gemm_4x16c2__neon_mull_padal_dup(benchmark::State& state, const char* net) { in qs8_gemm_4x16c2__neon_mull_padal_dup() argument
357 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x16c2__neon_mull_padal_dup, 4, 16, 2, 1, benchma… in qs8_gemm_4x16c2__neon_mull_padal_dup()
359 static void qs8_gemm_1x8c2__neon_mlal_padal_dup(benchmark::State& state, const char* net) { in qs8_gemm_1x8c2__neon_mlal_padal_dup() argument
360 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_1x8c2__neon_mlal_padal_dup, 1, 8, 2, 1, benchmark… in qs8_gemm_1x8c2__neon_mlal_padal_dup()
362 static void qs8_gemm_2x8c2__neon_mlal_padal_dup(benchmark::State& state, const char* net) { in qs8_gemm_2x8c2__neon_mlal_padal_dup() argument
363 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x8c2__neon_mlal_padal_dup, 2, 8, 2, 1, benchmark… in qs8_gemm_2x8c2__neon_mlal_padal_dup()
365 static void qs8_gemm_3x8c2__neon_mlal_padal_dup(benchmark::State& state, const char* net) { in qs8_gemm_3x8c2__neon_mlal_padal_dup() argument
366 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_3x8c2__neon_mlal_padal_dup, 3, 8, 2, 1, benchmark… in qs8_gemm_3x8c2__neon_mlal_padal_dup()
368 static void qs8_gemm_4x8c2__neon_mlal_padal_dup(benchmark::State& state, const char* net) { in qs8_gemm_4x8c2__neon_mlal_padal_dup() argument
369 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x8c2__neon_mlal_padal_dup, 4, 8, 2, 1, benchmark… in qs8_gemm_4x8c2__neon_mlal_padal_dup()
371 static void qs8_gemm_1x16c2__neon_mlal_padal_dup(benchmark::State& state, const char* net) { in qs8_gemm_1x16c2__neon_mlal_padal_dup() argument
372 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_1x16c2__neon_mlal_padal_dup, 1, 16, 2, 1, benchma… in qs8_gemm_1x16c2__neon_mlal_padal_dup()
374 static void qs8_gemm_2x16c2__neon_mlal_padal_dup(benchmark::State& state, const char* net) { in qs8_gemm_2x16c2__neon_mlal_padal_dup() argument
375 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x16c2__neon_mlal_padal_dup, 2, 16, 2, 1, benchma… in qs8_gemm_2x16c2__neon_mlal_padal_dup()
377 static void qs8_gemm_3x16c2__neon_mlal_padal_dup(benchmark::State& state, const char* net) { in qs8_gemm_3x16c2__neon_mlal_padal_dup() argument
378 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_3x16c2__neon_mlal_padal_dup, 3, 16, 2, 1, benchma… in qs8_gemm_3x16c2__neon_mlal_padal_dup()
380 static void qs8_gemm_4x16c2__neon_mlal_padal_dup(benchmark::State& state, const char* net) { in qs8_gemm_4x16c2__neon_mlal_padal_dup() argument
381 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x16c2__neon_mlal_padal_dup, 4, 16, 2, 1, benchma… in qs8_gemm_4x16c2__neon_mlal_padal_dup()
383 static void qs8_gemm_1x8c8__neon_mull_padal(benchmark::State& state, const char* net) { in qs8_gemm_1x8c8__neon_mull_padal() argument
384 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_1x8c8__neon_mull_padal, 1, 8, 8, 1, benchmark::ut… in qs8_gemm_1x8c8__neon_mull_padal()
386 static void qs8_gemm_2x8c8__neon_mull_padal(benchmark::State& state, const char* net) { in qs8_gemm_2x8c8__neon_mull_padal() argument
387 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x8c8__neon_mull_padal, 2, 8, 8, 1, benchmark::ut… in qs8_gemm_2x8c8__neon_mull_padal()
389 static void qs8_gemm_3x8c8__neon_mull_padal(benchmark::State& state, const char* net) { in qs8_gemm_3x8c8__neon_mull_padal() argument
390 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_3x8c8__neon_mull_padal, 3, 8, 8, 1, benchmark::ut… in qs8_gemm_3x8c8__neon_mull_padal()
392 static void qs8_gemm_4x8c8__neon_mull_padal(benchmark::State& state, const char* net) { in qs8_gemm_4x8c8__neon_mull_padal() argument
393 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mull_padal, 4, 8, 8, 1, benchmark::ut… in qs8_gemm_4x8c8__neon_mull_padal()
395 static void qs8_gemm_1x16c8__neon_mull_padal(benchmark::State& state, const char* net) { in qs8_gemm_1x16c8__neon_mull_padal() argument
396 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_1x16c8__neon_mull_padal, 1, 16, 8, 1, benchmark::… in qs8_gemm_1x16c8__neon_mull_padal()
398 static void qs8_gemm_2x16c8__neon_mull_padal(benchmark::State& state, const char* net) { in qs8_gemm_2x16c8__neon_mull_padal() argument
399 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x16c8__neon_mull_padal, 2, 16, 8, 1, benchmark::… in qs8_gemm_2x16c8__neon_mull_padal()
401 static void qs8_gemm_3x16c8__neon_mull_padal(benchmark::State& state, const char* net) { in qs8_gemm_3x16c8__neon_mull_padal() argument
402 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mull_padal, 3, 16, 8, 1, benchmark::… in qs8_gemm_3x16c8__neon_mull_padal()
404 static void qs8_gemm_4x16c8__neon_mull_padal(benchmark::State& state, const char* net) { in qs8_gemm_4x16c8__neon_mull_padal() argument
405 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mull_padal, 4, 16, 8, 1, benchmark::… in qs8_gemm_4x16c8__neon_mull_padal()
407 static void qs8_gemm_1x8c8__neon_mlal_padal(benchmark::State& state, const char* net) { in qs8_gemm_1x8c8__neon_mlal_padal() argument
408 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_1x8c8__neon_mlal_padal, 1, 8, 8, 1, benchmark::ut… in qs8_gemm_1x8c8__neon_mlal_padal()
410 static void qs8_gemm_2x8c8__neon_mlal_padal(benchmark::State& state, const char* net) { in qs8_gemm_2x8c8__neon_mlal_padal() argument
411 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x8c8__neon_mlal_padal, 2, 8, 8, 1, benchmark::ut… in qs8_gemm_2x8c8__neon_mlal_padal()
413 static void qs8_gemm_3x8c8__neon_mlal_padal(benchmark::State& state, const char* net) { in qs8_gemm_3x8c8__neon_mlal_padal() argument
414 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_3x8c8__neon_mlal_padal, 3, 8, 8, 1, benchmark::ut… in qs8_gemm_3x8c8__neon_mlal_padal()
416 static void qs8_gemm_4x8c8__neon_mlal_padal(benchmark::State& state, const char* net) { in qs8_gemm_4x8c8__neon_mlal_padal() argument
417 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mlal_padal, 4, 8, 8, 1, benchmark::ut… in qs8_gemm_4x8c8__neon_mlal_padal()
419 static void qs8_gemm_1x16c8__neon_mlal_padal(benchmark::State& state, const char* net) { in qs8_gemm_1x16c8__neon_mlal_padal() argument
420 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_1x16c8__neon_mlal_padal, 1, 16, 8, 1, benchmark::… in qs8_gemm_1x16c8__neon_mlal_padal()
422 static void qs8_gemm_2x16c8__neon_mlal_padal(benchmark::State& state, const char* net) { in qs8_gemm_2x16c8__neon_mlal_padal() argument
423 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x16c8__neon_mlal_padal, 2, 16, 8, 1, benchmark::… in qs8_gemm_2x16c8__neon_mlal_padal()
425 static void qs8_gemm_3x16c8__neon_mlal_padal(benchmark::State& state, const char* net) { in qs8_gemm_3x16c8__neon_mlal_padal() argument
426 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal, 3, 16, 8, 1, benchmark::… in qs8_gemm_3x16c8__neon_mlal_padal()
428 static void qs8_gemm_4x16c8__neon_mlal_padal(benchmark::State& state, const char* net) { in qs8_gemm_4x16c8__neon_mlal_padal() argument
429 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal, 4, 16, 8, 1, benchmark::… in qs8_gemm_4x16c8__neon_mlal_padal()
431 static void qs8_gemm_1x8c16__neon_mlal_padal(benchmark::State& state, const char* net) { in qs8_gemm_1x8c16__neon_mlal_padal() argument
432 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_1x8c16__neon_mlal_padal, 1, 8, 16, 1, benchmark::… in qs8_gemm_1x8c16__neon_mlal_padal()
434 static void qs8_gemm_2x8c16__neon_mlal_padal(benchmark::State& state, const char* net) { in qs8_gemm_2x8c16__neon_mlal_padal() argument
435 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x8c16__neon_mlal_padal, 2, 8, 16, 1, benchmark::… in qs8_gemm_2x8c16__neon_mlal_padal()
437 static void qs8_gemm_3x8c16__neon_mlal_padal(benchmark::State& state, const char* net) { in qs8_gemm_3x8c16__neon_mlal_padal() argument
438 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_3x8c16__neon_mlal_padal, 3, 8, 16, 1, benchmark::… in qs8_gemm_3x8c16__neon_mlal_padal()
440 static void qs8_gemm_4x8c16__neon_mlal_padal(benchmark::State& state, const char* net) { in qs8_gemm_4x8c16__neon_mlal_padal() argument
441 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x8c16__neon_mlal_padal, 4, 8, 16, 1, benchmark::… in qs8_gemm_4x8c16__neon_mlal_padal()
443 static void qs8_gemm_1x16c16__neon_mlal_padal(benchmark::State& state, const char* net) { in qs8_gemm_1x16c16__neon_mlal_padal() argument
444 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_1x16c16__neon_mlal_padal, 1, 16, 16, 1, benchmark… in qs8_gemm_1x16c16__neon_mlal_padal()
446 static void qs8_gemm_2x16c16__neon_mlal_padal(benchmark::State& state, const char* net) { in qs8_gemm_2x16c16__neon_mlal_padal() argument
447 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal, 2, 16, 16, 1, benchmark… in qs8_gemm_2x16c16__neon_mlal_padal()
449 static void qs8_gemm_3x16c16__neon_mlal_padal(benchmark::State& state, const char* net) { in qs8_gemm_3x16c16__neon_mlal_padal() argument
450 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_3x16c16__neon_mlal_padal, 3, 16, 16, 1, benchmark… in qs8_gemm_3x16c16__neon_mlal_padal()
452 static void qs8_gemm_4x16c16__neon_mlal_padal(benchmark::State& state, const char* net) { in qs8_gemm_4x16c16__neon_mlal_padal() argument
453 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x16c16__neon_mlal_padal, 4, 16, 16, 1, benchmark… in qs8_gemm_4x16c16__neon_mlal_padal()
455 static void qs8_gemm_1x8c4__neondot(benchmark::State& state, const char* net) { in qs8_gemm_1x8c4__neondot() argument
456 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_1x8c4__neondot, 1, 8, 4, 1, benchmark::utils::Che… in qs8_gemm_1x8c4__neondot()
458 static void qs8_gemm_4x8c4__neondot(benchmark::State& state, const char* net) { in qs8_gemm_4x8c4__neondot() argument
459 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x8c4__neondot, 4, 8, 4, 1, benchmark::utils::Che… in qs8_gemm_4x8c4__neondot()
461 static void qs8_gemm_6x8c4__neondot(benchmark::State& state, const char* net) { in qs8_gemm_6x8c4__neondot() argument
462 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_6x8c4__neondot, 6, 8, 4, 1, benchmark::utils::Che… in qs8_gemm_6x8c4__neondot()
464 static void qs8_gemm_8x8c4__neondot(benchmark::State& state, const char* net) { in qs8_gemm_8x8c4__neondot() argument
465 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_8x8c4__neondot, 8, 8, 4, 1, benchmark::utils::Che… in qs8_gemm_8x8c4__neondot()
467 static void qs8_gemm_1x16c4__neondot(benchmark::State& state, const char* net) { in qs8_gemm_1x16c4__neondot() argument
468 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_1x16c4__neondot, 1, 16, 4, 1, benchmark::utils::C… in qs8_gemm_1x16c4__neondot()
470 static void qs8_gemm_4x16c4__neondot(benchmark::State& state, const char* net) { in qs8_gemm_4x16c4__neondot() argument
471 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x16c4__neondot, 4, 16, 4, 1, benchmark::utils::C… in qs8_gemm_4x16c4__neondot()
473 static void qs8_gemm_6x16c4__neondot(benchmark::State& state, const char* net) { in qs8_gemm_6x16c4__neondot() argument
474 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_6x16c4__neondot, 6, 16, 4, 1, benchmark::utils::C… in qs8_gemm_6x16c4__neondot()
476 static void qs8_gemm_8x16c4__neondot(benchmark::State& state, const char* net) { in qs8_gemm_8x16c4__neondot() argument
477 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_8x16c4__neondot, 8, 16, 4, 1, benchmark::utils::C… in qs8_gemm_8x16c4__neondot()
547 …static void qs8_gemm_4x16c4__aarch64_neondot_cortex_a55(benchmark::State& state, const char* net) { in BENCHMARK_GEMM()
548 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x16c4__aarch64_neondot_cortex_a55, 4, 16, 4, 1, … in BENCHMARK_GEMM()
550 static void qs8_gemm_1x16c4__aarch64_neondot_ld32(benchmark::State& state, const char* net) { in qs8_gemm_1x16c4__aarch64_neondot_ld32() argument
551 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_1x16c4__aarch64_neondot_ld32, 1, 16, 4, 1, benchm… in qs8_gemm_1x16c4__aarch64_neondot_ld32()
553 static void qs8_gemm_1x16c4__aarch64_neondot_ld64(benchmark::State& state, const char* net) { in qs8_gemm_1x16c4__aarch64_neondot_ld64() argument
554 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_1x16c4__aarch64_neondot_ld64, 1, 16, 4, 1, benchm… in qs8_gemm_1x16c4__aarch64_neondot_ld64()
556 static void qs8_gemm_4x16c4__aarch64_neondot_ld32(benchmark::State& state, const char* net) { in qs8_gemm_4x16c4__aarch64_neondot_ld32() argument
557 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x16c4__aarch64_neondot_ld32, 4, 16, 4, 1, benchm… in qs8_gemm_4x16c4__aarch64_neondot_ld32()
559 static void qs8_gemm_4x16c4__aarch64_neondot_ld64(benchmark::State& state, const char* net) { in qs8_gemm_4x16c4__aarch64_neondot_ld64() argument
560 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x16c4__aarch64_neondot_ld64, 4, 16, 4, 1, benchm… in qs8_gemm_4x16c4__aarch64_neondot_ld64()
562 static void qs8_gemm_2x8c8__aarch64_neon_mull_padal(benchmark::State& state, const char* net) { in qs8_gemm_2x8c8__aarch64_neon_mull_padal() argument
563 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x8c8__aarch64_neon_mull_padal, 2, 8, 8, 1, bench… in qs8_gemm_2x8c8__aarch64_neon_mull_padal()
565 static void qs8_gemm_2x8c8__aarch64_neon_mlal_padal(benchmark::State& state, const char* net) { in qs8_gemm_2x8c8__aarch64_neon_mlal_padal() argument
566 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x8c8__aarch64_neon_mlal_padal, 2, 8, 8, 1, bench… in qs8_gemm_2x8c8__aarch64_neon_mlal_padal()
568 static void qs8_gemm_2x8c16__aarch64_neon_mlal_padal(benchmark::State& state, const char* net) { in qs8_gemm_2x8c16__aarch64_neon_mlal_padal() argument
569 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x8c16__aarch64_neon_mlal_padal, 2, 8, 16, 1, ben… in qs8_gemm_2x8c16__aarch64_neon_mlal_padal()
583 static void qs8_gemm_4x4c2__sse2_ld64(benchmark::State& state, const char* net) { in BENCHMARK_GEMM()
584 GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x4c2__sse2_ld64, 4, 4, 2, 1); in BENCHMARK_GEMM()
586 static void qs8_gemm_4x4c2__ssse3_ld64(benchmark::State& state, const char* net) { in qs8_gemm_4x4c2__ssse3_ld64() argument
587 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x4c2__ssse3_ld64, 4, 4, 2, 1, benchmark::utils::… in qs8_gemm_4x4c2__ssse3_ld64()
589 static void qs8_gemm_4x4c2__sse41_ld64(benchmark::State& state, const char* net) { in qs8_gemm_4x4c2__sse41_ld64() argument
590 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x4c2__sse41_ld64, 4, 4, 2, 1, benchmark::utils::… in qs8_gemm_4x4c2__sse41_ld64()
592 static void qs8_gemm_4x4c2__xop_ld64(benchmark::State& state, const char* net) { in qs8_gemm_4x4c2__xop_ld64() argument
593 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x4c2__xop_ld64, 4, 4, 2, 1, benchmark::utils::Ch… in qs8_gemm_4x4c2__xop_ld64()
596 static void qs8_gemm_4x4c2__sse2_ld128(benchmark::State& state, const char* net) { in qs8_gemm_4x4c2__sse2_ld128() argument
597 GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x4c2__sse2_ld128, 4, 4, 2, 1); in qs8_gemm_4x4c2__sse2_ld128()
599 static void qs8_gemm_4x4c2__ssse3_ld128(benchmark::State& state, const char* net) { in qs8_gemm_4x4c2__ssse3_ld128() argument
600 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x4c2__ssse3_ld128, 4, 4, 2, 1, benchmark::utils:… in qs8_gemm_4x4c2__ssse3_ld128()
602 static void qs8_gemm_4x4c2__sse41_ld128(benchmark::State& state, const char* net) { in qs8_gemm_4x4c2__sse41_ld128() argument
603 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x4c2__sse41_ld128, 4, 4, 2, 1, benchmark::utils:… in qs8_gemm_4x4c2__sse41_ld128()
605 static void qs8_gemm_4x4c2__xop_ld128(benchmark::State& state, const char* net) { in qs8_gemm_4x4c2__xop_ld128() argument
606 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x4c2__xop_ld128, 4, 4, 2, 1, benchmark::utils::C… in qs8_gemm_4x4c2__xop_ld128()
609 static void qs8_gemm_xw_4x4c2__sse2(benchmark::State& state, const char* net) { in qs8_gemm_xw_4x4c2__sse2() argument
610 GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_ukernel_4x4c2__sse2, 4, 4, 2, 1); in qs8_gemm_xw_4x4c2__sse2()
612 static void qs8_gemm_xw_4x4c2__ssse3(benchmark::State& state, const char* net) { in qs8_gemm_xw_4x4c2__ssse3() argument
613 …GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_ukernel_4x4c2__ssse3, 4, 4, 2, 1, benchmark::utils::Ch… in qs8_gemm_xw_4x4c2__ssse3()
615 static void qs8_gemm_xw_4x4c2__sse41(benchmark::State& state, const char* net) { in qs8_gemm_xw_4x4c2__sse41() argument
616 …GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_ukernel_4x4c2__sse41, 4, 4, 2, 1, benchmark::utils::Ch… in qs8_gemm_xw_4x4c2__sse41()
618 static void qs8_gemm_xw_4x4c2__xop(benchmark::State& state, const char* net) { in qs8_gemm_xw_4x4c2__xop() argument
619 …GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_ukernel_4x4c2__xop, 4, 4, 2, 1, benchmark::utils::Chec… in qs8_gemm_xw_4x4c2__xop()
622 static void qs8_gemm_2x4c8__sse2_ld64(benchmark::State& state, const char* net) { in qs8_gemm_2x4c8__sse2_ld64() argument
623 GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x4c8__sse2_ld64, 2, 4, 8, 1); in qs8_gemm_2x4c8__sse2_ld64()
625 static void qs8_gemm_2x4c8__ssse3_ld64(benchmark::State& state, const char* net) { in qs8_gemm_2x4c8__ssse3_ld64() argument
626 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x4c8__ssse3_ld64, 2, 4, 8, 1, benchmark::utils::… in qs8_gemm_2x4c8__ssse3_ld64()
628 static void qs8_gemm_2x4c8__sse41_ld64(benchmark::State& state, const char* net) { in qs8_gemm_2x4c8__sse41_ld64() argument
629 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x4c8__sse41_ld64, 2, 4, 8, 1, benchmark::utils::… in qs8_gemm_2x4c8__sse41_ld64()
631 static void qs8_gemm_2x4c8__xop_ld64(benchmark::State& state, const char* net) { in qs8_gemm_2x4c8__xop_ld64() argument
632 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x4c8__xop_ld64, 2, 4, 8, 1, benchmark::utils::Ch… in qs8_gemm_2x4c8__xop_ld64()
635 static void qs8_gemm_2x4c8__sse2_ld128(benchmark::State& state, const char* net) { in qs8_gemm_2x4c8__sse2_ld128() argument
636 GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x4c8__sse2_ld128, 2, 4, 8, 1); in qs8_gemm_2x4c8__sse2_ld128()
638 static void qs8_gemm_2x4c8__ssse3_ld128(benchmark::State& state, const char* net) { in qs8_gemm_2x4c8__ssse3_ld128() argument
639 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x4c8__ssse3_ld128, 2, 4, 8, 1, benchmark::utils:… in qs8_gemm_2x4c8__ssse3_ld128()
641 static void qs8_gemm_2x4c8__sse41_ld128(benchmark::State& state, const char* net) { in qs8_gemm_2x4c8__sse41_ld128() argument
642 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x4c8__sse41_ld128, 2, 4, 8, 1, benchmark::utils:… in qs8_gemm_2x4c8__sse41_ld128()
644 static void qs8_gemm_2x4c8__xop_ld128(benchmark::State& state, const char* net) { in qs8_gemm_2x4c8__xop_ld128() argument
645 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x4c8__xop_ld128, 2, 4, 8, 1, benchmark::utils::C… in qs8_gemm_2x4c8__xop_ld128()
648 static void qs8_gemm_xw_2x4c8__sse2(benchmark::State& state, const char* net) { in qs8_gemm_xw_2x4c8__sse2() argument
649 GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_ukernel_2x4c8__sse2, 2, 4, 8, 1); in qs8_gemm_xw_2x4c8__sse2()
651 static void qs8_gemm_xw_2x4c8__ssse3(benchmark::State& state, const char* net) { in qs8_gemm_xw_2x4c8__ssse3() argument
652 …GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_ukernel_2x4c8__ssse3, 2, 4, 8, 1, benchmark::utils::Ch… in qs8_gemm_xw_2x4c8__ssse3()
654 static void qs8_gemm_xw_2x4c8__sse41(benchmark::State& state, const char* net) { in qs8_gemm_xw_2x4c8__sse41() argument
655 …GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_ukernel_2x4c8__sse41, 2, 4, 8, 1, benchmark::utils::Ch… in qs8_gemm_xw_2x4c8__sse41()
657 static void qs8_gemm_xw_2x4c8__xop(benchmark::State& state, const char* net) { in qs8_gemm_xw_2x4c8__xop() argument
658 …GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_ukernel_2x4c8__xop, 2, 4, 8, 1, benchmark::utils::Chec… in qs8_gemm_xw_2x4c8__xop()
661 static void qs8_gemm_3x4c8__sse2_ld64(benchmark::State& state, const char* net) { in qs8_gemm_3x4c8__sse2_ld64() argument
662 GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_3x4c8__sse2_ld64, 3, 4, 8, 1); in qs8_gemm_3x4c8__sse2_ld64()
664 static void qs8_gemm_3x4c8__ssse3_ld64(benchmark::State& state, const char* net) { in qs8_gemm_3x4c8__ssse3_ld64() argument
665 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_3x4c8__ssse3_ld64, 3, 4, 8, 1, benchmark::utils::… in qs8_gemm_3x4c8__ssse3_ld64()
667 static void qs8_gemm_3x4c8__sse41_ld64(benchmark::State& state, const char* net) { in qs8_gemm_3x4c8__sse41_ld64() argument
668 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_3x4c8__sse41_ld64, 3, 4, 8, 1, benchmark::utils::… in qs8_gemm_3x4c8__sse41_ld64()
670 static void qs8_gemm_3x4c8__xop_ld64(benchmark::State& state, const char* net) { in qs8_gemm_3x4c8__xop_ld64() argument
671 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_3x4c8__xop_ld64, 3, 4, 8, 1, benchmark::utils::Ch… in qs8_gemm_3x4c8__xop_ld64()
674 static void qs8_gemm_3x4c8__sse2_ld128(benchmark::State& state, const char* net) { in qs8_gemm_3x4c8__sse2_ld128() argument
675 GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_3x4c8__sse2_ld128, 3, 4, 8, 1); in qs8_gemm_3x4c8__sse2_ld128()
677 static void qs8_gemm_3x4c8__ssse3_ld128(benchmark::State& state, const char* net) { in qs8_gemm_3x4c8__ssse3_ld128() argument
678 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_3x4c8__ssse3_ld128, 3, 4, 8, 1, benchmark::utils:… in qs8_gemm_3x4c8__ssse3_ld128()
680 static void qs8_gemm_3x4c8__sse41_ld128(benchmark::State& state, const char* net) { in qs8_gemm_3x4c8__sse41_ld128() argument
681 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_3x4c8__sse41_ld128, 3, 4, 8, 1, benchmark::utils:… in qs8_gemm_3x4c8__sse41_ld128()
683 static void qs8_gemm_3x4c8__xop_ld128(benchmark::State& state, const char* net) { in qs8_gemm_3x4c8__xop_ld128() argument
684 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_3x4c8__xop_ld128, 3, 4, 8, 1, benchmark::utils::C… in qs8_gemm_3x4c8__xop_ld128()
687 static void qs8_gemm_xw_3x4c8__sse2(benchmark::State& state, const char* net) { in qs8_gemm_xw_3x4c8__sse2() argument
688 GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_ukernel_3x4c8__sse2, 3, 4, 8, 1); in qs8_gemm_xw_3x4c8__sse2()
690 static void qs8_gemm_xw_3x4c8__ssse3(benchmark::State& state, const char* net) { in qs8_gemm_xw_3x4c8__ssse3() argument
691 …GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_ukernel_3x4c8__ssse3, 3, 4, 8, 1, benchmark::utils::Ch… in qs8_gemm_xw_3x4c8__ssse3()
693 static void qs8_gemm_xw_3x4c8__sse41(benchmark::State& state, const char* net) { in qs8_gemm_xw_3x4c8__sse41() argument
694 …GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_ukernel_3x4c8__sse41, 3, 4, 8, 1, benchmark::utils::Ch… in qs8_gemm_xw_3x4c8__sse41()
696 static void qs8_gemm_xw_3x4c8__xop(benchmark::State& state, const char* net) { in qs8_gemm_xw_3x4c8__xop() argument
697 …GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_ukernel_3x4c8__xop, 3, 4, 8, 1, benchmark::utils::Chec… in qs8_gemm_xw_3x4c8__xop()
700 static void qs8_gemm_2x8c8__avx2(benchmark::State& state, const char* net) { in qs8_gemm_2x8c8__avx2() argument
701 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x8c8__avx2, 2, 8, 8, 1, benchmark::utils::CheckA… in qs8_gemm_2x8c8__avx2()
703 static void qs8_gemm_3x8c8__avx2(benchmark::State& state, const char* net) { in qs8_gemm_3x8c8__avx2() argument
704 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_3x8c8__avx2, 3, 8, 8, 1, benchmark::utils::CheckA… in qs8_gemm_3x8c8__avx2()
707 static void qs8_gemm_2x16c8__avx512skx(benchmark::State& state, const char* net) { in qs8_gemm_2x16c8__avx512skx() argument
708 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x16c8__avx512skx, 2, 16, 8, 1, benchmark::utils:… in qs8_gemm_2x16c8__avx512skx()
710 static void qs8_gemm_3x16c8__avx512skx(benchmark::State& state, const char* net) { in qs8_gemm_3x16c8__avx512skx() argument
711 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_3x16c8__avx512skx, 3, 16, 8, 1, benchmark::utils:… in qs8_gemm_3x16c8__avx512skx()
713 static void qs8_gemm_4x16c8__avx512skx(benchmark::State& state, const char* net) { in qs8_gemm_4x16c8__avx512skx() argument
714 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_4x16c8__avx512skx, 4, 16, 8, 1, benchmark::utils:… in qs8_gemm_4x16c8__avx512skx()
717 static void qs8_gemm_xw_2x8c8__avx2(benchmark::State& state, const char* net) { in qs8_gemm_xw_2x8c8__avx2() argument
718 …GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_ukernel_2x8c8__avx2, 2, 8, 8, 1, benchmark::utils::Che… in qs8_gemm_xw_2x8c8__avx2()
720 static void qs8_gemm_xw_3x8c8__avx2(benchmark::State& state, const char* net) { in qs8_gemm_xw_3x8c8__avx2() argument
721 …GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_ukernel_3x8c8__avx2, 3, 8, 8, 1, benchmark::utils::Che… in qs8_gemm_xw_3x8c8__avx2()
775 static void qs8_gemm_2x4c8__wasmsimd_ld64(benchmark::State& state, const char* net) { in BENCHMARK_GEMM()
776 GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld64, 2, 4, 8, 1); in BENCHMARK_GEMM()
778 static void qs8_gemm_3x4c8__wasmsimd_ld64(benchmark::State& state, const char* net) { in qs8_gemm_3x4c8__wasmsimd_ld64() argument
779 GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld64, 3, 4, 8, 1); in qs8_gemm_3x4c8__wasmsimd_ld64()
782 static void qs8_gemm_2x4c8__wasmsimd_ld128(benchmark::State& state, const char* net) { in qs8_gemm_2x4c8__wasmsimd_ld128() argument
783 GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld128, 2, 4, 8, 1); in qs8_gemm_2x4c8__wasmsimd_ld128()
785 static void qs8_gemm_3x4c8__wasmsimd_ld128(benchmark::State& state, const char* net) { in qs8_gemm_3x4c8__wasmsimd_ld128() argument
786 GEMMBenchmark(state, xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld128, 3, 4, 8, 1); in qs8_gemm_3x4c8__wasmsimd_ld128()
789 static void qs8_gemm_xw_2x4c8__wasmsimd(benchmark::State& state, const char* net) { in qs8_gemm_xw_2x4c8__wasmsimd() argument
790 GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_ukernel_2x4c8__wasmsimd, 2, 4, 8, 1); in qs8_gemm_xw_2x4c8__wasmsimd()
792 static void qs8_gemm_xw_3x4c8__wasmsimd(benchmark::State& state, const char* net) { in qs8_gemm_xw_3x4c8__wasmsimd() argument
793 GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_ukernel_3x4c8__wasmsimd, 3, 4, 8, 1); in qs8_gemm_xw_3x4c8__wasmsimd()