35 static void GEMMBenchmark(benchmark::State& state,  in GEMMBenchmark()  argument
43     state.SkipWithError("cpuinfo initialization failed");  in GEMMBenchmark()
46   if (isa_check && !isa_check(state)) {  in GEMMBenchmark()
50   const size_t mc = state.range(0);  in GEMMBenchmark()
51   const size_t nc = state.range(1);  in GEMMBenchmark()
52   const size_t kc = state.range(2);  in GEMMBenchmark()
91   for (auto _ : state) {  in GEMMBenchmark()
92     // Use circular buffers (exceeding cache size) and prefetch to control cache state:  in GEMMBenchmark()
96     state.PauseTiming();  in GEMMBenchmark()
99     state.ResumeTiming();  in GEMMBenchmark()
117     state.counters["cpufreq"] = cpu_frequency;  in GEMMBenchmark()
120   state.counters["OPS"] = benchmark::Counter(  in GEMMBenchmark()
121     uint64_t(state.iterations()) * 2 * mc * nc * kc, benchmark::Counter::kIsRate);  in GEMMBenchmark()
125 static void RuyBenchmark(benchmark::State& state, size_t threads)  in RuyBenchmark()  argument
127   const size_t mc = state.range(0);  in RuyBenchmark()
128   const size_t nc = state.range(1);  in RuyBenchmark()
129   const size_t kc = state.range(2);  in RuyBenchmark()
184   for (auto _ : state) {  in RuyBenchmark()
185     // Use circular buffers (exceeding cache size) and prefetch to control cache state:  in RuyBenchmark()
190     state.PauseTiming();  in RuyBenchmark()
193     state.ResumeTiming();  in RuyBenchmark()
204     state.counters["cpufreq"] = cpu_frequency;  in RuyBenchmark()
207   state.counters["OPS"] = benchmark::Counter(  in RuyBenchmark()
208     uint64_t(state.iterations()) * 2 * mc * nc * kc, benchmark::Counter::kIsRate);  in RuyBenchmark()
211 static void ruy_st(benchmark::State& state, const char* net)  in ruy_st()  argument
213   RuyBenchmark(state, 1);  in ruy_st()
218   static void GEMMBenchmark(benchmark::State& state,  in GEMMBenchmark()  argument
227     const size_t nc = state.range(1);  in GEMMBenchmark()
228     const size_t kc = state.range(2);  in GEMMBenchmark()
232         state,  in GEMMBenchmark()
238   static void jit_qs8_gemm_4x8c4__aarch32_neondot_ld64(benchmark::State& state, const char* net) {  in jit_qs8_gemm_4x8c4__aarch32_neondot_ld64()  argument
239 …  GEMMBenchmark(state, xnn_generate_qs8_gemm_rndnu_ukernel_4x8c4__aarch32_neondot_ld64, 4, 8, 4, 1,  in jit_qs8_gemm_4x8c4__aarch32_neondot_ld64()
242 …static void jit_qs8_gemm_4x8__aarch32_neon_mlal_lane_ld64(benchmark::State& state, const char* net…  in jit_qs8_gemm_4x8__aarch32_neon_mlal_lane_ld64()  argument
243 …GEMMBenchmark(state, xnn_generate_qs8_gemm_rndnu_ukernel_4x8__aarch32_neon_mlal_lane_ld64, 4, 8, 1…  in jit_qs8_gemm_4x8__aarch32_neon_mlal_lane_ld64()
246 …static void jit_qs8_gemm_4x8__aarch32_neon_mlal_lane_prfm_ld64(benchmark::State& state, const char…  in jit_qs8_gemm_4x8__aarch32_neon_mlal_lane_prfm_ld64()  argument
247 …GEMMBenchmark(state, xnn_generate_qs8_gemm_rndnu_ukernel_4x8__aarch32_neon_mlal_lane_prfm_ld64, 4,…  in jit_qs8_gemm_4x8__aarch32_neon_mlal_lane_prfm_ld64()
256   static void qs8_gemm_4x8c4__aarch32_neondot_ld64(benchmark::State& state, const char* net) {  in BENCHMARK_GEMM()
257     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c4__aarch32_neondot_ld64, 4, 8, 4, 1,  in BENCHMARK_GEMM()
260   static void qs8_gemm_4x8c4__aarch32_neondot_cortex_a55(benchmark::State& state, const char* net) {  in qs8_gemm_4x8c4__aarch32_neondot_cortex_a55()  argument
261 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c4__aarch32_neondot_cortex_a55, 4, 8, 4,…  in qs8_gemm_4x8c4__aarch32_neondot_cortex_a55()
264 …static void qs8_gemm_4x8__aarch32_neon_mlal_lane_cortex_a53(benchmark::State& state, const char* n…  in qs8_gemm_4x8__aarch32_neon_mlal_lane_cortex_a53()  argument
265 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8__aarch32_neon_mlal_lane_cortex_a53, 4, …  in qs8_gemm_4x8__aarch32_neon_mlal_lane_cortex_a53()
268 …static void qs8_gemm_4x8__aarch32_neon_mlal_lane_prfm_cortex_a53(benchmark::State& state, const ch…  in qs8_gemm_4x8__aarch32_neon_mlal_lane_prfm_cortex_a53()  argument
269 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8__aarch32_neon_mlal_lane_prfm_cortex_a53…  in qs8_gemm_4x8__aarch32_neon_mlal_lane_prfm_cortex_a53()
272 …static void qs8_gemm_4x8__aarch32_neon_mlal_lane_cortex_a7(benchmark::State& state, const char* ne…  in qs8_gemm_4x8__aarch32_neon_mlal_lane_cortex_a7()  argument
273 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8__aarch32_neon_mlal_lane_cortex_a7, 4, 8…  in qs8_gemm_4x8__aarch32_neon_mlal_lane_cortex_a7()
276 …static void qs8_gemm_4x8__aarch32_neon_mlal_lane_prfm_cortex_a7(benchmark::State& state, const cha…  in qs8_gemm_4x8__aarch32_neon_mlal_lane_prfm_cortex_a7()  argument
277 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8__aarch32_neon_mlal_lane_prfm_cortex_a7,…  in qs8_gemm_4x8__aarch32_neon_mlal_lane_prfm_cortex_a7()
280   static void qs8_gemm_4x8__aarch32_neon_mlal_lane_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_4x8__aarch32_neon_mlal_lane_ld64()  argument
281 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8__aarch32_neon_mlal_lane_ld64, 4, 8, 1, …  in qs8_gemm_4x8__aarch32_neon_mlal_lane_ld64()
284 …static void qs8_gemm_4x8__aarch32_neon_mlal_lane_prfm_ld64(benchmark::State& state, const char* ne…  in qs8_gemm_4x8__aarch32_neon_mlal_lane_prfm_ld64()  argument
285 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8__aarch32_neon_mlal_lane_prfm_ld64, 4, 8…  in qs8_gemm_4x8__aarch32_neon_mlal_lane_prfm_ld64()
288 …static void qs8_gemm_1x8__aarch32_neon_mlal_lane_cortex_a7(benchmark::State& state, const char* ne…  in qs8_gemm_1x8__aarch32_neon_mlal_lane_cortex_a7()  argument
289 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8__aarch32_neon_mlal_lane_cortex_a7, 1, 8…  in qs8_gemm_1x8__aarch32_neon_mlal_lane_cortex_a7()
292 …static void qs8_gemm_1x8__aarch32_neon_mlal_lane_prfm_cortex_a7(benchmark::State& state, const cha…  in qs8_gemm_1x8__aarch32_neon_mlal_lane_prfm_cortex_a7()  argument
293 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8__aarch32_neon_mlal_lane_prfm_cortex_a7,…  in qs8_gemm_1x8__aarch32_neon_mlal_lane_prfm_cortex_a7()
310 …static void qs8_gemm_4x16c4__aarch64_neondot_cortex_a55(benchmark::State& state, const char* net) {  in BENCHMARK_GEMM()
311 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c4__aarch64_neondot_cortex_a55, 4, 16, …  in BENCHMARK_GEMM()
314   static void qs8_gemm_1x16c4__aarch64_neondot_ld32(benchmark::State& state, const char* net) {  in qs8_gemm_1x16c4__aarch64_neondot_ld32()  argument
315 …  GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c4__aarch64_neondot_ld32, 1, 16, 4, 1,  in qs8_gemm_1x16c4__aarch64_neondot_ld32()
318   static void qs8_gemm_1x16c4__aarch64_neondot_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_1x16c4__aarch64_neondot_ld64()  argument
319 …  GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c4__aarch64_neondot_ld64, 1, 16, 4, 1,  in qs8_gemm_1x16c4__aarch64_neondot_ld64()
322   static void qs8_gemm_4x16c4__aarch64_neondot_ld32(benchmark::State& state, const char* net) {  in qs8_gemm_4x16c4__aarch64_neondot_ld32()  argument
323 …  GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c4__aarch64_neondot_ld32, 4, 16, 4, 1,  in qs8_gemm_4x16c4__aarch64_neondot_ld32()
326   static void qs8_gemm_4x16c4__aarch64_neondot_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_4x16c4__aarch64_neondot_ld64()  argument
327 …  GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c4__aarch64_neondot_ld64, 4, 16, 4, 1,  in qs8_gemm_4x16c4__aarch64_neondot_ld64()
330   static void qs8_gemm_4x16c4__aarch64_neondot_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_4x16c4__aarch64_neondot_ld128()  argument
331 … GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c4__aarch64_neondot_ld128, 4, 16, 4, 1,  in qs8_gemm_4x16c4__aarch64_neondot_ld128()
334   static void qs8_gemm_4x8__aarch64_neon_mlal_lane_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_4x8__aarch64_neon_mlal_lane_ld64()  argument
335 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8__aarch64_neon_mlal_lane_ld64, 4, 8, 1, …  in qs8_gemm_4x8__aarch64_neon_mlal_lane_ld64()
338 …static void qs8_gemm_4x8__aarch64_neon_mlal_lane_prfm_ld64(benchmark::State& state, const char* ne…  in qs8_gemm_4x8__aarch64_neon_mlal_lane_prfm_ld64()  argument
339 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8__aarch64_neon_mlal_lane_prfm_ld64, 4, 8…  in qs8_gemm_4x8__aarch64_neon_mlal_lane_prfm_ld64()
342 …static void qs8_gemm_4x16__aarch64_neon_mlal_lane_cortex_a53(benchmark::State& state, const char* …  in qs8_gemm_4x16__aarch64_neon_mlal_lane_cortex_a53()  argument
343 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16__aarch64_neon_mlal_lane_cortex_a53, 4,…  in qs8_gemm_4x16__aarch64_neon_mlal_lane_cortex_a53()
346 …static void qs8_gemm_4x16__aarch64_neon_mlal_lane_prfm_cortex_a53(benchmark::State& state, const c…  in qs8_gemm_4x16__aarch64_neon_mlal_lane_prfm_cortex_a53()  argument
347 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16__aarch64_neon_mlal_lane_prfm_cortex_a5…  in qs8_gemm_4x16__aarch64_neon_mlal_lane_prfm_cortex_a53()
350   static void qs8_gemm_4x16__aarch64_neon_mlal_lane_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_4x16__aarch64_neon_mlal_lane_ld64()  argument
351 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16__aarch64_neon_mlal_lane_ld64, 4, 16, 1…  in qs8_gemm_4x16__aarch64_neon_mlal_lane_ld64()
354 …static void qs8_gemm_4x16__aarch64_neon_mlal_lane_prfm_ld64(benchmark::State& state, const char* n…  in qs8_gemm_4x16__aarch64_neon_mlal_lane_prfm_ld64()  argument
355 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16__aarch64_neon_mlal_lane_prfm_ld64, 4, …  in qs8_gemm_4x16__aarch64_neon_mlal_lane_prfm_ld64()
358   static void qs8_gemm_1x8c8__aarch64_neon_mlal_prfm(benchmark::State& state, const char* net) {  in qs8_gemm_1x8c8__aarch64_neon_mlal_prfm()  argument
359 …  GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c8__aarch64_neon_mlal_prfm, 1, 8, 8, 1,  in qs8_gemm_1x8c8__aarch64_neon_mlal_prfm()
362   static void qs8_gemm_1x8c8__aarch64_neon_mlal(benchmark::State& state, const char* net) {  in qs8_gemm_1x8c8__aarch64_neon_mlal()  argument
363     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c8__aarch64_neon_mlal, 1, 8, 8, 1,  in qs8_gemm_1x8c8__aarch64_neon_mlal()
366 …static void qs8_gemm_1x8c8__aarch64_neon_mlal_cortex_a53(benchmark::State& state, const char* net)…  in qs8_gemm_1x8c8__aarch64_neon_mlal_cortex_a53()  argument
367 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c8__aarch64_neon_mlal_cortex_a53, 1, 8, …  in qs8_gemm_1x8c8__aarch64_neon_mlal_cortex_a53()
370 …static void qs8_gemm_1x8c8__aarch64_neon_mlal_prfm_cortex_a53(benchmark::State& state, const char*…  in qs8_gemm_1x8c8__aarch64_neon_mlal_prfm_cortex_a53()  argument
371 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c8__aarch64_neon_mlal_prfm_cortex_a53, 1…  in qs8_gemm_1x8c8__aarch64_neon_mlal_prfm_cortex_a53()
374   static void qs8_gemm_2x8c8__aarch64_neon_mull(benchmark::State& state, const char* net) {  in qs8_gemm_2x8c8__aarch64_neon_mull()  argument
375     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c8__aarch64_neon_mull, 2, 8, 8, 1,  in qs8_gemm_2x8c8__aarch64_neon_mull()
378   static void qs8_gemm_2x8c8__aarch64_neon_mlal(benchmark::State& state, const char* net) {  in qs8_gemm_2x8c8__aarch64_neon_mlal()  argument
379     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c8__aarch64_neon_mlal, 2, 8, 8, 1,  in qs8_gemm_2x8c8__aarch64_neon_mlal()
382   static void qs8_gemm_2x8c8__aarch64_neon_mlal_prfm(benchmark::State& state, const char* net) {  in qs8_gemm_2x8c8__aarch64_neon_mlal_prfm()  argument
383 …  GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c8__aarch64_neon_mlal_prfm, 2, 8, 8, 1,  in qs8_gemm_2x8c8__aarch64_neon_mlal_prfm()
386 …static void qs8_gemm_2x8c8__aarch64_neon_mlal_cortex_a53(benchmark::State& state, const char* net)…  in qs8_gemm_2x8c8__aarch64_neon_mlal_cortex_a53()  argument
387 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c8__aarch64_neon_mlal_cortex_a53, 2, 8, …  in qs8_gemm_2x8c8__aarch64_neon_mlal_cortex_a53()
390 …static void qs8_gemm_2x8c8__aarch64_neon_mlal_prfm_cortex_a53(benchmark::State& state, const char*…  in qs8_gemm_2x8c8__aarch64_neon_mlal_prfm_cortex_a53()  argument
391 …GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c8__aarch64_neon_mlal_prfm_cortex_a53, 2…  in qs8_gemm_2x8c8__aarch64_neon_mlal_prfm_cortex_a53()
394   static void qs8_gemm_2x8c16__aarch64_neon_mlal(benchmark::State& state, const char* net) {  in qs8_gemm_2x8c16__aarch64_neon_mlal()  argument
395     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c16__aarch64_neon_mlal, 2, 8, 16, 1,  in qs8_gemm_2x8c16__aarch64_neon_mlal()
425   static void qs8_gemm_1x8c4__neondot(benchmark::State& state, const char* net) {  in BENCHMARK_GEMM()
426     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c4__neondot, 1, 8, 4, 1,  in BENCHMARK_GEMM()
429   static void qs8_gemm_4x8c4__neondot(benchmark::State& state, const char* net) {  in qs8_gemm_4x8c4__neondot()  argument
430     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c4__neondot, 4, 8, 4, 1,  in qs8_gemm_4x8c4__neondot()
433   static void qs8_gemm_6x8c4__neondot(benchmark::State& state, const char* net) {  in qs8_gemm_6x8c4__neondot()  argument
434     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_6x8c4__neondot, 6, 8, 4, 1,  in qs8_gemm_6x8c4__neondot()
437   static void qs8_gemm_8x8c4__neondot(benchmark::State& state, const char* net) {  in qs8_gemm_8x8c4__neondot()  argument
438     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_8x8c4__neondot, 8, 8, 4, 1,  in qs8_gemm_8x8c4__neondot()
441   static void qs8_gemm_1x16c4__neondot(benchmark::State& state, const char* net) {  in qs8_gemm_1x16c4__neondot()  argument
442     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c4__neondot, 1, 16, 4, 1,  in qs8_gemm_1x16c4__neondot()
445   static void qs8_gemm_4x16c4__neondot(benchmark::State& state, const char* net) {  in qs8_gemm_4x16c4__neondot()  argument
446     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c4__neondot, 4, 16, 4, 1,  in qs8_gemm_4x16c4__neondot()
449   static void qs8_gemm_6x16c4__neondot(benchmark::State& state, const char* net) {  in qs8_gemm_6x16c4__neondot()  argument
450     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_6x16c4__neondot, 6, 16, 4, 1,  in qs8_gemm_6x16c4__neondot()
453   static void qs8_gemm_8x16c4__neondot(benchmark::State& state, const char* net) {  in qs8_gemm_8x16c4__neondot()  argument
454     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_8x16c4__neondot, 8, 16, 4, 1,  in qs8_gemm_8x16c4__neondot()
470   static void qs8_gemm_1x8__neon_mlal_lane(benchmark::State& state, const char* net) {  in BENCHMARK_GEMM()
471     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8__neon_mlal_lane, 1, 8, 1, 1,  in BENCHMARK_GEMM()
474   static void qs8_gemm_2x8__neon_mlal_lane(benchmark::State& state, const char* net) {  in qs8_gemm_2x8__neon_mlal_lane()  argument
475     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x8__neon_mlal_lane, 2, 8, 1, 1,  in qs8_gemm_2x8__neon_mlal_lane()
478   static void qs8_gemm_3x8__neon_mlal_lane(benchmark::State& state, const char* net) {  in qs8_gemm_3x8__neon_mlal_lane()  argument
479     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x8__neon_mlal_lane, 3, 8, 1, 1,  in qs8_gemm_3x8__neon_mlal_lane()
482   static void qs8_gemm_4x8__neon_mlal_lane(benchmark::State& state, const char* net) {  in qs8_gemm_4x8__neon_mlal_lane()  argument
483     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8__neon_mlal_lane, 4, 8, 1, 1,  in qs8_gemm_4x8__neon_mlal_lane()
486   static void qs8_gemm_6x8__neon_mlal_lane(benchmark::State& state, const char* net) {  in qs8_gemm_6x8__neon_mlal_lane()  argument
487     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_6x8__neon_mlal_lane, 6, 8, 1, 1,  in qs8_gemm_6x8__neon_mlal_lane()
490   static void qs8_gemm_1x16__neon_mlal_lane(benchmark::State& state, const char* net) {  in qs8_gemm_1x16__neon_mlal_lane()  argument
491     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x16__neon_mlal_lane, 1, 16, 1, 1,  in qs8_gemm_1x16__neon_mlal_lane()
494   static void qs8_gemm_2x16__neon_mlal_lane(benchmark::State& state, const char* net) {  in qs8_gemm_2x16__neon_mlal_lane()  argument
495     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x16__neon_mlal_lane, 2, 16, 1, 1,  in qs8_gemm_2x16__neon_mlal_lane()
498   static void qs8_gemm_3x16__neon_mlal_lane(benchmark::State& state, const char* net) {  in qs8_gemm_3x16__neon_mlal_lane()  argument
499     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x16__neon_mlal_lane, 3, 16, 1, 1,  in qs8_gemm_3x16__neon_mlal_lane()
502   static void qs8_gemm_4x16__neon_mlal_lane(benchmark::State& state, const char* net) {  in qs8_gemm_4x16__neon_mlal_lane()  argument
503     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16__neon_mlal_lane, 4, 16, 1, 1,  in qs8_gemm_4x16__neon_mlal_lane()
506   static void qs8_gemm_6x16__neon_mlal_lane(benchmark::State& state, const char* net) {  in qs8_gemm_6x16__neon_mlal_lane()  argument
507     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_6x16__neon_mlal_lane, 6, 16, 1, 1,  in qs8_gemm_6x16__neon_mlal_lane()
510   static void qs8_gemm_1x8__neon_mlal_lane_prfm(benchmark::State& state, const char* net) {  in qs8_gemm_1x8__neon_mlal_lane_prfm()  argument
511     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8__neon_mlal_lane_prfm, 1, 8, 1, 1,  in qs8_gemm_1x8__neon_mlal_lane_prfm()
514   static void qs8_gemm_2x8__neon_mlal_lane_prfm(benchmark::State& state, const char* net) {  in qs8_gemm_2x8__neon_mlal_lane_prfm()  argument
515     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x8__neon_mlal_lane_prfm, 2, 8, 1, 1,  in qs8_gemm_2x8__neon_mlal_lane_prfm()
518   static void qs8_gemm_3x8__neon_mlal_lane_prfm(benchmark::State& state, const char* net) {  in qs8_gemm_3x8__neon_mlal_lane_prfm()  argument
519     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x8__neon_mlal_lane_prfm, 3, 8, 1, 1,  in qs8_gemm_3x8__neon_mlal_lane_prfm()
522   static void qs8_gemm_4x8__neon_mlal_lane_prfm(benchmark::State& state, const char* net) {  in qs8_gemm_4x8__neon_mlal_lane_prfm()  argument
523     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8__neon_mlal_lane_prfm, 4, 8, 1, 1,  in qs8_gemm_4x8__neon_mlal_lane_prfm()
526   static void qs8_gemm_6x8__neon_mlal_lane_prfm(benchmark::State& state, const char* net) {  in qs8_gemm_6x8__neon_mlal_lane_prfm()  argument
527     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_6x8__neon_mlal_lane_prfm, 6, 8, 1, 1,  in qs8_gemm_6x8__neon_mlal_lane_prfm()
530   static void qs8_gemm_1x16__neon_mlal_lane_prfm(benchmark::State& state, const char* net) {  in qs8_gemm_1x16__neon_mlal_lane_prfm()  argument
531     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x16__neon_mlal_lane_prfm, 1, 16, 1, 1,  in qs8_gemm_1x16__neon_mlal_lane_prfm()
534   static void qs8_gemm_2x16__neon_mlal_lane_prfm(benchmark::State& state, const char* net) {  in qs8_gemm_2x16__neon_mlal_lane_prfm()  argument
535     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x16__neon_mlal_lane_prfm, 2, 16, 1, 1,  in qs8_gemm_2x16__neon_mlal_lane_prfm()
538   static void qs8_gemm_3x16__neon_mlal_lane_prfm(benchmark::State& state, const char* net) {  in qs8_gemm_3x16__neon_mlal_lane_prfm()  argument
539     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x16__neon_mlal_lane_prfm, 3, 16, 1, 1,  in qs8_gemm_3x16__neon_mlal_lane_prfm()
542   static void qs8_gemm_4x16__neon_mlal_lane_prfm(benchmark::State& state, const char* net) {  in qs8_gemm_4x16__neon_mlal_lane_prfm()  argument
543     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16__neon_mlal_lane_prfm, 4, 16, 1, 1,  in qs8_gemm_4x16__neon_mlal_lane_prfm()
546   static void qs8_gemm_6x16__neon_mlal_lane_prfm(benchmark::State& state, const char* net) {  in qs8_gemm_6x16__neon_mlal_lane_prfm()  argument
547     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_6x16__neon_mlal_lane_prfm, 6, 16, 1, 1,  in qs8_gemm_6x16__neon_mlal_lane_prfm()
550   static void qs8_gemm_1x8c2__neon_mull_dup(benchmark::State& state, const char* net) {  in qs8_gemm_1x8c2__neon_mull_dup()  argument
551     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c2__neon_mull_dup, 1, 8, 2, 1,  in qs8_gemm_1x8c2__neon_mull_dup()
554   static void qs8_gemm_2x8c2__neon_mull_dup(benchmark::State& state, const char* net) {  in qs8_gemm_2x8c2__neon_mull_dup()  argument
555     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c2__neon_mull_dup, 2, 8, 2, 1,  in qs8_gemm_2x8c2__neon_mull_dup()
558   static void qs8_gemm_3x8c2__neon_mull_dup(benchmark::State& state, const char* net) {  in qs8_gemm_3x8c2__neon_mull_dup()  argument
559     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c2__neon_mull_dup, 3, 8, 2, 1,  in qs8_gemm_3x8c2__neon_mull_dup()
562   static void qs8_gemm_4x8c2__neon_mull_dup(benchmark::State& state, const char* net) {  in qs8_gemm_4x8c2__neon_mull_dup()  argument
563     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c2__neon_mull_dup, 4, 8, 2, 1,  in qs8_gemm_4x8c2__neon_mull_dup()
566   static void qs8_gemm_1x16c2__neon_mull_dup(benchmark::State& state, const char* net) {  in qs8_gemm_1x16c2__neon_mull_dup()  argument
567     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c2__neon_mull_dup, 1, 16, 2, 1,  in qs8_gemm_1x16c2__neon_mull_dup()
570   static void qs8_gemm_2x16c2__neon_mull_dup(benchmark::State& state, const char* net) {  in qs8_gemm_2x16c2__neon_mull_dup()  argument
571     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c2__neon_mull_dup, 2, 16, 2, 1,  in qs8_gemm_2x16c2__neon_mull_dup()
574   static void qs8_gemm_3x16c2__neon_mull_dup(benchmark::State& state, const char* net) {  in qs8_gemm_3x16c2__neon_mull_dup()  argument
575     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c2__neon_mull_dup, 3, 16, 2, 1,  in qs8_gemm_3x16c2__neon_mull_dup()
578   static void qs8_gemm_4x16c2__neon_mull_dup(benchmark::State& state, const char* net) {  in qs8_gemm_4x16c2__neon_mull_dup()  argument
579     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c2__neon_mull_dup, 4, 16, 2, 1,  in qs8_gemm_4x16c2__neon_mull_dup()
582   static void qs8_gemm_1x8c2__neon_mlal_dup(benchmark::State& state, const char* net) {  in qs8_gemm_1x8c2__neon_mlal_dup()  argument
583     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c2__neon_mlal_dup, 1, 8, 2, 1,  in qs8_gemm_1x8c2__neon_mlal_dup()
586   static void qs8_gemm_2x8c2__neon_mlal_dup(benchmark::State& state, const char* net) {  in qs8_gemm_2x8c2__neon_mlal_dup()  argument
587     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c2__neon_mlal_dup, 2, 8, 2, 1,  in qs8_gemm_2x8c2__neon_mlal_dup()
590   static void qs8_gemm_3x8c2__neon_mlal_dup(benchmark::State& state, const char* net) {  in qs8_gemm_3x8c2__neon_mlal_dup()  argument
591     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c2__neon_mlal_dup, 3, 8, 2, 1,  in qs8_gemm_3x8c2__neon_mlal_dup()
594   static void qs8_gemm_4x8c2__neon_mlal_dup(benchmark::State& state, const char* net) {  in qs8_gemm_4x8c2__neon_mlal_dup()  argument
595     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c2__neon_mlal_dup, 4, 8, 2, 1,  in qs8_gemm_4x8c2__neon_mlal_dup()
598   static void qs8_gemm_1x16c2__neon_mlal_dup(benchmark::State& state, const char* net) {  in qs8_gemm_1x16c2__neon_mlal_dup()  argument
599     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c2__neon_mlal_dup, 1, 16, 2, 1,  in qs8_gemm_1x16c2__neon_mlal_dup()
602   static void qs8_gemm_2x16c2__neon_mlal_dup(benchmark::State& state, const char* net) {  in qs8_gemm_2x16c2__neon_mlal_dup()  argument
603     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c2__neon_mlal_dup, 2, 16, 2, 1,  in qs8_gemm_2x16c2__neon_mlal_dup()
606   static void qs8_gemm_3x16c2__neon_mlal_dup(benchmark::State& state, const char* net) {  in qs8_gemm_3x16c2__neon_mlal_dup()  argument
607     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c2__neon_mlal_dup, 3, 16, 2, 1,  in qs8_gemm_3x16c2__neon_mlal_dup()
610   static void qs8_gemm_4x16c2__neon_mlal_dup(benchmark::State& state, const char* net) {  in qs8_gemm_4x16c2__neon_mlal_dup()  argument
611     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c2__neon_mlal_dup, 4, 16, 2, 1,  in qs8_gemm_4x16c2__neon_mlal_dup()
614   static void qs8_gemm_1x8c2__neon_mull_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_1x8c2__neon_mull_ld1r()  argument
615     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c2__neon_mull_ld1r, 1, 8, 2, 1,  in qs8_gemm_1x8c2__neon_mull_ld1r()
618   static void qs8_gemm_2x8c2__neon_mull_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_2x8c2__neon_mull_ld1r()  argument
619     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c2__neon_mull_ld1r, 2, 8, 2, 1,  in qs8_gemm_2x8c2__neon_mull_ld1r()
622   static void qs8_gemm_3x8c2__neon_mull_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_3x8c2__neon_mull_ld1r()  argument
623     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c2__neon_mull_ld1r, 3, 8, 2, 1,  in qs8_gemm_3x8c2__neon_mull_ld1r()
626   static void qs8_gemm_4x8c2__neon_mull_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_4x8c2__neon_mull_ld1r()  argument
627     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c2__neon_mull_ld1r, 4, 8, 2, 1,  in qs8_gemm_4x8c2__neon_mull_ld1r()
630   static void qs8_gemm_1x16c2__neon_mull_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_1x16c2__neon_mull_ld1r()  argument
631     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c2__neon_mull_ld1r, 1, 16, 2, 1,  in qs8_gemm_1x16c2__neon_mull_ld1r()
634   static void qs8_gemm_2x16c2__neon_mull_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_2x16c2__neon_mull_ld1r()  argument
635     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c2__neon_mull_ld1r, 2, 16, 2, 1,  in qs8_gemm_2x16c2__neon_mull_ld1r()
638   static void qs8_gemm_3x16c2__neon_mull_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_3x16c2__neon_mull_ld1r()  argument
639     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c2__neon_mull_ld1r, 3, 16, 2, 1,  in qs8_gemm_3x16c2__neon_mull_ld1r()
642   static void qs8_gemm_4x16c2__neon_mull_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_4x16c2__neon_mull_ld1r()  argument
643     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c2__neon_mull_ld1r, 4, 16, 2, 1,  in qs8_gemm_4x16c2__neon_mull_ld1r()
646   static void qs8_gemm_1x8c2__neon_mlal_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_1x8c2__neon_mlal_ld1r()  argument
647     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c2__neon_mlal_ld1r, 1, 8, 2, 1,  in qs8_gemm_1x8c2__neon_mlal_ld1r()
650   static void qs8_gemm_2x8c2__neon_mlal_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_2x8c2__neon_mlal_ld1r()  argument
651     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c2__neon_mlal_ld1r, 2, 8, 2, 1,  in qs8_gemm_2x8c2__neon_mlal_ld1r()
654   static void qs8_gemm_3x8c2__neon_mlal_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_3x8c2__neon_mlal_ld1r()  argument
655     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c2__neon_mlal_ld1r, 3, 8, 2, 1,  in qs8_gemm_3x8c2__neon_mlal_ld1r()
658   static void qs8_gemm_4x8c2__neon_mlal_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_4x8c2__neon_mlal_ld1r()  argument
659     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c2__neon_mlal_ld1r, 4, 8, 2, 1,  in qs8_gemm_4x8c2__neon_mlal_ld1r()
662   static void qs8_gemm_1x16c2__neon_mlal_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_1x16c2__neon_mlal_ld1r()  argument
663     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c2__neon_mlal_ld1r, 1, 16, 2, 1,  in qs8_gemm_1x16c2__neon_mlal_ld1r()
666   static void qs8_gemm_2x16c2__neon_mlal_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_2x16c2__neon_mlal_ld1r()  argument
667     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c2__neon_mlal_ld1r, 2, 16, 2, 1,  in qs8_gemm_2x16c2__neon_mlal_ld1r()
670   static void qs8_gemm_3x16c2__neon_mlal_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_3x16c2__neon_mlal_ld1r()  argument
671     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c2__neon_mlal_ld1r, 3, 16, 2, 1,  in qs8_gemm_3x16c2__neon_mlal_ld1r()
674   static void qs8_gemm_4x16c2__neon_mlal_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_4x16c2__neon_mlal_ld1r()  argument
675     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c2__neon_mlal_ld1r, 4, 16, 2, 1,  in qs8_gemm_4x16c2__neon_mlal_ld1r()
678   static void qs8_gemm_1x8c2__neon_mull_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_1x8c2__neon_mull_ld2r()  argument
679     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c2__neon_mull_ld2r, 1, 8, 2, 1,  in qs8_gemm_1x8c2__neon_mull_ld2r()
682   static void qs8_gemm_2x8c2__neon_mull_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_2x8c2__neon_mull_ld2r()  argument
683     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c2__neon_mull_ld2r, 2, 8, 2, 1,  in qs8_gemm_2x8c2__neon_mull_ld2r()
686   static void qs8_gemm_3x8c2__neon_mull_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_3x8c2__neon_mull_ld2r()  argument
687     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c2__neon_mull_ld2r, 3, 8, 2, 1,  in qs8_gemm_3x8c2__neon_mull_ld2r()
690   static void qs8_gemm_4x8c2__neon_mull_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_4x8c2__neon_mull_ld2r()  argument
691     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c2__neon_mull_ld2r, 4, 8, 2, 1,  in qs8_gemm_4x8c2__neon_mull_ld2r()
694   static void qs8_gemm_1x16c2__neon_mull_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_1x16c2__neon_mull_ld2r()  argument
695     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c2__neon_mull_ld2r, 1, 16, 2, 1,  in qs8_gemm_1x16c2__neon_mull_ld2r()
698   static void qs8_gemm_2x16c2__neon_mull_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_2x16c2__neon_mull_ld2r()  argument
699     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c2__neon_mull_ld2r, 2, 16, 2, 1,  in qs8_gemm_2x16c2__neon_mull_ld2r()
702   static void qs8_gemm_3x16c2__neon_mull_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_3x16c2__neon_mull_ld2r()  argument
703     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c2__neon_mull_ld2r, 3, 16, 2, 1,  in qs8_gemm_3x16c2__neon_mull_ld2r()
706   static void qs8_gemm_4x16c2__neon_mull_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_4x16c2__neon_mull_ld2r()  argument
707     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c2__neon_mull_ld2r, 4, 16, 2, 1,  in qs8_gemm_4x16c2__neon_mull_ld2r()
710   static void qs8_gemm_1x8c2__neon_mlal_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_1x8c2__neon_mlal_ld2r()  argument
711     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c2__neon_mlal_ld2r, 1, 8, 2, 1,  in qs8_gemm_1x8c2__neon_mlal_ld2r()
714   static void qs8_gemm_2x8c2__neon_mlal_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_2x8c2__neon_mlal_ld2r()  argument
715     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c2__neon_mlal_ld2r, 2, 8, 2, 1,  in qs8_gemm_2x8c2__neon_mlal_ld2r()
718   static void qs8_gemm_3x8c2__neon_mlal_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_3x8c2__neon_mlal_ld2r()  argument
719     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c2__neon_mlal_ld2r, 3, 8, 2, 1,  in qs8_gemm_3x8c2__neon_mlal_ld2r()
722   static void qs8_gemm_4x8c2__neon_mlal_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_4x8c2__neon_mlal_ld2r()  argument
723     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c2__neon_mlal_ld2r, 4, 8, 2, 1,  in qs8_gemm_4x8c2__neon_mlal_ld2r()
726   static void qs8_gemm_1x16c2__neon_mlal_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_1x16c2__neon_mlal_ld2r()  argument
727     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c2__neon_mlal_ld2r, 1, 16, 2, 1,  in qs8_gemm_1x16c2__neon_mlal_ld2r()
730   static void qs8_gemm_2x16c2__neon_mlal_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_2x16c2__neon_mlal_ld2r()  argument
731     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c2__neon_mlal_ld2r, 2, 16, 2, 1,  in qs8_gemm_2x16c2__neon_mlal_ld2r()
734   static void qs8_gemm_3x16c2__neon_mlal_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_3x16c2__neon_mlal_ld2r()  argument
735     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c2__neon_mlal_ld2r, 3, 16, 2, 1,  in qs8_gemm_3x16c2__neon_mlal_ld2r()
738   static void qs8_gemm_4x16c2__neon_mlal_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_4x16c2__neon_mlal_ld2r()  argument
739     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c2__neon_mlal_ld2r, 4, 16, 2, 1,  in qs8_gemm_4x16c2__neon_mlal_ld2r()
742   static void qs8_gemm_1x8c2__neon_mull_ld4r(benchmark::State& state, const char* net) {  in qs8_gemm_1x8c2__neon_mull_ld4r()  argument
743     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c2__neon_mull_ld4r, 1, 8, 2, 1,  in qs8_gemm_1x8c2__neon_mull_ld4r()
746   static void qs8_gemm_2x8c2__neon_mull_ld4r(benchmark::State& state, const char* net) {  in qs8_gemm_2x8c2__neon_mull_ld4r()  argument
747     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c2__neon_mull_ld4r, 2, 8, 2, 1,  in qs8_gemm_2x8c2__neon_mull_ld4r()
750   static void qs8_gemm_3x8c2__neon_mull_ld4r(benchmark::State& state, const char* net) {  in qs8_gemm_3x8c2__neon_mull_ld4r()  argument
751     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c2__neon_mull_ld4r, 3, 8, 2, 1,  in qs8_gemm_3x8c2__neon_mull_ld4r()
754   static void qs8_gemm_4x8c2__neon_mull_ld4r(benchmark::State& state, const char* net) {  in qs8_gemm_4x8c2__neon_mull_ld4r()  argument
755     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c2__neon_mull_ld4r, 4, 8, 2, 1,  in qs8_gemm_4x8c2__neon_mull_ld4r()
758   static void qs8_gemm_1x16c2__neon_mull_ld4r(benchmark::State& state, const char* net) {  in qs8_gemm_1x16c2__neon_mull_ld4r()  argument
759     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c2__neon_mull_ld4r, 1, 16, 2, 1,  in qs8_gemm_1x16c2__neon_mull_ld4r()
762   static void qs8_gemm_2x16c2__neon_mull_ld4r(benchmark::State& state, const char* net) {  in qs8_gemm_2x16c2__neon_mull_ld4r()  argument
763     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c2__neon_mull_ld4r, 2, 16, 2, 1,  in qs8_gemm_2x16c2__neon_mull_ld4r()
766   static void qs8_gemm_3x16c2__neon_mull_ld4r(benchmark::State& state, const char* net) {  in qs8_gemm_3x16c2__neon_mull_ld4r()  argument
767     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c2__neon_mull_ld4r, 3, 16, 2, 1,  in qs8_gemm_3x16c2__neon_mull_ld4r()
770   static void qs8_gemm_4x16c2__neon_mull_ld4r(benchmark::State& state, const char* net) {  in qs8_gemm_4x16c2__neon_mull_ld4r()  argument
771     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c2__neon_mull_ld4r, 4, 16, 2, 1,  in qs8_gemm_4x16c2__neon_mull_ld4r()
774   static void qs8_gemm_1x8c2__neon_mlal_ld4r(benchmark::State& state, const char* net) {  in qs8_gemm_1x8c2__neon_mlal_ld4r()  argument
775     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c2__neon_mlal_ld4r, 1, 8, 2, 1,  in qs8_gemm_1x8c2__neon_mlal_ld4r()
778   static void qs8_gemm_2x8c2__neon_mlal_ld4r(benchmark::State& state, const char* net) {  in qs8_gemm_2x8c2__neon_mlal_ld4r()  argument
779     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c2__neon_mlal_ld4r, 2, 8, 2, 1,  in qs8_gemm_2x8c2__neon_mlal_ld4r()
782   static void qs8_gemm_3x8c2__neon_mlal_ld4r(benchmark::State& state, const char* net) {  in qs8_gemm_3x8c2__neon_mlal_ld4r()  argument
783     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c2__neon_mlal_ld4r, 3, 8, 2, 1,  in qs8_gemm_3x8c2__neon_mlal_ld4r()
786   static void qs8_gemm_4x8c2__neon_mlal_ld4r(benchmark::State& state, const char* net) {  in qs8_gemm_4x8c2__neon_mlal_ld4r()  argument
787     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c2__neon_mlal_ld4r, 4, 8, 2, 1,  in qs8_gemm_4x8c2__neon_mlal_ld4r()
790   static void qs8_gemm_1x16c2__neon_mlal_ld4r(benchmark::State& state, const char* net) {  in qs8_gemm_1x16c2__neon_mlal_ld4r()  argument
791     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c2__neon_mlal_ld4r, 1, 16, 2, 1,  in qs8_gemm_1x16c2__neon_mlal_ld4r()
794   static void qs8_gemm_2x16c2__neon_mlal_ld4r(benchmark::State& state, const char* net) {  in qs8_gemm_2x16c2__neon_mlal_ld4r()  argument
795     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c2__neon_mlal_ld4r, 2, 16, 2, 1,  in qs8_gemm_2x16c2__neon_mlal_ld4r()
798   static void qs8_gemm_3x16c2__neon_mlal_ld4r(benchmark::State& state, const char* net) {  in qs8_gemm_3x16c2__neon_mlal_ld4r()  argument
799     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c2__neon_mlal_ld4r, 3, 16, 2, 1,  in qs8_gemm_3x16c2__neon_mlal_ld4r()
802   static void qs8_gemm_4x16c2__neon_mlal_ld4r(benchmark::State& state, const char* net) {  in qs8_gemm_4x16c2__neon_mlal_ld4r()  argument
803     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c2__neon_mlal_ld4r, 4, 16, 2, 1,  in qs8_gemm_4x16c2__neon_mlal_ld4r()
806   static void qs8_gemm_1x8c2s4__neon_mull(benchmark::State& state, const char* net) {  in qs8_gemm_1x8c2s4__neon_mull()  argument
807     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c2s4__neon_mull, 1, 8, 2, 4,  in qs8_gemm_1x8c2s4__neon_mull()
810   static void qs8_gemm_2x8c2s4__neon_mull(benchmark::State& state, const char* net) {  in qs8_gemm_2x8c2s4__neon_mull()  argument
811     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c2s4__neon_mull, 2, 8, 2, 4,  in qs8_gemm_2x8c2s4__neon_mull()
814   static void qs8_gemm_3x8c2s4__neon_mull(benchmark::State& state, const char* net) {  in qs8_gemm_3x8c2s4__neon_mull()  argument
815     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c2s4__neon_mull, 3, 8, 2, 4,  in qs8_gemm_3x8c2s4__neon_mull()
818   static void qs8_gemm_4x8c2s4__neon_mull(benchmark::State& state, const char* net) {  in qs8_gemm_4x8c2s4__neon_mull()  argument
819     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c2s4__neon_mull, 4, 8, 2, 4,  in qs8_gemm_4x8c2s4__neon_mull()
822   static void qs8_gemm_1x16c2s4__neon_mull(benchmark::State& state, const char* net) {  in qs8_gemm_1x16c2s4__neon_mull()  argument
823     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c2s4__neon_mull, 1, 16, 2, 4,  in qs8_gemm_1x16c2s4__neon_mull()
826   static void qs8_gemm_2x16c2s4__neon_mull(benchmark::State& state, const char* net) {  in qs8_gemm_2x16c2s4__neon_mull()  argument
827     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c2s4__neon_mull, 2, 16, 2, 4,  in qs8_gemm_2x16c2s4__neon_mull()
830   static void qs8_gemm_3x16c2s4__neon_mull(benchmark::State& state, const char* net) {  in qs8_gemm_3x16c2s4__neon_mull()  argument
831     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c2s4__neon_mull, 3, 16, 2, 4,  in qs8_gemm_3x16c2s4__neon_mull()
834   static void qs8_gemm_4x16c2s4__neon_mull(benchmark::State& state, const char* net) {  in qs8_gemm_4x16c2s4__neon_mull()  argument
835     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c2s4__neon_mull, 4, 16, 2, 4,  in qs8_gemm_4x16c2s4__neon_mull()
838   static void qs8_gemm_1x8c2s4__neon_mlal(benchmark::State& state, const char* net) {  in qs8_gemm_1x8c2s4__neon_mlal()  argument
839     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c2s4__neon_mlal, 1, 8, 2, 4,  in qs8_gemm_1x8c2s4__neon_mlal()
842   static void qs8_gemm_2x8c2s4__neon_mlal(benchmark::State& state, const char* net) {  in qs8_gemm_2x8c2s4__neon_mlal()  argument
843     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c2s4__neon_mlal, 2, 8, 2, 4,  in qs8_gemm_2x8c2s4__neon_mlal()
846   static void qs8_gemm_3x8c2s4__neon_mlal(benchmark::State& state, const char* net) {  in qs8_gemm_3x8c2s4__neon_mlal()  argument
847     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c2s4__neon_mlal, 3, 8, 2, 4,  in qs8_gemm_3x8c2s4__neon_mlal()
850   static void qs8_gemm_4x8c2s4__neon_mlal(benchmark::State& state, const char* net) {  in qs8_gemm_4x8c2s4__neon_mlal()  argument
851     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c2s4__neon_mlal, 4, 8, 2, 4,  in qs8_gemm_4x8c2s4__neon_mlal()
854   static void qs8_gemm_1x16c2s4__neon_mlal(benchmark::State& state, const char* net) {  in qs8_gemm_1x16c2s4__neon_mlal()  argument
855     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c2s4__neon_mlal, 1, 16, 2, 4,  in qs8_gemm_1x16c2s4__neon_mlal()
858   static void qs8_gemm_2x16c2s4__neon_mlal(benchmark::State& state, const char* net) {  in qs8_gemm_2x16c2s4__neon_mlal()  argument
859     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c2s4__neon_mlal, 2, 16, 2, 4,  in qs8_gemm_2x16c2s4__neon_mlal()
862   static void qs8_gemm_3x16c2s4__neon_mlal(benchmark::State& state, const char* net) {  in qs8_gemm_3x16c2s4__neon_mlal()  argument
863     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c2s4__neon_mlal, 3, 16, 2, 4,  in qs8_gemm_3x16c2s4__neon_mlal()
866   static void qs8_gemm_4x16c2s4__neon_mlal(benchmark::State& state, const char* net) {  in qs8_gemm_4x16c2s4__neon_mlal()  argument
867     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c2s4__neon_mlal, 4, 16, 2, 4,  in qs8_gemm_4x16c2s4__neon_mlal()
870   static void qs8_gemm_1x8c4__neon_mull_dup(benchmark::State& state, const char* net) {  in qs8_gemm_1x8c4__neon_mull_dup()  argument
871     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c4__neon_mull_dup, 1, 8, 4, 1,  in qs8_gemm_1x8c4__neon_mull_dup()
874   static void qs8_gemm_2x8c4__neon_mull_dup(benchmark::State& state, const char* net) {  in qs8_gemm_2x8c4__neon_mull_dup()  argument
875     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c4__neon_mull_dup, 2, 8, 4, 1,  in qs8_gemm_2x8c4__neon_mull_dup()
878   static void qs8_gemm_3x8c4__neon_mull_dup(benchmark::State& state, const char* net) {  in qs8_gemm_3x8c4__neon_mull_dup()  argument
879     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c4__neon_mull_dup, 3, 8, 4, 1,  in qs8_gemm_3x8c4__neon_mull_dup()
882   static void qs8_gemm_4x8c4__neon_mull_dup(benchmark::State& state, const char* net) {  in qs8_gemm_4x8c4__neon_mull_dup()  argument
883     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c4__neon_mull_dup, 4, 8, 4, 1,  in qs8_gemm_4x8c4__neon_mull_dup()
886   static void qs8_gemm_1x16c4__neon_mull_dup(benchmark::State& state, const char* net) {  in qs8_gemm_1x16c4__neon_mull_dup()  argument
887     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c4__neon_mull_dup, 1, 16, 4, 1,  in qs8_gemm_1x16c4__neon_mull_dup()
890   static void qs8_gemm_2x16c4__neon_mull_dup(benchmark::State& state, const char* net) {  in qs8_gemm_2x16c4__neon_mull_dup()  argument
891     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c4__neon_mull_dup, 2, 16, 4, 1,  in qs8_gemm_2x16c4__neon_mull_dup()
894   static void qs8_gemm_3x16c4__neon_mull_dup(benchmark::State& state, const char* net) {  in qs8_gemm_3x16c4__neon_mull_dup()  argument
895     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c4__neon_mull_dup, 3, 16, 4, 1,  in qs8_gemm_3x16c4__neon_mull_dup()
898   static void qs8_gemm_4x16c4__neon_mull_dup(benchmark::State& state, const char* net) {  in qs8_gemm_4x16c4__neon_mull_dup()  argument
899     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c4__neon_mull_dup, 4, 16, 4, 1,  in qs8_gemm_4x16c4__neon_mull_dup()
902   static void qs8_gemm_1x8c4__neon_mlal_dup(benchmark::State& state, const char* net) {  in qs8_gemm_1x8c4__neon_mlal_dup()  argument
903     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c4__neon_mlal_dup, 1, 8, 4, 1,  in qs8_gemm_1x8c4__neon_mlal_dup()
906   static void qs8_gemm_2x8c4__neon_mlal_dup(benchmark::State& state, const char* net) {  in qs8_gemm_2x8c4__neon_mlal_dup()  argument
907     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c4__neon_mlal_dup, 2, 8, 4, 1,  in qs8_gemm_2x8c4__neon_mlal_dup()
910   static void qs8_gemm_3x8c4__neon_mlal_dup(benchmark::State& state, const char* net) {  in qs8_gemm_3x8c4__neon_mlal_dup()  argument
911     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c4__neon_mlal_dup, 3, 8, 4, 1,  in qs8_gemm_3x8c4__neon_mlal_dup()
914   static void qs8_gemm_4x8c4__neon_mlal_dup(benchmark::State& state, const char* net) {  in qs8_gemm_4x8c4__neon_mlal_dup()  argument
915     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c4__neon_mlal_dup, 4, 8, 4, 1,  in qs8_gemm_4x8c4__neon_mlal_dup()
918   static void qs8_gemm_1x16c4__neon_mlal_dup(benchmark::State& state, const char* net) {  in qs8_gemm_1x16c4__neon_mlal_dup()  argument
919     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c4__neon_mlal_dup, 1, 16, 4, 1,  in qs8_gemm_1x16c4__neon_mlal_dup()
922   static void qs8_gemm_2x16c4__neon_mlal_dup(benchmark::State& state, const char* net) {  in qs8_gemm_2x16c4__neon_mlal_dup()  argument
923     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c4__neon_mlal_dup, 2, 16, 4, 1,  in qs8_gemm_2x16c4__neon_mlal_dup()
926   static void qs8_gemm_3x16c4__neon_mlal_dup(benchmark::State& state, const char* net) {  in qs8_gemm_3x16c4__neon_mlal_dup()  argument
927     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c4__neon_mlal_dup, 3, 16, 4, 1,  in qs8_gemm_3x16c4__neon_mlal_dup()
930   static void qs8_gemm_4x16c4__neon_mlal_dup(benchmark::State& state, const char* net) {  in qs8_gemm_4x16c4__neon_mlal_dup()  argument
931     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c4__neon_mlal_dup, 4, 16, 4, 1,  in qs8_gemm_4x16c4__neon_mlal_dup()
934   static void qs8_gemm_1x8c4__neon_mull_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_1x8c4__neon_mull_ld1r()  argument
935     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c4__neon_mull_ld1r, 1, 8, 4, 1,  in qs8_gemm_1x8c4__neon_mull_ld1r()
938   static void qs8_gemm_2x8c4__neon_mull_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_2x8c4__neon_mull_ld1r()  argument
939     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c4__neon_mull_ld1r, 2, 8, 4, 1,  in qs8_gemm_2x8c4__neon_mull_ld1r()
942   static void qs8_gemm_3x8c4__neon_mull_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_3x8c4__neon_mull_ld1r()  argument
943     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c4__neon_mull_ld1r, 3, 8, 4, 1,  in qs8_gemm_3x8c4__neon_mull_ld1r()
946   static void qs8_gemm_4x8c4__neon_mull_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_4x8c4__neon_mull_ld1r()  argument
947     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c4__neon_mull_ld1r, 4, 8, 4, 1,  in qs8_gemm_4x8c4__neon_mull_ld1r()
950   static void qs8_gemm_1x16c4__neon_mull_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_1x16c4__neon_mull_ld1r()  argument
951     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c4__neon_mull_ld1r, 1, 16, 4, 1,  in qs8_gemm_1x16c4__neon_mull_ld1r()
954   static void qs8_gemm_2x16c4__neon_mull_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_2x16c4__neon_mull_ld1r()  argument
955     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c4__neon_mull_ld1r, 2, 16, 4, 1,  in qs8_gemm_2x16c4__neon_mull_ld1r()
958   static void qs8_gemm_3x16c4__neon_mull_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_3x16c4__neon_mull_ld1r()  argument
959     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c4__neon_mull_ld1r, 3, 16, 4, 1,  in qs8_gemm_3x16c4__neon_mull_ld1r()
962   static void qs8_gemm_4x16c4__neon_mull_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_4x16c4__neon_mull_ld1r()  argument
963     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c4__neon_mull_ld1r, 4, 16, 4, 1,  in qs8_gemm_4x16c4__neon_mull_ld1r()
966   static void qs8_gemm_1x8c4__neon_mlal_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_1x8c4__neon_mlal_ld1r()  argument
967     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c4__neon_mlal_ld1r, 1, 8, 4, 1,  in qs8_gemm_1x8c4__neon_mlal_ld1r()
970   static void qs8_gemm_2x8c4__neon_mlal_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_2x8c4__neon_mlal_ld1r()  argument
971     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c4__neon_mlal_ld1r, 2, 8, 4, 1,  in qs8_gemm_2x8c4__neon_mlal_ld1r()
974   static void qs8_gemm_3x8c4__neon_mlal_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_3x8c4__neon_mlal_ld1r()  argument
975     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c4__neon_mlal_ld1r, 3, 8, 4, 1,  in qs8_gemm_3x8c4__neon_mlal_ld1r()
978   static void qs8_gemm_4x8c4__neon_mlal_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_4x8c4__neon_mlal_ld1r()  argument
979     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c4__neon_mlal_ld1r, 4, 8, 4, 1,  in qs8_gemm_4x8c4__neon_mlal_ld1r()
982   static void qs8_gemm_1x16c4__neon_mlal_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_1x16c4__neon_mlal_ld1r()  argument
983     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c4__neon_mlal_ld1r, 1, 16, 4, 1,  in qs8_gemm_1x16c4__neon_mlal_ld1r()
986   static void qs8_gemm_2x16c4__neon_mlal_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_2x16c4__neon_mlal_ld1r()  argument
987     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c4__neon_mlal_ld1r, 2, 16, 4, 1,  in qs8_gemm_2x16c4__neon_mlal_ld1r()
990   static void qs8_gemm_3x16c4__neon_mlal_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_3x16c4__neon_mlal_ld1r()  argument
991     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c4__neon_mlal_ld1r, 3, 16, 4, 1,  in qs8_gemm_3x16c4__neon_mlal_ld1r()
994   static void qs8_gemm_4x16c4__neon_mlal_ld1r(benchmark::State& state, const char* net) {  in qs8_gemm_4x16c4__neon_mlal_ld1r()  argument
995     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c4__neon_mlal_ld1r, 4, 16, 4, 1,  in qs8_gemm_4x16c4__neon_mlal_ld1r()
998   static void qs8_gemm_1x8c4__neon_mull_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_1x8c4__neon_mull_ld2r()  argument
999     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c4__neon_mull_ld2r, 1, 8, 4, 1,  in qs8_gemm_1x8c4__neon_mull_ld2r()
1002   static void qs8_gemm_2x8c4__neon_mull_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_2x8c4__neon_mull_ld2r()  argument
1003     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c4__neon_mull_ld2r, 2, 8, 4, 1,  in qs8_gemm_2x8c4__neon_mull_ld2r()
1006   static void qs8_gemm_3x8c4__neon_mull_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_3x8c4__neon_mull_ld2r()  argument
1007     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c4__neon_mull_ld2r, 3, 8, 4, 1,  in qs8_gemm_3x8c4__neon_mull_ld2r()
1010   static void qs8_gemm_4x8c4__neon_mull_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_4x8c4__neon_mull_ld2r()  argument
1011     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c4__neon_mull_ld2r, 4, 8, 4, 1,  in qs8_gemm_4x8c4__neon_mull_ld2r()
1014   static void qs8_gemm_1x16c4__neon_mull_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_1x16c4__neon_mull_ld2r()  argument
1015     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c4__neon_mull_ld2r, 1, 16, 4, 1,  in qs8_gemm_1x16c4__neon_mull_ld2r()
1018   static void qs8_gemm_2x16c4__neon_mull_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_2x16c4__neon_mull_ld2r()  argument
1019     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c4__neon_mull_ld2r, 2, 16, 4, 1,  in qs8_gemm_2x16c4__neon_mull_ld2r()
1022   static void qs8_gemm_3x16c4__neon_mull_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_3x16c4__neon_mull_ld2r()  argument
1023     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c4__neon_mull_ld2r, 3, 16, 4, 1,  in qs8_gemm_3x16c4__neon_mull_ld2r()
1026   static void qs8_gemm_4x16c4__neon_mull_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_4x16c4__neon_mull_ld2r()  argument
1027     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c4__neon_mull_ld2r, 4, 16, 4, 1,  in qs8_gemm_4x16c4__neon_mull_ld2r()
1030   static void qs8_gemm_1x8c4__neon_mlal_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_1x8c4__neon_mlal_ld2r()  argument
1031     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c4__neon_mlal_ld2r, 1, 8, 4, 1,  in qs8_gemm_1x8c4__neon_mlal_ld2r()
1034   static void qs8_gemm_2x8c4__neon_mlal_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_2x8c4__neon_mlal_ld2r()  argument
1035     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c4__neon_mlal_ld2r, 2, 8, 4, 1,  in qs8_gemm_2x8c4__neon_mlal_ld2r()
1038   static void qs8_gemm_3x8c4__neon_mlal_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_3x8c4__neon_mlal_ld2r()  argument
1039     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c4__neon_mlal_ld2r, 3, 8, 4, 1,  in qs8_gemm_3x8c4__neon_mlal_ld2r()
1042   static void qs8_gemm_4x8c4__neon_mlal_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_4x8c4__neon_mlal_ld2r()  argument
1043     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c4__neon_mlal_ld2r, 4, 8, 4, 1,  in qs8_gemm_4x8c4__neon_mlal_ld2r()
1046   static void qs8_gemm_1x16c4__neon_mlal_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_1x16c4__neon_mlal_ld2r()  argument
1047     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c4__neon_mlal_ld2r, 1, 16, 4, 1,  in qs8_gemm_1x16c4__neon_mlal_ld2r()
1050   static void qs8_gemm_2x16c4__neon_mlal_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_2x16c4__neon_mlal_ld2r()  argument
1051     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c4__neon_mlal_ld2r, 2, 16, 4, 1,  in qs8_gemm_2x16c4__neon_mlal_ld2r()
1054   static void qs8_gemm_3x16c4__neon_mlal_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_3x16c4__neon_mlal_ld2r()  argument
1055     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c4__neon_mlal_ld2r, 3, 16, 4, 1,  in qs8_gemm_3x16c4__neon_mlal_ld2r()
1058   static void qs8_gemm_4x16c4__neon_mlal_ld2r(benchmark::State& state, const char* net) {  in qs8_gemm_4x16c4__neon_mlal_ld2r()  argument
1059     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c4__neon_mlal_ld2r, 4, 16, 4, 1,  in qs8_gemm_4x16c4__neon_mlal_ld2r()
1062   static void qs8_gemm_1x8c8__neon_mull(benchmark::State& state, const char* net) {  in qs8_gemm_1x8c8__neon_mull()  argument
1063     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c8__neon_mull, 1, 8, 8, 1,  in qs8_gemm_1x8c8__neon_mull()
1066   static void qs8_gemm_2x8c8__neon_mull(benchmark::State& state, const char* net) {  in qs8_gemm_2x8c8__neon_mull()  argument
1067     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c8__neon_mull, 2, 8, 8, 1,  in qs8_gemm_2x8c8__neon_mull()
1070   static void qs8_gemm_3x8c8__neon_mull(benchmark::State& state, const char* net) {  in qs8_gemm_3x8c8__neon_mull()  argument
1071     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c8__neon_mull, 3, 8, 8, 1,  in qs8_gemm_3x8c8__neon_mull()
1074   static void qs8_gemm_4x8c8__neon_mull(benchmark::State& state, const char* net) {  in qs8_gemm_4x8c8__neon_mull()  argument
1075     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c8__neon_mull, 4, 8, 8, 1,  in qs8_gemm_4x8c8__neon_mull()
1078   static void qs8_gemm_1x16c8__neon_mull(benchmark::State& state, const char* net) {  in qs8_gemm_1x16c8__neon_mull()  argument
1079     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c8__neon_mull, 1, 16, 8, 1,  in qs8_gemm_1x16c8__neon_mull()
1082   static void qs8_gemm_2x16c8__neon_mull(benchmark::State& state, const char* net) {  in qs8_gemm_2x16c8__neon_mull()  argument
1083     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c8__neon_mull, 2, 16, 8, 1,  in qs8_gemm_2x16c8__neon_mull()
1086   static void qs8_gemm_3x16c8__neon_mull(benchmark::State& state, const char* net) {  in qs8_gemm_3x16c8__neon_mull()  argument
1087     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c8__neon_mull, 3, 16, 8, 1,  in qs8_gemm_3x16c8__neon_mull()
1090   static void qs8_gemm_4x16c8__neon_mull(benchmark::State& state, const char* net) {  in qs8_gemm_4x16c8__neon_mull()  argument
1091     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c8__neon_mull, 4, 16, 8, 1,  in qs8_gemm_4x16c8__neon_mull()
1094   static void qs8_gemm_1x8c8__neon_mlal(benchmark::State& state, const char* net) {  in qs8_gemm_1x8c8__neon_mlal()  argument
1095     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c8__neon_mlal, 1, 8, 8, 1,  in qs8_gemm_1x8c8__neon_mlal()
1098   static void qs8_gemm_2x8c8__neon_mlal(benchmark::State& state, const char* net) {  in qs8_gemm_2x8c8__neon_mlal()  argument
1099     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c8__neon_mlal, 2, 8, 8, 1,  in qs8_gemm_2x8c8__neon_mlal()
1102   static void qs8_gemm_3x8c8__neon_mlal(benchmark::State& state, const char* net) {  in qs8_gemm_3x8c8__neon_mlal()  argument
1103     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c8__neon_mlal, 3, 8, 8, 1,  in qs8_gemm_3x8c8__neon_mlal()
1106   static void qs8_gemm_4x8c8__neon_mlal(benchmark::State& state, const char* net) {  in qs8_gemm_4x8c8__neon_mlal()  argument
1107     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c8__neon_mlal, 4, 8, 8, 1,  in qs8_gemm_4x8c8__neon_mlal()
1110   static void qs8_gemm_1x16c8__neon_mlal(benchmark::State& state, const char* net) {  in qs8_gemm_1x16c8__neon_mlal()  argument
1111     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c8__neon_mlal, 1, 16, 8, 1,  in qs8_gemm_1x16c8__neon_mlal()
1114   static void qs8_gemm_2x16c8__neon_mlal(benchmark::State& state, const char* net) {  in qs8_gemm_2x16c8__neon_mlal()  argument
1115     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c8__neon_mlal, 2, 16, 8, 1,  in qs8_gemm_2x16c8__neon_mlal()
1118   static void qs8_gemm_3x16c8__neon_mlal(benchmark::State& state, const char* net) {  in qs8_gemm_3x16c8__neon_mlal()  argument
1119     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c8__neon_mlal, 3, 16, 8, 1,  in qs8_gemm_3x16c8__neon_mlal()
1122   static void qs8_gemm_4x16c8__neon_mlal(benchmark::State& state, const char* net) {  in qs8_gemm_4x16c8__neon_mlal()  argument
1123     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c8__neon_mlal, 4, 16, 8, 1,  in qs8_gemm_4x16c8__neon_mlal()
1126   static void qs8_gemm_1x8c16__neon_mlal(benchmark::State& state, const char* net) {  in qs8_gemm_1x8c16__neon_mlal()  argument
1127     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x8c16__neon_mlal, 1, 8, 16, 1,  in qs8_gemm_1x8c16__neon_mlal()
1130   static void qs8_gemm_2x8c16__neon_mlal(benchmark::State& state, const char* net) {  in qs8_gemm_2x8c16__neon_mlal()  argument
1131     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x8c16__neon_mlal, 2, 8, 16, 1,  in qs8_gemm_2x8c16__neon_mlal()
1134   static void qs8_gemm_3x8c16__neon_mlal(benchmark::State& state, const char* net) {  in qs8_gemm_3x8c16__neon_mlal()  argument
1135     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x8c16__neon_mlal, 3, 8, 16, 1,  in qs8_gemm_3x8c16__neon_mlal()
1138   static void qs8_gemm_4x8c16__neon_mlal(benchmark::State& state, const char* net) {  in qs8_gemm_4x8c16__neon_mlal()  argument
1139     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x8c16__neon_mlal, 4, 8, 16, 1,  in qs8_gemm_4x8c16__neon_mlal()
1142   static void qs8_gemm_1x16c16__neon_mlal(benchmark::State& state, const char* net) {  in qs8_gemm_1x16c16__neon_mlal()  argument
1143     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_1x16c16__neon_mlal, 1, 16, 16, 1,  in qs8_gemm_1x16c16__neon_mlal()
1146   static void qs8_gemm_2x16c16__neon_mlal(benchmark::State& state, const char* net) {  in qs8_gemm_2x16c16__neon_mlal()  argument
1147     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_2x16c16__neon_mlal, 2, 16, 16, 1,  in qs8_gemm_2x16c16__neon_mlal()
1150   static void qs8_gemm_3x16c16__neon_mlal(benchmark::State& state, const char* net) {  in qs8_gemm_3x16c16__neon_mlal()  argument
1151     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_3x16c16__neon_mlal, 3, 16, 16, 1,  in qs8_gemm_3x16c16__neon_mlal()
1154   static void qs8_gemm_4x16c16__neon_mlal(benchmark::State& state, const char* net) {  in qs8_gemm_4x16c16__neon_mlal()  argument
1155     GEMMBenchmark(state, xnn_qs8_gemm_minmax_rndnu_ukernel_4x16c16__neon_mlal, 4, 16, 16, 1,  in qs8_gemm_4x16c16__neon_mlal()
1335   static void qs8_gemm_1x1c4__armsimd32(benchmark::State& state, const char* net) {  in BENCHMARK_GEMM()
1336     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_1x1c4__armsimd32, 1, 1, 4, 1,  in BENCHMARK_GEMM()
1339   static void qs8_gemm_2x1c4__armsimd32(benchmark::State& state, const char* net) {  in qs8_gemm_2x1c4__armsimd32()  argument
1340     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_1x1c4__armsimd32, 2, 1, 4, 1,  in qs8_gemm_2x1c4__armsimd32()
1343   static void qs8_gemm_1x2c4__armsimd32(benchmark::State& state, const char* net) {  in qs8_gemm_1x2c4__armsimd32()  argument
1344     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_1x2c4__armsimd32, 1, 2, 4, 1,  in qs8_gemm_1x2c4__armsimd32()
1347   static void qs8_gemm_2x2c4__armsimd32(benchmark::State& state, const char* net) {  in qs8_gemm_2x2c4__armsimd32()  argument
1348     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_1x2c4__armsimd32, 2, 2, 4, 1,  in qs8_gemm_2x2c4__armsimd32()
1360   static void qs8_gemm_2x16c8__avx512skx(benchmark::State& state, const char* net) {  in BENCHMARK_GEMM()
1361     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x16c8__avx512skx, 2, 16, 8, 1,  in BENCHMARK_GEMM()
1364   static void qs8_gemm_3x16c8__avx512skx(benchmark::State& state, const char* net) {  in qs8_gemm_3x16c8__avx512skx()  argument
1365     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x16c8__avx512skx, 3, 16, 8, 1,  in qs8_gemm_3x16c8__avx512skx()
1368   static void qs8_gemm_4x16c8__avx512skx(benchmark::State& state, const char* net) {  in qs8_gemm_4x16c8__avx512skx()  argument
1369     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x16c8__avx512skx, 4, 16, 8, 1,  in qs8_gemm_4x16c8__avx512skx()
1373   static void qs8_gemm_2x8c8__avx2(benchmark::State& state, const char* net) {  in qs8_gemm_2x8c8__avx2()  argument
1374     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x8c8__avx2, 2, 8, 8, 1,  in qs8_gemm_2x8c8__avx2()
1377   static void qs8_gemm_3x8c8__avx2(benchmark::State& state, const char* net) {  in qs8_gemm_3x8c8__avx2()  argument
1378     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x8c8__avx2, 3, 8, 8, 1,  in qs8_gemm_3x8c8__avx2()
1382   static void qs8_gemm_xw_2x8c8__avx2(benchmark::State& state, const char* net) {  in qs8_gemm_xw_2x8c8__avx2()  argument
1383     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_2x8c8__avx2, 2, 8, 8, 1,  in qs8_gemm_xw_2x8c8__avx2()
1386   static void qs8_gemm_xw_3x8c8__avx2(benchmark::State& state, const char* net) {  in qs8_gemm_xw_3x8c8__avx2()  argument
1387     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_3x8c8__avx2, 3, 8, 8, 1,  in qs8_gemm_xw_3x8c8__avx2()
1391   static void qs8_gemm_2x4c2__xop_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c2__xop_ld64()  argument
1392     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c2__xop_ld64, 2, 4, 2, 1,  in qs8_gemm_2x4c2__xop_ld64()
1395   static void qs8_gemm_3x4c2__xop_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c2__xop_ld64()  argument
1396     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c2__xop_ld64, 3, 4, 2, 1,  in qs8_gemm_3x4c2__xop_ld64()
1399   static void qs8_gemm_4x4c2__xop_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_4x4c2__xop_ld64()  argument
1400     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x4c2__xop_ld64, 4, 4, 2, 1,  in qs8_gemm_4x4c2__xop_ld64()
1404   static void qs8_gemm_2x4c2__xop_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c2__xop_ld128()  argument
1405     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c2__xop_ld128, 2, 4, 2, 1,  in qs8_gemm_2x4c2__xop_ld128()
1408   static void qs8_gemm_3x4c2__xop_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c2__xop_ld128()  argument
1409     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c2__xop_ld128, 3, 4, 2, 1,  in qs8_gemm_3x4c2__xop_ld128()
1412   static void qs8_gemm_4x4c2__xop_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_4x4c2__xop_ld128()  argument
1413     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x4c2__xop_ld128, 4, 4, 2, 1,  in qs8_gemm_4x4c2__xop_ld128()
1417   static void qs8_gemm_xw_2x4c2__xop(benchmark::State& state, const char* net) {  in qs8_gemm_xw_2x4c2__xop()  argument
1418     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_2x4c2__xop, 2, 4, 2, 1,  in qs8_gemm_xw_2x4c2__xop()
1421   static void qs8_gemm_xw_3x4c2__xop(benchmark::State& state, const char* net) {  in qs8_gemm_xw_3x4c2__xop()  argument
1422     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_3x4c2__xop, 3, 4, 2, 1,  in qs8_gemm_xw_3x4c2__xop()
1425   static void qs8_gemm_xw_4x4c2__xop(benchmark::State& state, const char* net) {  in qs8_gemm_xw_4x4c2__xop()  argument
1426     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_4x4c2__xop, 4, 4, 2, 1,  in qs8_gemm_xw_4x4c2__xop()
1430   static void qs8_gemm_2x4c2s4__xop_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c2s4__xop_ld64()  argument
1431     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c2s4__xop_ld64, 2, 4, 2, 4,  in qs8_gemm_2x4c2s4__xop_ld64()
1434   static void qs8_gemm_3x4c2s4__xop_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c2s4__xop_ld64()  argument
1435     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c2s4__xop_ld64, 3, 4, 2, 4,  in qs8_gemm_3x4c2s4__xop_ld64()
1438   static void qs8_gemm_4x4c2s4__xop_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_4x4c2s4__xop_ld64()  argument
1439     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x4c2s4__xop_ld64, 4, 4, 2, 4,  in qs8_gemm_4x4c2s4__xop_ld64()
1443   static void qs8_gemm_2x4c2s4__xop_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c2s4__xop_ld128()  argument
1444     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c2s4__xop_ld128, 2, 4, 2, 4,  in qs8_gemm_2x4c2s4__xop_ld128()
1447   static void qs8_gemm_3x4c2s4__xop_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c2s4__xop_ld128()  argument
1448     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c2s4__xop_ld128, 3, 4, 2, 4,  in qs8_gemm_3x4c2s4__xop_ld128()
1451   static void qs8_gemm_4x4c2s4__xop_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_4x4c2s4__xop_ld128()  argument
1452     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x4c2s4__xop_ld128, 4, 4, 2, 4,  in qs8_gemm_4x4c2s4__xop_ld128()
1456   static void qs8_gemm_xw_2x4c2s4__xop(benchmark::State& state, const char* net) {  in qs8_gemm_xw_2x4c2s4__xop()  argument
1457     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_2x4c2s4__xop, 2, 4, 2, 4,  in qs8_gemm_xw_2x4c2s4__xop()
1460   static void qs8_gemm_xw_3x4c2s4__xop(benchmark::State& state, const char* net) {  in qs8_gemm_xw_3x4c2s4__xop()  argument
1461     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_3x4c2s4__xop, 3, 4, 2, 4,  in qs8_gemm_xw_3x4c2s4__xop()
1464   static void qs8_gemm_xw_4x4c2s4__xop(benchmark::State& state, const char* net) {  in qs8_gemm_xw_4x4c2s4__xop()  argument
1465     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_4x4c2s4__xop, 4, 4, 2, 4,  in qs8_gemm_xw_4x4c2s4__xop()
1469   static void qs8_gemm_2x4c8__xop_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c8__xop_ld64()  argument
1470     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c8__xop_ld64, 2, 4, 8, 1,  in qs8_gemm_2x4c8__xop_ld64()
1473   static void qs8_gemm_3x4c8__xop_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c8__xop_ld64()  argument
1474     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c8__xop_ld64, 3, 4, 8, 1,  in qs8_gemm_3x4c8__xop_ld64()
1478   static void qs8_gemm_2x4c8__xop_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c8__xop_ld128()  argument
1479     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c8__xop_ld128, 2, 4, 8, 1,  in qs8_gemm_2x4c8__xop_ld128()
1482   static void qs8_gemm_3x4c8__xop_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c8__xop_ld128()  argument
1483     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c8__xop_ld128, 3, 4, 8, 1,  in qs8_gemm_3x4c8__xop_ld128()
1487   static void qs8_gemm_xw_2x4c8__xop(benchmark::State& state, const char* net) {  in qs8_gemm_xw_2x4c8__xop()  argument
1488     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_2x4c8__xop, 2, 4, 8, 1,  in qs8_gemm_xw_2x4c8__xop()
1491   static void qs8_gemm_xw_3x4c8__xop(benchmark::State& state, const char* net) {  in qs8_gemm_xw_3x4c8__xop()  argument
1492     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_3x4c8__xop, 3, 4, 8, 1,  in qs8_gemm_xw_3x4c8__xop()
1496   static void qs8_gemm_2x4c2__avx_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c2__avx_ld64()  argument
1497     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c2__avx_ld64, 2, 4, 2, 1,  in qs8_gemm_2x4c2__avx_ld64()
1500   static void qs8_gemm_3x4c2__avx_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c2__avx_ld64()  argument
1501     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c2__avx_ld64, 3, 4, 2, 1,  in qs8_gemm_3x4c2__avx_ld64()
1504   static void qs8_gemm_4x4c2__avx_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_4x4c2__avx_ld64()  argument
1505     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x4c2__avx_ld64, 4, 4, 2, 1,  in qs8_gemm_4x4c2__avx_ld64()
1509   static void qs8_gemm_2x4c2__avx_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c2__avx_ld128()  argument
1510     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c2__avx_ld128, 2, 4, 2, 1,  in qs8_gemm_2x4c2__avx_ld128()
1513   static void qs8_gemm_3x4c2__avx_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c2__avx_ld128()  argument
1514     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c2__avx_ld128, 3, 4, 2, 1,  in qs8_gemm_3x4c2__avx_ld128()
1517   static void qs8_gemm_4x4c2__avx_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_4x4c2__avx_ld128()  argument
1518     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x4c2__avx_ld128, 4, 4, 2, 1,  in qs8_gemm_4x4c2__avx_ld128()
1522   static void qs8_gemm_xw_2x4c2__avx(benchmark::State& state, const char* net) {  in qs8_gemm_xw_2x4c2__avx()  argument
1523     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_2x4c2__avx, 2, 4, 2, 1,  in qs8_gemm_xw_2x4c2__avx()
1526   static void qs8_gemm_xw_3x4c2__avx(benchmark::State& state, const char* net) {  in qs8_gemm_xw_3x4c2__avx()  argument
1527     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_3x4c2__avx, 3, 4, 2, 1,  in qs8_gemm_xw_3x4c2__avx()
1530   static void qs8_gemm_xw_4x4c2__avx(benchmark::State& state, const char* net) {  in qs8_gemm_xw_4x4c2__avx()  argument
1531     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_4x4c2__avx, 4, 4, 2, 1,  in qs8_gemm_xw_4x4c2__avx()
1535   static void qs8_gemm_2x4c2s4__avx_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c2s4__avx_ld64()  argument
1536     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c2s4__avx_ld64, 2, 4, 2, 4,  in qs8_gemm_2x4c2s4__avx_ld64()
1539   static void qs8_gemm_3x4c2s4__avx_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c2s4__avx_ld64()  argument
1540     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c2s4__avx_ld64, 3, 4, 2, 4,  in qs8_gemm_3x4c2s4__avx_ld64()
1543   static void qs8_gemm_4x4c2s4__avx_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_4x4c2s4__avx_ld64()  argument
1544     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x4c2s4__avx_ld64, 4, 4, 2, 4,  in qs8_gemm_4x4c2s4__avx_ld64()
1548   static void qs8_gemm_2x4c2s4__avx_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c2s4__avx_ld128()  argument
1549     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c2s4__avx_ld128, 2, 4, 2, 4,  in qs8_gemm_2x4c2s4__avx_ld128()
1552   static void qs8_gemm_3x4c2s4__avx_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c2s4__avx_ld128()  argument
1553     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c2s4__avx_ld128, 3, 4, 2, 4,  in qs8_gemm_3x4c2s4__avx_ld128()
1556   static void qs8_gemm_4x4c2s4__avx_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_4x4c2s4__avx_ld128()  argument
1557     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x4c2s4__avx_ld128, 4, 4, 2, 4,  in qs8_gemm_4x4c2s4__avx_ld128()
1561   static void qs8_gemm_xw_2x4c2s4__avx(benchmark::State& state, const char* net) {  in qs8_gemm_xw_2x4c2s4__avx()  argument
1562     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_2x4c2s4__avx, 2, 4, 2, 4,  in qs8_gemm_xw_2x4c2s4__avx()
1565   static void qs8_gemm_xw_3x4c2s4__avx(benchmark::State& state, const char* net) {  in qs8_gemm_xw_3x4c2s4__avx()  argument
1566     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_3x4c2s4__avx, 3, 4, 2, 4,  in qs8_gemm_xw_3x4c2s4__avx()
1569   static void qs8_gemm_xw_4x4c2s4__avx(benchmark::State& state, const char* net) {  in qs8_gemm_xw_4x4c2s4__avx()  argument
1570     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_4x4c2s4__avx, 4, 4, 2, 4,  in qs8_gemm_xw_4x4c2s4__avx()
1574   static void qs8_gemm_2x4c8__avx_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c8__avx_ld64()  argument
1575     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c8__avx_ld64, 2, 4, 8, 1,  in qs8_gemm_2x4c8__avx_ld64()
1578   static void qs8_gemm_3x4c8__avx_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c8__avx_ld64()  argument
1579     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c8__avx_ld64, 3, 4, 8, 1,  in qs8_gemm_3x4c8__avx_ld64()
1583   static void qs8_gemm_2x4c8__avx_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c8__avx_ld128()  argument
1584     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c8__avx_ld128, 2, 4, 8, 1,  in qs8_gemm_2x4c8__avx_ld128()
1587   static void qs8_gemm_3x4c8__avx_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c8__avx_ld128()  argument
1588     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c8__avx_ld128, 3, 4, 8, 1,  in qs8_gemm_3x4c8__avx_ld128()
1592   static void qs8_gemm_xw_2x4c8__avx(benchmark::State& state, const char* net) {  in qs8_gemm_xw_2x4c8__avx()  argument
1593     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_2x4c8__avx, 2, 4, 8, 1,  in qs8_gemm_xw_2x4c8__avx()
1596   static void qs8_gemm_xw_3x4c8__avx(benchmark::State& state, const char* net) {  in qs8_gemm_xw_3x4c8__avx()  argument
1597     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_3x4c8__avx, 3, 4, 8, 1,  in qs8_gemm_xw_3x4c8__avx()
1601   static void qs8_gemm_2x4c2__sse41_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c2__sse41_ld64()  argument
1602     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c2__sse41_ld64, 2, 4, 2, 1,  in qs8_gemm_2x4c2__sse41_ld64()
1605   static void qs8_gemm_3x4c2__sse41_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c2__sse41_ld64()  argument
1606     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c2__sse41_ld64, 3, 4, 2, 1,  in qs8_gemm_3x4c2__sse41_ld64()
1609   static void qs8_gemm_4x4c2__sse41_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_4x4c2__sse41_ld64()  argument
1610     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x4c2__sse41_ld64, 4, 4, 2, 1,  in qs8_gemm_4x4c2__sse41_ld64()
1614   static void qs8_gemm_2x4c2__sse41_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c2__sse41_ld128()  argument
1615     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c2__sse41_ld128, 2, 4, 2, 1,  in qs8_gemm_2x4c2__sse41_ld128()
1618   static void qs8_gemm_3x4c2__sse41_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c2__sse41_ld128()  argument
1619     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c2__sse41_ld128, 3, 4, 2, 1,  in qs8_gemm_3x4c2__sse41_ld128()
1622   static void qs8_gemm_4x4c2__sse41_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_4x4c2__sse41_ld128()  argument
1623     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x4c2__sse41_ld128, 4, 4, 2, 1,  in qs8_gemm_4x4c2__sse41_ld128()
1627   static void qs8_gemm_xw_2x4c2__sse41(benchmark::State& state, const char* net) {  in qs8_gemm_xw_2x4c2__sse41()  argument
1628     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_2x4c2__sse41, 2, 4, 2, 1,  in qs8_gemm_xw_2x4c2__sse41()
1631   static void qs8_gemm_xw_3x4c2__sse41(benchmark::State& state, const char* net) {  in qs8_gemm_xw_3x4c2__sse41()  argument
1632     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_3x4c2__sse41, 3, 4, 2, 1,  in qs8_gemm_xw_3x4c2__sse41()
1635   static void qs8_gemm_xw_4x4c2__sse41(benchmark::State& state, const char* net) {  in qs8_gemm_xw_4x4c2__sse41()  argument
1636     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_4x4c2__sse41, 4, 4, 2, 1,  in qs8_gemm_xw_4x4c2__sse41()
1640   static void qs8_gemm_2x4c2s4__sse41_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c2s4__sse41_ld64()  argument
1641     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c2s4__sse41_ld64, 2, 4, 2, 4,  in qs8_gemm_2x4c2s4__sse41_ld64()
1644   static void qs8_gemm_3x4c2s4__sse41_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c2s4__sse41_ld64()  argument
1645     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c2s4__sse41_ld64, 3, 4, 2, 4,  in qs8_gemm_3x4c2s4__sse41_ld64()
1648   static void qs8_gemm_4x4c2s4__sse41_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_4x4c2s4__sse41_ld64()  argument
1649     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x4c2s4__sse41_ld64, 4, 4, 2, 4,  in qs8_gemm_4x4c2s4__sse41_ld64()
1653   static void qs8_gemm_2x4c2s4__sse41_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c2s4__sse41_ld128()  argument
1654     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c2s4__sse41_ld128, 2, 4, 2, 4,  in qs8_gemm_2x4c2s4__sse41_ld128()
1657   static void qs8_gemm_3x4c2s4__sse41_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c2s4__sse41_ld128()  argument
1658     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c2s4__sse41_ld128, 3, 4, 2, 4,  in qs8_gemm_3x4c2s4__sse41_ld128()
1661   static void qs8_gemm_4x4c2s4__sse41_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_4x4c2s4__sse41_ld128()  argument
1662     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x4c2s4__sse41_ld128, 4, 4, 2, 4,  in qs8_gemm_4x4c2s4__sse41_ld128()
1666   static void qs8_gemm_xw_2x4c2s4__sse41(benchmark::State& state, const char* net) {  in qs8_gemm_xw_2x4c2s4__sse41()  argument
1667     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_2x4c2s4__sse41, 2, 4, 2, 4,  in qs8_gemm_xw_2x4c2s4__sse41()
1670   static void qs8_gemm_xw_3x4c2s4__sse41(benchmark::State& state, const char* net) {  in qs8_gemm_xw_3x4c2s4__sse41()  argument
1671     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_3x4c2s4__sse41, 3, 4, 2, 4,  in qs8_gemm_xw_3x4c2s4__sse41()
1674   static void qs8_gemm_xw_4x4c2s4__sse41(benchmark::State& state, const char* net) {  in qs8_gemm_xw_4x4c2s4__sse41()  argument
1675     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_4x4c2s4__sse41, 4, 4, 2, 4,  in qs8_gemm_xw_4x4c2s4__sse41()
1679   static void qs8_gemm_2x4c8__sse41_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c8__sse41_ld64()  argument
1680     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c8__sse41_ld64, 2, 4, 8, 1,  in qs8_gemm_2x4c8__sse41_ld64()
1683   static void qs8_gemm_3x4c8__sse41_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c8__sse41_ld64()  argument
1684     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c8__sse41_ld64, 3, 4, 8, 1,  in qs8_gemm_3x4c8__sse41_ld64()
1688   static void qs8_gemm_2x4c8__sse41_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c8__sse41_ld128()  argument
1689     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c8__sse41_ld128, 2, 4, 8, 1,  in qs8_gemm_2x4c8__sse41_ld128()
1692   static void qs8_gemm_3x4c8__sse41_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c8__sse41_ld128()  argument
1693     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c8__sse41_ld128, 3, 4, 8, 1,  in qs8_gemm_3x4c8__sse41_ld128()
1697   static void qs8_gemm_xw_2x4c8__sse41(benchmark::State& state, const char* net) {  in qs8_gemm_xw_2x4c8__sse41()  argument
1698     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_2x4c8__sse41, 2, 4, 8, 1,  in qs8_gemm_xw_2x4c8__sse41()
1701   static void qs8_gemm_xw_3x4c8__sse41(benchmark::State& state, const char* net) {  in qs8_gemm_xw_3x4c8__sse41()  argument
1702     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_3x4c8__sse41, 3, 4, 8, 1,  in qs8_gemm_xw_3x4c8__sse41()
1706   static void qs8_gemm_2x4c8__ssse3_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c8__ssse3_ld64()  argument
1707     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c8__ssse3_ld64, 2, 4, 8, 1,  in qs8_gemm_2x4c8__ssse3_ld64()
1710   static void qs8_gemm_3x4c8__ssse3_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c8__ssse3_ld64()  argument
1711     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c8__ssse3_ld64, 3, 4, 8, 1,  in qs8_gemm_3x4c8__ssse3_ld64()
1715   static void qs8_gemm_2x4c8__ssse3_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c8__ssse3_ld128()  argument
1716     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c8__ssse3_ld128, 2, 4, 8, 1,  in qs8_gemm_2x4c8__ssse3_ld128()
1719   static void qs8_gemm_3x4c8__ssse3_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c8__ssse3_ld128()  argument
1720     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c8__ssse3_ld128, 3, 4, 8, 1,  in qs8_gemm_3x4c8__ssse3_ld128()
1724   static void qs8_gemm_xw_2x4c8__ssse3(benchmark::State& state, const char* net) {  in qs8_gemm_xw_2x4c8__ssse3()  argument
1725     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_2x4c8__ssse3, 2, 4, 8, 1,  in qs8_gemm_xw_2x4c8__ssse3()
1728   static void qs8_gemm_xw_3x4c8__ssse3(benchmark::State& state, const char* net) {  in qs8_gemm_xw_3x4c8__ssse3()  argument
1729     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_3x4c8__ssse3, 3, 4, 8, 1,  in qs8_gemm_xw_3x4c8__ssse3()
1733   static void qs8_gemm_2x4c2__sse2_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c2__sse2_ld64()  argument
1734     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c2__sse2_ld64, 2, 4, 2, 1,  in qs8_gemm_2x4c2__sse2_ld64()
1737   static void qs8_gemm_3x4c2__sse2_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c2__sse2_ld64()  argument
1738     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c2__sse2_ld64, 3, 4, 2, 1,  in qs8_gemm_3x4c2__sse2_ld64()
1741   static void qs8_gemm_4x4c2__sse2_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_4x4c2__sse2_ld64()  argument
1742     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x4c2__sse2_ld64, 4, 4, 2, 1,  in qs8_gemm_4x4c2__sse2_ld64()
1746   static void qs8_gemm_2x4c2__sse2_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c2__sse2_ld128()  argument
1747     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c2__sse2_ld128, 2, 4, 2, 1,  in qs8_gemm_2x4c2__sse2_ld128()
1750   static void qs8_gemm_3x4c2__sse2_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c2__sse2_ld128()  argument
1751     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c2__sse2_ld128, 3, 4, 2, 1,  in qs8_gemm_3x4c2__sse2_ld128()
1754   static void qs8_gemm_4x4c2__sse2_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_4x4c2__sse2_ld128()  argument
1755     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x4c2__sse2_ld128, 4, 4, 2, 1,  in qs8_gemm_4x4c2__sse2_ld128()
1759   static void qs8_gemm_xw_2x4c2__sse2(benchmark::State& state, const char* net) {  in qs8_gemm_xw_2x4c2__sse2()  argument
1760     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_2x4c2__sse2, 2, 4, 2, 1,  in qs8_gemm_xw_2x4c2__sse2()
1763   static void qs8_gemm_xw_3x4c2__sse2(benchmark::State& state, const char* net) {  in qs8_gemm_xw_3x4c2__sse2()  argument
1764     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_3x4c2__sse2, 3, 4, 2, 1,  in qs8_gemm_xw_3x4c2__sse2()
1767   static void qs8_gemm_xw_4x4c2__sse2(benchmark::State& state, const char* net) {  in qs8_gemm_xw_4x4c2__sse2()  argument
1768     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_4x4c2__sse2, 4, 4, 2, 1,  in qs8_gemm_xw_4x4c2__sse2()
1772   static void qs8_gemm_2x4c2s4__sse2_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c2s4__sse2_ld64()  argument
1773     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c2s4__sse2_ld64, 2, 4, 2, 4,  in qs8_gemm_2x4c2s4__sse2_ld64()
1776   static void qs8_gemm_3x4c2s4__sse2_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c2s4__sse2_ld64()  argument
1777     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c2s4__sse2_ld64, 3, 4, 2, 4,  in qs8_gemm_3x4c2s4__sse2_ld64()
1780   static void qs8_gemm_4x4c2s4__sse2_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_4x4c2s4__sse2_ld64()  argument
1781     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x4c2s4__sse2_ld64, 4, 4, 2, 4,  in qs8_gemm_4x4c2s4__sse2_ld64()
1785   static void qs8_gemm_2x4c2s4__sse2_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c2s4__sse2_ld128()  argument
1786     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c2s4__sse2_ld128, 2, 4, 2, 4,  in qs8_gemm_2x4c2s4__sse2_ld128()
1789   static void qs8_gemm_3x4c2s4__sse2_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c2s4__sse2_ld128()  argument
1790     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c2s4__sse2_ld128, 3, 4, 2, 4,  in qs8_gemm_3x4c2s4__sse2_ld128()
1793   static void qs8_gemm_4x4c2s4__sse2_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_4x4c2s4__sse2_ld128()  argument
1794     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x4c2s4__sse2_ld128, 4, 4, 2, 4,  in qs8_gemm_4x4c2s4__sse2_ld128()
1798   static void qs8_gemm_xw_2x4c2s4__sse2(benchmark::State& state, const char* net) {  in qs8_gemm_xw_2x4c2s4__sse2()  argument
1799     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_2x4c2s4__sse2, 2, 4, 2, 4,  in qs8_gemm_xw_2x4c2s4__sse2()
1802   static void qs8_gemm_xw_3x4c2s4__sse2(benchmark::State& state, const char* net) {  in qs8_gemm_xw_3x4c2s4__sse2()  argument
1803     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_3x4c2s4__sse2, 3, 4, 2, 4,  in qs8_gemm_xw_3x4c2s4__sse2()
1806   static void qs8_gemm_xw_4x4c2s4__sse2(benchmark::State& state, const char* net) {  in qs8_gemm_xw_4x4c2s4__sse2()  argument
1807     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_4x4c2s4__sse2, 4, 4, 2, 4,  in qs8_gemm_xw_4x4c2s4__sse2()
1811   static void qs8_gemm_2x4c8__sse2_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c8__sse2_ld64()  argument
1812     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c8__sse2_ld64, 2, 4, 8, 1,  in qs8_gemm_2x4c8__sse2_ld64()
1815   static void qs8_gemm_3x4c8__sse2_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c8__sse2_ld64()  argument
1816     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c8__sse2_ld64, 3, 4, 8, 1,  in qs8_gemm_3x4c8__sse2_ld64()
1820   static void qs8_gemm_2x4c8__sse2_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c8__sse2_ld128()  argument
1821     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c8__sse2_ld128, 2, 4, 8, 1,  in qs8_gemm_2x4c8__sse2_ld128()
1824   static void qs8_gemm_3x4c8__sse2_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c8__sse2_ld128()  argument
1825     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c8__sse2_ld128, 3, 4, 8, 1,  in qs8_gemm_3x4c8__sse2_ld128()
1829   static void qs8_gemm_xw_2x4c8__sse2(benchmark::State& state, const char* net) {  in qs8_gemm_xw_2x4c8__sse2()  argument
1830     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_2x4c8__sse2, 2, 4, 8, 1,  in qs8_gemm_xw_2x4c8__sse2()
1833   static void qs8_gemm_xw_3x4c8__sse2(benchmark::State& state, const char* net) {  in qs8_gemm_xw_3x4c8__sse2()  argument
1834     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_3x4c8__sse2, 3, 4, 8, 1,  in qs8_gemm_xw_3x4c8__sse2()
1921   static void qs8_gemm_2x4c2__wasmsimd_dot16x2_ld64(benchmark::State& state, const char* net) {  in BENCHMARK_GEMM()
1922     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c2__wasmsimd_dot16x2_ld64, 2, 4, 2, 1,  in BENCHMARK_GEMM()
1925   static void qs8_gemm_3x4c2__wasmsimd_dot16x2_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c2__wasmsimd_dot16x2_ld64()  argument
1926     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c2__wasmsimd_dot16x2_ld64, 3, 4, 2, 1,  in qs8_gemm_3x4c2__wasmsimd_dot16x2_ld64()
1929   static void qs8_gemm_4x4c2__wasmsimd_dot16x2_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_4x4c2__wasmsimd_dot16x2_ld64()  argument
1930     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x4c2__wasmsimd_dot16x2_ld64, 4, 4, 2, 1,  in qs8_gemm_4x4c2__wasmsimd_dot16x2_ld64()
1934   static void qs8_gemm_2x4c2__wasmsimd_dot16x2_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c2__wasmsimd_dot16x2_ld128()  argument
1935     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c2__wasmsimd_dot16x2_ld128, 2, 4, 2, 1,  in qs8_gemm_2x4c2__wasmsimd_dot16x2_ld128()
1938   static void qs8_gemm_3x4c2__wasmsimd_dot16x2_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c2__wasmsimd_dot16x2_ld128()  argument
1939     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c2__wasmsimd_dot16x2_ld128, 3, 4, 2, 1,  in qs8_gemm_3x4c2__wasmsimd_dot16x2_ld128()
1942   static void qs8_gemm_4x4c2__wasmsimd_dot16x2_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_4x4c2__wasmsimd_dot16x2_ld128()  argument
1943     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x4c2__wasmsimd_dot16x2_ld128, 4, 4, 2, 1,  in qs8_gemm_4x4c2__wasmsimd_dot16x2_ld128()
1947   static void qs8_gemm_xw_2x4c2__wasmsimd_dot16x2(benchmark::State& state, const char* net) {  in qs8_gemm_xw_2x4c2__wasmsimd_dot16x2()  argument
1948     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_2x4c2__wasmsimd_dot16x2, 2, 4, 2, 1,  in qs8_gemm_xw_2x4c2__wasmsimd_dot16x2()
1951   static void qs8_gemm_xw_3x4c2__wasmsimd_dot16x2(benchmark::State& state, const char* net) {  in qs8_gemm_xw_3x4c2__wasmsimd_dot16x2()  argument
1952     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_3x4c2__wasmsimd_dot16x2, 3, 4, 2, 1,  in qs8_gemm_xw_3x4c2__wasmsimd_dot16x2()
1955   static void qs8_gemm_xw_4x4c2__wasmsimd_dot16x2(benchmark::State& state, const char* net) {  in qs8_gemm_xw_4x4c2__wasmsimd_dot16x2()  argument
1956     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_4x4c2__wasmsimd_dot16x2, 4, 4, 2, 1,  in qs8_gemm_xw_4x4c2__wasmsimd_dot16x2()
1960   static void qs8_gemm_2x4c2s4__wasmsimd_dot16x2_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c2s4__wasmsimd_dot16x2_ld64()  argument
1961 …  GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c2s4__wasmsimd_dot16x2_ld64, 2, 4, 2, 4,  in qs8_gemm_2x4c2s4__wasmsimd_dot16x2_ld64()
1964   static void qs8_gemm_3x4c2s4__wasmsimd_dot16x2_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c2s4__wasmsimd_dot16x2_ld64()  argument
1965 …  GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c2s4__wasmsimd_dot16x2_ld64, 3, 4, 2, 4,  in qs8_gemm_3x4c2s4__wasmsimd_dot16x2_ld64()
1968   static void qs8_gemm_4x4c2s4__wasmsimd_dot16x2_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_4x4c2s4__wasmsimd_dot16x2_ld64()  argument
1969 …  GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x4c2s4__wasmsimd_dot16x2_ld64, 4, 4, 2, 4,  in qs8_gemm_4x4c2s4__wasmsimd_dot16x2_ld64()
1973   static void qs8_gemm_2x4c2s4__wasmsimd_dot16x2_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c2s4__wasmsimd_dot16x2_ld128()  argument
1974 … GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c2s4__wasmsimd_dot16x2_ld128, 2, 4, 2, 4,  in qs8_gemm_2x4c2s4__wasmsimd_dot16x2_ld128()
1977   static void qs8_gemm_3x4c2s4__wasmsimd_dot16x2_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c2s4__wasmsimd_dot16x2_ld128()  argument
1978 … GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c2s4__wasmsimd_dot16x2_ld128, 3, 4, 2, 4,  in qs8_gemm_3x4c2s4__wasmsimd_dot16x2_ld128()
1981   static void qs8_gemm_4x4c2s4__wasmsimd_dot16x2_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_4x4c2s4__wasmsimd_dot16x2_ld128()  argument
1982 … GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x4c2s4__wasmsimd_dot16x2_ld128, 4, 4, 2, 4,  in qs8_gemm_4x4c2s4__wasmsimd_dot16x2_ld128()
1986   static void qs8_gemm_2x4c8__wasmsimd_dot16x2_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c8__wasmsimd_dot16x2_ld64()  argument
1987     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c8__wasmsimd_dot16x2_ld64, 2, 4, 8, 1,  in qs8_gemm_2x4c8__wasmsimd_dot16x2_ld64()
1990   static void qs8_gemm_3x4c8__wasmsimd_dot16x2_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c8__wasmsimd_dot16x2_ld64()  argument
1991     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c8__wasmsimd_dot16x2_ld64, 3, 4, 8, 1,  in qs8_gemm_3x4c8__wasmsimd_dot16x2_ld64()
1994   static void qs8_gemm_4x4c8__wasmsimd_dot16x2_ld64(benchmark::State& state, const char* net) {  in qs8_gemm_4x4c8__wasmsimd_dot16x2_ld64()  argument
1995     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x4c8__wasmsimd_dot16x2_ld64, 4, 4, 8, 1,  in qs8_gemm_4x4c8__wasmsimd_dot16x2_ld64()
1999   static void qs8_gemm_2x4c8__wasmsimd_dot16x2_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_2x4c8__wasmsimd_dot16x2_ld128()  argument
2000     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4c8__wasmsimd_dot16x2_ld128, 2, 4, 8, 1,  in qs8_gemm_2x4c8__wasmsimd_dot16x2_ld128()
2003   static void qs8_gemm_3x4c8__wasmsimd_dot16x2_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_3x4c8__wasmsimd_dot16x2_ld128()  argument
2004     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4c8__wasmsimd_dot16x2_ld128, 3, 4, 8, 1,  in qs8_gemm_3x4c8__wasmsimd_dot16x2_ld128()
2007   static void qs8_gemm_4x4c8__wasmsimd_dot16x2_ld128(benchmark::State& state, const char* net) {  in qs8_gemm_4x4c8__wasmsimd_dot16x2_ld128()  argument
2008     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x4c8__wasmsimd_dot16x2_ld128, 4, 4, 8, 1,  in qs8_gemm_4x4c8__wasmsimd_dot16x2_ld128()
2012   static void qs8_gemm_xw_2x4c8__wasmsimd_dot16x2(benchmark::State& state, const char* net) {  in qs8_gemm_xw_2x4c8__wasmsimd_dot16x2()  argument
2013     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_2x4c8__wasmsimd_dot16x2, 2, 4, 8, 1,  in qs8_gemm_xw_2x4c8__wasmsimd_dot16x2()
2016   static void qs8_gemm_xw_3x4c8__wasmsimd_dot16x2(benchmark::State& state, const char* net) {  in qs8_gemm_xw_3x4c8__wasmsimd_dot16x2()  argument
2017     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_3x4c8__wasmsimd_dot16x2, 3, 4, 8, 1,  in qs8_gemm_xw_3x4c8__wasmsimd_dot16x2()
2020   static void qs8_gemm_xw_4x4c8__wasmsimd_dot16x2(benchmark::State& state, const char* net) {  in qs8_gemm_xw_4x4c8__wasmsimd_dot16x2()  argument
2021     GEMMBenchmark(state, xnn_qs8_gemm_xw_minmax_fp32_ukernel_4x4c8__wasmsimd_dot16x2, 4, 4, 8, 1,  in qs8_gemm_xw_4x4c8__wasmsimd_dot16x2()
2055   static void qs8_gemm_2x2__wasm_fmagic(benchmark::State& state, const char* net) {  in BENCHMARK_GEMM()
2056     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x2__wasm_fmagic, 2, 2, 1, 1,  in BENCHMARK_GEMM()
2059   static void qs8_gemm_3x2__wasm_fmagic(benchmark::State& state, const char* net) {  in qs8_gemm_3x2__wasm_fmagic()  argument
2060     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x2__wasm_fmagic, 3, 2, 1, 1,  in qs8_gemm_3x2__wasm_fmagic()
2063   static void qs8_gemm_4x2__wasm_fmagic(benchmark::State& state, const char* net) {  in qs8_gemm_4x2__wasm_fmagic()  argument
2064     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x2__wasm_fmagic, 4, 2, 1, 1,  in qs8_gemm_4x2__wasm_fmagic()
2067   static void qs8_gemm_2x4__wasm_fmagic(benchmark::State& state, const char* net) {  in qs8_gemm_2x4__wasm_fmagic()  argument
2068     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4__wasm_fmagic, 2, 4, 1, 1,  in qs8_gemm_2x4__wasm_fmagic()
2071   static void qs8_gemm_3x4__wasm_fmagic(benchmark::State& state, const char* net) {  in qs8_gemm_3x4__wasm_fmagic()  argument
2072     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4__wasm_fmagic, 3, 4, 1, 1,  in qs8_gemm_3x4__wasm_fmagic()
2075   static void qs8_gemm_4x4__wasm_fmagic(benchmark::State& state, const char* net) {  in qs8_gemm_4x4__wasm_fmagic()  argument
2076     GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x4__wasm_fmagic, 4, 4, 1, 1,  in qs8_gemm_4x4__wasm_fmagic()
2089 static void qs8_gemm_2x2__scalar_fmagic(benchmark::State& state, const char* net) {  in BENCHMARK_GEMM()
2090   GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x2__scalar_fmagic, 2, 2, 1, 1,  in BENCHMARK_GEMM()
2093 static void qs8_gemm_3x2__scalar_fmagic(benchmark::State& state, const char* net) {  in qs8_gemm_3x2__scalar_fmagic()  argument
2094   GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x2__scalar_fmagic, 3, 2, 1, 1,  in qs8_gemm_3x2__scalar_fmagic()
2097 static void qs8_gemm_4x2__scalar_fmagic(benchmark::State& state, const char* net) {  in qs8_gemm_4x2__scalar_fmagic()  argument
2098   GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x2__scalar_fmagic, 4, 2, 1, 1,  in qs8_gemm_4x2__scalar_fmagic()
2101 static void qs8_gemm_2x4__scalar_fmagic(benchmark::State& state, const char* net) {  in qs8_gemm_2x4__scalar_fmagic()  argument
2102   GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4__scalar_fmagic, 2, 4, 1, 1,  in qs8_gemm_2x4__scalar_fmagic()
2105 static void qs8_gemm_3x4__scalar_fmagic(benchmark::State& state, const char* net) {  in qs8_gemm_3x4__scalar_fmagic()  argument
2106   GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4__scalar_fmagic, 3, 4, 1, 1,  in qs8_gemm_3x4__scalar_fmagic()
2109 static void qs8_gemm_4x4__scalar_fmagic(benchmark::State& state, const char* net) {  in qs8_gemm_4x4__scalar_fmagic()  argument
2110   GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x4__scalar_fmagic, 4, 4, 1, 1,  in qs8_gemm_4x4__scalar_fmagic()
2114 static void qs8_gemm_2x2__scalar_imagic(benchmark::State& state, const char* net) {  in qs8_gemm_2x2__scalar_imagic()  argument
2115   GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x2__scalar_imagic, 2, 2, 1, 1,  in qs8_gemm_2x2__scalar_imagic()
2118 static void qs8_gemm_3x2__scalar_imagic(benchmark::State& state, const char* net) {  in qs8_gemm_3x2__scalar_imagic()  argument
2119   GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x2__scalar_imagic, 3, 2, 1, 1,  in qs8_gemm_3x2__scalar_imagic()
2122 static void qs8_gemm_4x2__scalar_imagic(benchmark::State& state, const char* net) {  in qs8_gemm_4x2__scalar_imagic()  argument
2123   GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x2__scalar_imagic, 4, 2, 1, 1,  in qs8_gemm_4x2__scalar_imagic()
2126 static void qs8_gemm_2x4__scalar_imagic(benchmark::State& state, const char* net) {  in qs8_gemm_2x4__scalar_imagic()  argument
2127   GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4__scalar_imagic, 2, 4, 1, 1,  in qs8_gemm_2x4__scalar_imagic()
2130 static void qs8_gemm_3x4__scalar_imagic(benchmark::State& state, const char* net) {  in qs8_gemm_3x4__scalar_imagic()  argument
2131   GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4__scalar_imagic, 3, 4, 1, 1,  in qs8_gemm_3x4__scalar_imagic()
2134 static void qs8_gemm_4x4__scalar_imagic(benchmark::State& state, const char* net) {  in qs8_gemm_4x4__scalar_imagic()  argument
2135   GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x4__scalar_imagic, 4, 4, 1, 1,  in qs8_gemm_4x4__scalar_imagic()
2139 static void qs8_gemm_2x2__scalar_lrintf(benchmark::State& state, const char* net) {  in qs8_gemm_2x2__scalar_lrintf()  argument
2140   GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x2__scalar_lrintf, 2, 2, 1, 1,  in qs8_gemm_2x2__scalar_lrintf()
2143 static void qs8_gemm_3x2__scalar_lrintf(benchmark::State& state, const char* net) {  in qs8_gemm_3x2__scalar_lrintf()  argument
2144   GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x2__scalar_lrintf, 3, 2, 1, 1,  in qs8_gemm_3x2__scalar_lrintf()
2147 static void qs8_gemm_4x2__scalar_lrintf(benchmark::State& state, const char* net) {  in qs8_gemm_4x2__scalar_lrintf()  argument
2148   GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x2__scalar_lrintf, 4, 2, 1, 1,  in qs8_gemm_4x2__scalar_lrintf()
2151 static void qs8_gemm_2x4__scalar_lrintf(benchmark::State& state, const char* net) {  in qs8_gemm_2x4__scalar_lrintf()  argument
2152   GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_2x4__scalar_lrintf, 2, 4, 1, 1,  in qs8_gemm_2x4__scalar_lrintf()
2155 static void qs8_gemm_3x4__scalar_lrintf(benchmark::State& state, const char* net) {  in qs8_gemm_3x4__scalar_lrintf()  argument
2156   GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_3x4__scalar_lrintf, 3, 4, 1, 1,  in qs8_gemm_3x4__scalar_lrintf()
2159 static void qs8_gemm_4x4__scalar_lrintf(benchmark::State& state, const char* net) {  in qs8_gemm_4x4__scalar_lrintf()  argument
2160   GEMMBenchmark(state, xnn_qs8_gemm_minmax_fp32_ukernel_4x4__scalar_lrintf, 4, 4, 1, 1,  in qs8_gemm_4x4__scalar_lrintf()