38 static void GEMMBenchmark(benchmark::State& state,  in GEMMBenchmark()  argument
44   if (isa_check && !isa_check(state)) {  in GEMMBenchmark()
48   const size_t mc = state.range(0);  in GEMMBenchmark()
49   const size_t nc = state.range(1);  in GEMMBenchmark()
50   const size_t kc = state.range(2);  in GEMMBenchmark()
83   for (auto _ : state) {  in GEMMBenchmark()
84     // Use circular buffers (exceeding cache size) and prefetch to control cache state:  in GEMMBenchmark()
88     state.PauseTiming();  in GEMMBenchmark()
91     state.ResumeTiming();  in GEMMBenchmark()
106     state.counters["cpufreq"] = cpu_frequency;  in GEMMBenchmark()
109   state.counters["FLOPS"] = benchmark::Counter(  in GEMMBenchmark()
110     uint64_t(state.iterations()) * 2 * mc * nc * kc, benchmark::Counter::kIsRate);  in GEMMBenchmark()
113 static void PPMM1PBenchmark(benchmark::State& state,  in PPMM1PBenchmark()  argument
120   if (isa_check && !isa_check(state)) {  in PPMM1PBenchmark()
124   const size_t mc = state.range(0);  in PPMM1PBenchmark()
125   const size_t nc = state.range(1);  in PPMM1PBenchmark()
126   const size_t kc = state.range(2);  in PPMM1PBenchmark()
160   for (auto _ : state) {  in PPMM1PBenchmark()
161     // Use circular buffers (exceeding cache size) and prefetch to control cache state:  in PPMM1PBenchmark()
165     state.PauseTiming();  in PPMM1PBenchmark()
168     state.ResumeTiming();  in PPMM1PBenchmark()
184     state.counters["cpufreq"] = cpu_frequency;  in PPMM1PBenchmark()
187   state.counters["FLOPS"] = benchmark::Counter(  in PPMM1PBenchmark()
188     uint64_t(state.iterations()) * 2 * mc * nc * kc, benchmark::Counter::kIsRate);  in PPMM1PBenchmark()
191 static void PPMM2PBenchmark(benchmark::State& state,  in PPMM2PBenchmark()  argument
198   if (isa_check && !isa_check(state)) {  in PPMM2PBenchmark()
202   const size_t mc = state.range(0);  in PPMM2PBenchmark()
203   const size_t nc = state.range(1);  in PPMM2PBenchmark()
204   const size_t kc = state.range(2);  in PPMM2PBenchmark()
239   for (auto _ : state) {  in PPMM2PBenchmark()
240     // Use circular buffers (exceeding cache size) and prefetch to control cache state:  in PPMM2PBenchmark()
244     state.PauseTiming();  in PPMM2PBenchmark()
247     state.ResumeTiming();  in PPMM2PBenchmark()
266     state.counters["cpufreq"] = cpu_frequency;  in PPMM2PBenchmark()
269   state.counters["FLOPS"] = benchmark::Counter(  in PPMM2PBenchmark()
270     uint64_t(state.iterations()) * 2 * mc * nc * kc, benchmark::Counter::kIsRate);  in PPMM2PBenchmark()
274 static void RuyBenchmark(benchmark::State& state, uint32_t threads)  in RuyBenchmark()  argument
280   const size_t mc = state.range(0);  in RuyBenchmark()
281   const size_t nc = state.range(1);  in RuyBenchmark()
282   const size_t kc = state.range(2);  in RuyBenchmark()
328   for (auto _ : state) {  in RuyBenchmark()
329     // Use circular buffers (exceeding cache size) and prefetch to control cache state:  in RuyBenchmark()
334     state.PauseTiming();  in RuyBenchmark()
337     state.ResumeTiming();  in RuyBenchmark()
348     state.counters["cpufreq"] = cpu_frequency;  in RuyBenchmark()
351   state.counters["FLOPS"] = benchmark::Counter(  in RuyBenchmark()
352     uint64_t(state.iterations()) * 2 * mc * nc * kc, benchmark::Counter::kIsRate);  in RuyBenchmark()
355 static void ruy_st(benchmark::State& state, const char* net)  in ruy_st()  argument
357   RuyBenchmark(state, 1);  in ruy_st()
362 static void GEMMBenchmark(benchmark::State& state,  in GEMMBenchmark()  argument
368   if (isa_check && !isa_check(state)) {  in GEMMBenchmark()
372   const size_t mc = state.range(0);  in GEMMBenchmark()
373   const size_t nc = state.range(1);  in GEMMBenchmark()
374   const size_t kc = state.range(2);  in GEMMBenchmark()
420   for (auto _ : state) {  in GEMMBenchmark()
421     // Use circular buffers (exceeding cache size) and prefetch to control cache state:  in GEMMBenchmark()
425     state.PauseTiming();  in GEMMBenchmark()
428     state.ResumeTiming();  in GEMMBenchmark()
445     state.counters["cpufreq"] = cpu_frequency;  in GEMMBenchmark()
448   state.counters["FLOPS"] = benchmark::Counter(  in GEMMBenchmark()
449     uint64_t(state.iterations()) * 2 * mc * nc * kc, benchmark::Counter::kIsRate);  in GEMMBenchmark()
454   static void f32_gemm_1x8__aarch64_neonfma_ld64(benchmark::State& state, const char* net) {  in f32_gemm_1x8__aarch64_neonfma_ld64()  argument
455     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_1x8__aarch64_neonfma_ld64, 1, 8, 1, 1,  in f32_gemm_1x8__aarch64_neonfma_ld64()
458   static void f32_gemm_1x12__aarch64_neonfma_cortex_a53(benchmark::State& state, const char* net) {  in f32_gemm_1x12__aarch64_neonfma_cortex_a53()  argument
459     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_1x12__aarch64_neonfma_cortex_a53, 1, 12, 1, 1,  in f32_gemm_1x12__aarch64_neonfma_cortex_a53()
462   static void f32_gemm_1x8__aarch64_neonfma_cortex_a53(benchmark::State& state, const char* net) {  in f32_gemm_1x8__aarch64_neonfma_cortex_a53()  argument
463     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_1x8__aarch64_neonfma_cortex_a53, 1, 8, 1, 1,  in f32_gemm_1x8__aarch64_neonfma_cortex_a53()
466 …static void f32_gemm_1x8__aarch64_neonfma_prfm_cortex_a53(benchmark::State& state, const char* net…  in f32_gemm_1x8__aarch64_neonfma_prfm_cortex_a53()  argument
467 … GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_1x8__aarch64_neonfma_prfm_cortex_a53, 1, 8, 1, 1,  in f32_gemm_1x8__aarch64_neonfma_prfm_cortex_a53()
470   static void f32_gemm_1x8__aarch64_neonfma_cortex_a75(benchmark::State& state, const char* net) {  in f32_gemm_1x8__aarch64_neonfma_cortex_a75()  argument
471     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_1x8__aarch64_neonfma_cortex_a75, 1, 8, 1, 1,  in f32_gemm_1x8__aarch64_neonfma_cortex_a75()
474 …static void f32_gemm_1x8__aarch64_neonfma_prfm_cortex_a75(benchmark::State& state, const char* net…  in f32_gemm_1x8__aarch64_neonfma_prfm_cortex_a75()  argument
475 … GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_1x8__aarch64_neonfma_prfm_cortex_a75, 1, 8, 1, 1,  in f32_gemm_1x8__aarch64_neonfma_prfm_cortex_a75()
478   static void f32_gemm_4x12__aarch64_neonfma_cortex_a53(benchmark::State& state, const char* net) {  in f32_gemm_4x12__aarch64_neonfma_cortex_a53()  argument
479     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x12__aarch64_neonfma_cortex_a53, 4, 12, 1, 1,  in f32_gemm_4x12__aarch64_neonfma_cortex_a53()
482   static void f32_gemm_4x8__aarch64_neonfma_cortex_a53(benchmark::State& state, const char* net) {  in f32_gemm_4x8__aarch64_neonfma_cortex_a53()  argument
483     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__aarch64_neonfma_cortex_a53, 4, 8, 1, 1,  in f32_gemm_4x8__aarch64_neonfma_cortex_a53()
486 …static void f32_gemm_4x8__aarch64_neonfma_prfm_cortex_a53(benchmark::State& state, const char* net…  in f32_gemm_4x8__aarch64_neonfma_prfm_cortex_a53()  argument
487 … GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__aarch64_neonfma_prfm_cortex_a53, 4, 8, 1, 1,  in f32_gemm_4x8__aarch64_neonfma_prfm_cortex_a53()
490   static void f32_gemm_4x8__aarch64_neonfma_cortex_a55(benchmark::State& state, const char* net) {  in f32_gemm_4x8__aarch64_neonfma_cortex_a55()  argument
491     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__aarch64_neonfma_cortex_a55, 4, 8, 1, 1,  in f32_gemm_4x8__aarch64_neonfma_cortex_a55()
494   static void f32_gemm_4x8__aarch64_neonfma_cortex_a75(benchmark::State& state, const char* net) {  in f32_gemm_4x8__aarch64_neonfma_cortex_a75()  argument
495     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__aarch64_neonfma_cortex_a75, 4, 8, 1, 1,  in f32_gemm_4x8__aarch64_neonfma_cortex_a75()
498 …static void f32_gemm_4x8__aarch64_neonfma_prfm_cortex_a75(benchmark::State& state, const char* net…  in f32_gemm_4x8__aarch64_neonfma_prfm_cortex_a75()  argument
499 … GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__aarch64_neonfma_prfm_cortex_a75, 4, 8, 1, 1,  in f32_gemm_4x8__aarch64_neonfma_prfm_cortex_a75()
502   static void f32_gemm_4x2__aarch64_neonfma_cortex_a75(benchmark::State& state, const char* net) {  in f32_gemm_4x2__aarch64_neonfma_cortex_a75()  argument
503     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x2__aarch64_neonfma_cortex_a75, 4, 2, 1, 1,  in f32_gemm_4x2__aarch64_neonfma_cortex_a75()
506 …static void f32_gemm_4x2__aarch64_neonfma_prfm_cortex_a75(benchmark::State& state, const char* net…  in f32_gemm_4x2__aarch64_neonfma_prfm_cortex_a75()  argument
507 … GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x2__aarch64_neonfma_prfm_cortex_a75, 4, 2, 1, 1,  in f32_gemm_4x2__aarch64_neonfma_prfm_cortex_a75()
510   static void f32_gemm_4x2__aarch64_neonfma_ld64(benchmark::State& state, const char* net) {  in f32_gemm_4x2__aarch64_neonfma_ld64()  argument
511     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x2__aarch64_neonfma_ld64, 4, 2, 1, 1,  in f32_gemm_4x2__aarch64_neonfma_ld64()
514   static void f32_gemm_4x8__aarch64_neonfma_ld64(benchmark::State& state, const char* net) {  in f32_gemm_4x8__aarch64_neonfma_ld64()  argument
515     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__aarch64_neonfma_ld64, 4, 8, 1, 1,  in f32_gemm_4x8__aarch64_neonfma_ld64()
518   static void f32_gemm_4x8__aarch64_neonfma_ld128(benchmark::State& state, const char* net) {  in f32_gemm_4x8__aarch64_neonfma_ld128()  argument
519     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__aarch64_neonfma_ld128, 4, 8, 1, 1,  in f32_gemm_4x8__aarch64_neonfma_ld128()
522   static void f32_gemm_5x8__aarch64_neonfma_cortex_a75(benchmark::State& state, const char* net) {  in f32_gemm_5x8__aarch64_neonfma_cortex_a75()  argument
523     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_5x8__aarch64_neonfma_cortex_a75, 5, 8, 1, 1,  in f32_gemm_5x8__aarch64_neonfma_cortex_a75()
526 …static void f32_gemm_5x8__aarch64_neonfma_prfm_cortex_a75(benchmark::State& state, const char* net…  in f32_gemm_5x8__aarch64_neonfma_prfm_cortex_a75()  argument
527 … GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_5x8__aarch64_neonfma_prfm_cortex_a75, 5, 8, 1, 1,  in f32_gemm_5x8__aarch64_neonfma_prfm_cortex_a75()
530   static void f32_gemm_6x8__aarch64_neonfma_ld64(benchmark::State& state, const char* net) {  in f32_gemm_6x8__aarch64_neonfma_ld64()  argument
531     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8__aarch64_neonfma_ld64, 6, 8, 1, 1,  in f32_gemm_6x8__aarch64_neonfma_ld64()
534   static void f32_gemm_6x8__aarch64_neonfma_ld128(benchmark::State& state, const char* net) {  in f32_gemm_6x8__aarch64_neonfma_ld128()  argument
535     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8__aarch64_neonfma_ld128, 6, 8, 1, 1,  in f32_gemm_6x8__aarch64_neonfma_ld128()
538   static void f32_gemm_6x8__aarch64_neonfma_cortex_a53(benchmark::State& state, const char* net) {  in f32_gemm_6x8__aarch64_neonfma_cortex_a53()  argument
539     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8__aarch64_neonfma_cortex_a53, 6, 8, 1, 1,  in f32_gemm_6x8__aarch64_neonfma_cortex_a53()
542 …static void f32_gemm_6x8__aarch64_neonfma_prfm_cortex_a53(benchmark::State& state, const char* net…  in f32_gemm_6x8__aarch64_neonfma_prfm_cortex_a53()  argument
543 … GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8__aarch64_neonfma_prfm_cortex_a53, 6, 8, 1, 1,  in f32_gemm_6x8__aarch64_neonfma_prfm_cortex_a53()
546   static void f32_gemm_6x8__aarch64_neonfma_cortex_a55(benchmark::State& state, const char* net) {  in f32_gemm_6x8__aarch64_neonfma_cortex_a55()  argument
547     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8__aarch64_neonfma_cortex_a55, 6, 8, 1, 1,  in f32_gemm_6x8__aarch64_neonfma_cortex_a55()
550   static void f32_gemm_6x8__aarch64_neonfma_cortex_a73(benchmark::State& state, const char* net) {  in f32_gemm_6x8__aarch64_neonfma_cortex_a73()  argument
551     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8__aarch64_neonfma_cortex_a73, 6, 8, 1, 1,  in f32_gemm_6x8__aarch64_neonfma_cortex_a73()
554   static void f32_gemm_6x8__aarch64_neonfma_cortex_a75(benchmark::State& state, const char* net) {  in f32_gemm_6x8__aarch64_neonfma_cortex_a75()  argument
555     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8__aarch64_neonfma_cortex_a75, 6, 8, 1, 1,  in f32_gemm_6x8__aarch64_neonfma_cortex_a75()
558 …static void f32_gemm_6x8__aarch64_neonfma_prfm_cortex_a75(benchmark::State& state, const char* net…  in f32_gemm_6x8__aarch64_neonfma_prfm_cortex_a75()  argument
559 … GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8__aarch64_neonfma_prfm_cortex_a75, 6, 8, 1, 1,  in f32_gemm_6x8__aarch64_neonfma_prfm_cortex_a75()
562   static void f32_gemm_1x8__neonfma_lane_ld64(benchmark::State& state, const char* net) {  in f32_gemm_1x8__neonfma_lane_ld64()  argument
563     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_1x8__neonfma_lane_ld64, 1, 8, 1, 1,  in f32_gemm_1x8__neonfma_lane_ld64()
566   static void f32_gemm_4x2__neonfma_lane_ld64(benchmark::State& state, const char* net) {  in f32_gemm_4x2__neonfma_lane_ld64()  argument
567     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x2__neonfma_lane_ld64, 4, 2, 1, 1,  in f32_gemm_4x2__neonfma_lane_ld64()
570   static void f32_gemm_6x2__neonfma_lane_ld64(benchmark::State& state, const char* net) {  in f32_gemm_6x2__neonfma_lane_ld64()  argument
571     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x2__neonfma_lane_ld64, 6, 2, 1, 1,  in f32_gemm_6x2__neonfma_lane_ld64()
574   static void f32_gemm_4x8__neonfma_lane_ld64(benchmark::State& state, const char* net) {  in f32_gemm_4x8__neonfma_lane_ld64()  argument
575     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__neonfma_lane_ld64, 4, 8, 1, 1,  in f32_gemm_4x8__neonfma_lane_ld64()
578   static void f32_gemm_4x8__neonfma_lane_ld128(benchmark::State& state, const char* net) {  in f32_gemm_4x8__neonfma_lane_ld128()  argument
579     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__neonfma_lane_ld128, 4, 8, 1, 1,  in f32_gemm_4x8__neonfma_lane_ld128()
582   static void f32_gemm_5x8__neonfma_lane_ld64(benchmark::State& state, const char* net) {  in f32_gemm_5x8__neonfma_lane_ld64()  argument
583     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_5x8__neonfma_lane_ld64, 5, 8, 1, 1,  in f32_gemm_5x8__neonfma_lane_ld64()
586   static void f32_gemm_6x8__neonfma_lane_ld64(benchmark::State& state, const char* net) {  in f32_gemm_6x8__neonfma_lane_ld64()  argument
587     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8__neonfma_lane_ld64, 6, 8, 1, 1,  in f32_gemm_6x8__neonfma_lane_ld64()
590   static void f32_gemm_6x8__neonfma_lane_ld128(benchmark::State& state, const char* net) {  in f32_gemm_6x8__neonfma_lane_ld128()  argument
591     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8__neonfma_lane_ld128, 6, 8, 1, 1,  in f32_gemm_6x8__neonfma_lane_ld128()
633   static void f32_gemm_4x4__aarch32_vfp_ld64(benchmark::State& state, const char* net) {  in BENCHMARK_GEMM()
634     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x4__aarch32_vfp_ld64, 4, 4, 1, 1,  in BENCHMARK_GEMM()
638   static void f32_gemm_4x8__aarch32_neon_ld64(benchmark::State& state, const char* net) {  in f32_gemm_4x8__aarch32_neon_ld64()  argument
639     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__aarch32_neon_ld64, 4, 8, 1, 1,  in f32_gemm_4x8__aarch32_neon_ld64()
642   static void f32_gemm_4x8__aarch32_neon_cortex_a7(benchmark::State& state, const char* net) {  in f32_gemm_4x8__aarch32_neon_cortex_a7()  argument
643     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__aarch32_neon_cortex_a7, 4, 8, 1, 1,  in f32_gemm_4x8__aarch32_neon_cortex_a7()
646   static void f32_gemm_4x8__aarch32_neon_cortex_a53(benchmark::State& state, const char* net) {  in f32_gemm_4x8__aarch32_neon_cortex_a53()  argument
647     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__aarch32_neon_cortex_a53, 4, 8, 1, 1,  in f32_gemm_4x8__aarch32_neon_cortex_a53()
650   static void f32_gemm_4x8__aarch32_neon_prfm_cortex_a53(benchmark::State& state, const char* net) {  in f32_gemm_4x8__aarch32_neon_prfm_cortex_a53()  argument
651     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__aarch32_neon_prfm_cortex_a53, 4, 8, 1, 1,  in f32_gemm_4x8__aarch32_neon_prfm_cortex_a53()
654   static void f32_gemm_4x8__aarch32_neon_cortex_a55(benchmark::State& state, const char* net) {  in f32_gemm_4x8__aarch32_neon_cortex_a55()  argument
655     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__aarch32_neon_cortex_a55, 4, 8, 1, 1,  in f32_gemm_4x8__aarch32_neon_cortex_a55()
658   static void f32_gemm_4x8__aarch32_neon_cortex_a75(benchmark::State& state, const char* net) {  in f32_gemm_4x8__aarch32_neon_cortex_a75()  argument
659     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__aarch32_neon_cortex_a75, 4, 8, 1, 1,  in f32_gemm_4x8__aarch32_neon_cortex_a75()
662   static void f32_gemm_4x8__aarch32_neon_prfm_cortex_a75(benchmark::State& state, const char* net) {  in f32_gemm_4x8__aarch32_neon_prfm_cortex_a75()  argument
663     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__aarch32_neon_prfm_cortex_a75, 4, 8, 1, 1,  in f32_gemm_4x8__aarch32_neon_prfm_cortex_a75()
678   static void f32_gemm_1x8__neon_lane_ld64(benchmark::State& state, const char* net) {  in BENCHMARK_GEMM()
679     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_1x8__neon_lane_ld64, 1, 8, 1, 1,  in BENCHMARK_GEMM()
682   static void f32_gemm_4x2__neon_lane_ld64(benchmark::State& state, const char* net) {  in f32_gemm_4x2__neon_lane_ld64()  argument
683     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x2__neon_lane_ld64, 4, 2, 1, 1,  in f32_gemm_4x2__neon_lane_ld64()
686   static void f32_gemm_6x2__neon_lane_ld64(benchmark::State& state, const char* net) {  in f32_gemm_6x2__neon_lane_ld64()  argument
687     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x2__neon_lane_ld64, 6, 2, 1, 1,  in f32_gemm_6x2__neon_lane_ld64()
690   static void f32_gemm_4x8__neon_lane_ld64(benchmark::State& state, const char* net) {  in f32_gemm_4x8__neon_lane_ld64()  argument
691     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__neon_lane_ld64, 4, 8, 1, 1,  in f32_gemm_4x8__neon_lane_ld64()
694   static void f32_gemm_4x8__neon_lane_ld128(benchmark::State& state, const char* net) {  in f32_gemm_4x8__neon_lane_ld128()  argument
695     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__neon_lane_ld128, 4, 8, 1, 1,  in f32_gemm_4x8__neon_lane_ld128()
698   static void f32_gemm_5x8__neon_lane_ld64(benchmark::State& state, const char* net) {  in f32_gemm_5x8__neon_lane_ld64()  argument
699     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_5x8__neon_lane_ld64, 5, 8, 1, 1,  in f32_gemm_5x8__neon_lane_ld64()
702   static void f32_gemm_6x8__neon_lane_ld64(benchmark::State& state, const char* net) {  in f32_gemm_6x8__neon_lane_ld64()  argument
703     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8__neon_lane_ld64, 6, 8, 1, 1,  in f32_gemm_6x8__neon_lane_ld64()
706   static void f32_gemm_6x8__neon_lane_ld128(benchmark::State& state, const char* net) {  in f32_gemm_6x8__neon_lane_ld128()  argument
707     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8__neon_lane_ld128, 6, 8, 1, 1,  in f32_gemm_6x8__neon_lane_ld128()
710   static void f32_gemm_1x8__neonfma_dup_ld64(benchmark::State& state, const char* net) {  in f32_gemm_1x8__neonfma_dup_ld64()  argument
711     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_1x8__neonfma_dup_ld64, 1, 8, 1, 1,  in f32_gemm_1x8__neonfma_dup_ld64()
714   static void f32_gemm_4x8__neonfma_dup_ld64(benchmark::State& state, const char* net) {  in f32_gemm_4x8__neonfma_dup_ld64()  argument
715     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__neonfma_dup_ld64, 4, 8, 1, 1,  in f32_gemm_4x8__neonfma_dup_ld64()
718   static void f32_gemm_4x8__neonfma_dup_ld128(benchmark::State& state, const char* net) {  in f32_gemm_4x8__neonfma_dup_ld128()  argument
719     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__neonfma_dup_ld128, 4, 8, 1, 1,  in f32_gemm_4x8__neonfma_dup_ld128()
722   static void f32_gemm_6x8__neonfma_dup_ld64(benchmark::State& state, const char* net) {  in f32_gemm_6x8__neonfma_dup_ld64()  argument
723     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8__neonfma_dup_ld64, 6, 8, 1, 1,  in f32_gemm_6x8__neonfma_dup_ld64()
726   static void f32_gemm_6x8__neonfma_dup_ld128(benchmark::State& state, const char* net) {  in f32_gemm_6x8__neonfma_dup_ld128()  argument
727     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8__neonfma_dup_ld128, 6, 8, 1, 1,  in f32_gemm_6x8__neonfma_dup_ld128()
730   static void f32_gemm_1x8s4__neon(benchmark::State& state, const char* net) {  in f32_gemm_1x8s4__neon()  argument
731     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_1x8s4__neon, 1, 8, 1, 4,  in f32_gemm_1x8s4__neon()
734   static void f32_gemm_1x8s4__neonfma(benchmark::State& state, const char* net) {  in f32_gemm_1x8s4__neonfma()  argument
735     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_1x8s4__neonfma, 1, 8, 1, 4,  in f32_gemm_1x8s4__neonfma()
738   static void f32_gemm_4x8s4__neon(benchmark::State& state, const char* net) {  in f32_gemm_4x8s4__neon()  argument
739     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8s4__neon, 4, 8, 1, 4,  in f32_gemm_4x8s4__neon()
742   static void f32_gemm_4x8s4__neonfma(benchmark::State& state, const char* net) {  in f32_gemm_4x8s4__neonfma()  argument
743     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8s4__neonfma, 4, 8, 1, 4,  in f32_gemm_4x8s4__neonfma()
746   static void f32_gemm_6x8s4__neon(benchmark::State& state, const char* net) {  in f32_gemm_6x8s4__neon()  argument
747     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8s4__neon, 6, 8, 1, 4,  in f32_gemm_6x8s4__neon()
750   static void f32_gemm_6x8s4__neonfma(benchmark::State& state, const char* net) {  in f32_gemm_6x8s4__neonfma()  argument
751     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8s4__neonfma, 6, 8, 1, 4,  in f32_gemm_6x8s4__neonfma()
754   static void f32_gemm_8x8s4__neon(benchmark::State& state, const char* net) {  in f32_gemm_8x8s4__neon()  argument
755     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_8x8s4__neon, 8, 8, 1, 4,  in f32_gemm_8x8s4__neon()
758   static void f32_gemm_8x8s4__neonfma(benchmark::State& state, const char* net) {  in f32_gemm_8x8s4__neonfma()  argument
759     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_8x8s4__neonfma, 8, 8, 1, 4,  in f32_gemm_8x8s4__neonfma()
762   static void f32_ppmm_4x8_unipass__neonfma(benchmark::State& state, const char* net) {  in f32_ppmm_4x8_unipass__neonfma()  argument
763 …PPMM1PBenchmark(state, xnn_f32_ppmm_minmax_ukernel_4x8__neonfma, xnn_x32_packx_ukernel_4x__neon_st…  in f32_ppmm_4x8_unipass__neonfma()
766   static void f32_ppmm_4x8_twopass__neonfma(benchmark::State& state, const char* net) {  in f32_ppmm_4x8_twopass__neonfma()  argument
767 …PPMM2PBenchmark(state, xnn_f32_ppmm_minmax_ukernel_4x8__neonfma, xnn_x32_packx_ukernel_4x__neon_st…  in f32_ppmm_4x8_twopass__neonfma()
802   static void jit_f32_gemm_4x8__aarch32_neon_ld64(benchmark::State& state, const char* net)  in BENCHMARK_GEMM()
804     GEMMBenchmark(state, xnn_generate_f32_gemm_ukernel_4x8__aarch32_neon_ld64, 4, 8, 1, 1,  in BENCHMARK_GEMM()
807   static void jit_f32_gemm_4x8__aarch32_neon_cortex_a7(benchmark::State& state, const char* net)  in jit_f32_gemm_4x8__aarch32_neon_cortex_a7()  argument
809     GEMMBenchmark(state, xnn_generate_f32_gemm_ukernel_4x8__aarch32_neon_cortex_a7, 4, 8, 1, 1,  in jit_f32_gemm_4x8__aarch32_neon_cortex_a7()
812   static void jit_f32_gemm_4x8__aarch32_neon_cortex_a53(benchmark::State& state, const char* net)  in jit_f32_gemm_4x8__aarch32_neon_cortex_a53()  argument
814     GEMMBenchmark(state, xnn_generate_f32_gemm_ukernel_4x8__aarch32_neon_cortex_a53, 4, 8, 1, 1,  in jit_f32_gemm_4x8__aarch32_neon_cortex_a53()
817   static void jit_f32_gemm_4x8__aarch32_neon_cortex_a55(benchmark::State& state, const char* net)  in jit_f32_gemm_4x8__aarch32_neon_cortex_a55()  argument
819     GEMMBenchmark(state, xnn_generate_f32_gemm_ukernel_4x8__aarch32_neon_cortex_a55, 4, 8, 1, 1,  in jit_f32_gemm_4x8__aarch32_neon_cortex_a55()
822   static void jit_f32_gemm_4x8__aarch32_neon_cortex_a75(benchmark::State& state, const char* net)  in jit_f32_gemm_4x8__aarch32_neon_cortex_a75()  argument
824     GEMMBenchmark(state, xnn_generate_f32_gemm_ukernel_4x8__aarch32_neon_cortex_a75, 4, 8, 1, 1,  in jit_f32_gemm_4x8__aarch32_neon_cortex_a75()
827 …static void jit_f32_gemm_4x8__aarch32_neon_prfm_cortex_a75(benchmark::State& state, const char* ne…  in jit_f32_gemm_4x8__aarch32_neon_prfm_cortex_a75()  argument
829 …  GEMMBenchmark(state, xnn_generate_f32_gemm_ukernel_4x8__aarch32_neon_prfm_cortex_a75, 4, 8, 1, 1,  in jit_f32_gemm_4x8__aarch32_neon_prfm_cortex_a75()
842   static void jit_f32_gemm_1x8__aarch64_neonfma_cortex_a75(benchmark::State& state, const char* net)  in BENCHMARK_GEMM()
844     GEMMBenchmark(state, xnn_generate_f32_gemm_ukernel_1x8__aarch64_neonfma_cortex_a75, 1, 8, 1, 1,  in BENCHMARK_GEMM()
847 …static void jit_f32_gemm_1x8__aarch64_neonfma_prfm_cortex_a75(benchmark::State& state, const char*…  in jit_f32_gemm_1x8__aarch64_neonfma_prfm_cortex_a75()  argument
849 …GEMMBenchmark(state, xnn_generate_f32_gemm_ukernel_1x8__aarch64_neonfma_prfm_cortex_a75, 1, 8, 1, …  in jit_f32_gemm_1x8__aarch64_neonfma_prfm_cortex_a75()
852   static void jit_f32_gemm_4x8__aarch64_neonfma_cortex_a75(benchmark::State& state, const char* net)  in jit_f32_gemm_4x8__aarch64_neonfma_cortex_a75()  argument
854     GEMMBenchmark(state, xnn_generate_f32_gemm_ukernel_4x8__aarch64_neonfma_cortex_a75, 4, 8, 1, 1,  in jit_f32_gemm_4x8__aarch64_neonfma_cortex_a75()
857 …static void jit_f32_gemm_4x8__aarch64_neonfma_prfm_cortex_a75(benchmark::State& state, const char*…  in jit_f32_gemm_4x8__aarch64_neonfma_prfm_cortex_a75()  argument
859 …GEMMBenchmark(state, xnn_generate_f32_gemm_ukernel_4x8__aarch64_neonfma_prfm_cortex_a75, 4, 8, 1, …  in jit_f32_gemm_4x8__aarch64_neonfma_prfm_cortex_a75()
862   static void jit_f32_gemm_6x8__aarch64_neonfma_ld128(benchmark::State& state, const char* net)  in jit_f32_gemm_6x8__aarch64_neonfma_ld128()  argument
864     GEMMBenchmark(state, xnn_generate_f32_gemm_ukernel_6x8__aarch64_neonfma_ld128, 6, 8, 1, 1,  in jit_f32_gemm_6x8__aarch64_neonfma_ld128()
874   static void name(benchmark::State &state, const char *net) {                  \
876         state,                                                                  \
893   static void f32_gemm_1x16__avx512f_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_1x16__avx512f_broadcast()  argument
894     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_1x16__avx512f_broadcast, 1, 16, 1, 1,  in f32_gemm_1x16__avx512f_broadcast()
897   static void f32_gemm_4x16__avx512f_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_4x16__avx512f_broadcast()  argument
898     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x16__avx512f_broadcast, 4, 16, 1, 1,  in f32_gemm_4x16__avx512f_broadcast()
901   static void f32_gemm_5x16__avx512f_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_5x16__avx512f_broadcast()  argument
902     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_5x16__avx512f_broadcast, 5, 16, 1, 1,  in f32_gemm_5x16__avx512f_broadcast()
905   static void f32_gemm_6x16__avx512f_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_6x16__avx512f_broadcast()  argument
906     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x16__avx512f_broadcast, 6, 16, 1, 1,  in f32_gemm_6x16__avx512f_broadcast()
909   static void f32_gemm_7x16__avx512f_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_7x16__avx512f_broadcast()  argument
910     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_7x16__avx512f_broadcast, 7, 16, 1, 1,  in f32_gemm_7x16__avx512f_broadcast()
913   static void f32_gemm_8x16__avx512f_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_8x16__avx512f_broadcast()  argument
914     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_8x16__avx512f_broadcast, 8, 16, 1, 1,  in f32_gemm_8x16__avx512f_broadcast()
918   static void f32_gemm_1x8__fma3_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_1x8__fma3_broadcast()  argument
919     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_1x8__fma3_broadcast, 1, 8, 1, 1,  in f32_gemm_1x8__fma3_broadcast()
922   static void f32_gemm_4x8__fma3_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_4x8__fma3_broadcast()  argument
923     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__fma3_broadcast, 4, 8, 1, 1,  in f32_gemm_4x8__fma3_broadcast()
926   static void f32_gemm_5x8__fma3_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_5x8__fma3_broadcast()  argument
927     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_5x8__fma3_broadcast, 5, 8, 1, 1,  in f32_gemm_5x8__fma3_broadcast()
930   static void f32_gemm_6x8__fma3_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_6x8__fma3_broadcast()  argument
931     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8__fma3_broadcast, 6, 8, 1, 1,  in f32_gemm_6x8__fma3_broadcast()
934   static void f32_gemm_7x8__fma3_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_7x8__fma3_broadcast()  argument
935     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_7x8__fma3_broadcast, 7, 8, 1, 1,  in f32_gemm_7x8__fma3_broadcast()
938   static void f32_gemm_8x8__fma3_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_8x8__fma3_broadcast()  argument
939     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_8x8__fma3_broadcast, 8, 8, 1, 1,  in f32_gemm_8x8__fma3_broadcast()
942   static void f32_gemm_1x16__fma3_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_1x16__fma3_broadcast()  argument
943     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_1x16__fma3_broadcast, 1, 16, 1, 1,  in f32_gemm_1x16__fma3_broadcast()
946   static void f32_gemm_3x16__fma3_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_3x16__fma3_broadcast()  argument
947     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_3x16__fma3_broadcast, 4, 16, 1, 1,  in f32_gemm_3x16__fma3_broadcast()
950   static void f32_gemm_4x16__fma3_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_4x16__fma3_broadcast()  argument
951     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x16__fma3_broadcast, 4, 16, 1, 1,  in f32_gemm_4x16__fma3_broadcast()
954   static void f32_gemm_5x16__fma3_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_5x16__fma3_broadcast()  argument
955     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_5x16__fma3_broadcast, 5, 16, 1, 1,  in f32_gemm_5x16__fma3_broadcast()
959   static void f32_gemm_1x16s4__fma3_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_1x16s4__fma3_broadcast()  argument
960     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_1x16s4__fma3_broadcast, 1, 16, 1, 4,  in f32_gemm_1x16s4__fma3_broadcast()
963   static void f32_gemm_3x16s4__fma3_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_3x16s4__fma3_broadcast()  argument
964     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_3x16s4__fma3_broadcast, 4, 16, 1, 4,  in f32_gemm_3x16s4__fma3_broadcast()
967   static void f32_gemm_4x16s4__fma3_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_4x16s4__fma3_broadcast()  argument
968     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x16s4__fma3_broadcast, 4, 16, 1, 4,  in f32_gemm_4x16s4__fma3_broadcast()
971   static void f32_gemm_5x16s4__fma3_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_5x16s4__fma3_broadcast()  argument
972     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_5x16s4__fma3_broadcast, 5, 16, 1, 4,  in f32_gemm_5x16s4__fma3_broadcast()
976   static void f32_gemm_1x8__avx_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_1x8__avx_broadcast()  argument
977     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_1x8__avx_broadcast, 1, 8, 1, 1,  in f32_gemm_1x8__avx_broadcast()
980   static void f32_gemm_4x8__avx_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_4x8__avx_broadcast()  argument
981     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__avx_broadcast, 4, 8, 1, 1,  in f32_gemm_4x8__avx_broadcast()
984   static void f32_gemm_5x8__avx_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_5x8__avx_broadcast()  argument
985     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_5x8__avx_broadcast, 5, 8, 1, 1,  in f32_gemm_5x8__avx_broadcast()
988   static void f32_gemm_6x8__avx_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_6x8__avx_broadcast()  argument
989     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8__avx_broadcast, 6, 8, 1, 1,  in f32_gemm_6x8__avx_broadcast()
992   static void f32_gemm_7x8__avx_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_7x8__avx_broadcast()  argument
993     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_7x8__avx_broadcast, 7, 8, 1, 1,  in f32_gemm_7x8__avx_broadcast()
996   static void f32_gemm_1x16__avx_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_1x16__avx_broadcast()  argument
997     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_1x16__avx_broadcast, 1, 16, 1, 1,  in f32_gemm_1x16__avx_broadcast()
1000   static void f32_gemm_3x16__avx_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_3x16__avx_broadcast()  argument
1001     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_3x16__avx_broadcast, 4, 16, 1, 1,  in f32_gemm_3x16__avx_broadcast()
1004   static void f32_gemm_4x16__avx_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_4x16__avx_broadcast()  argument
1005     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x16__avx_broadcast, 4, 16, 1, 1,  in f32_gemm_4x16__avx_broadcast()
1008   static void f32_gemm_5x16__avx_broadcast(benchmark::State& state, const char* net) {  in f32_gemm_5x16__avx_broadcast()  argument
1009     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_5x16__avx_broadcast, 5, 16, 1, 1,  in f32_gemm_5x16__avx_broadcast()
1013   static void f32_gemm_1x8__sse2_dup(benchmark::State& state, const char* net) {  in f32_gemm_1x8__sse2_dup()  argument
1014     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_1x8__sse2_dup, 1, 8, 1, 1,  in f32_gemm_1x8__sse2_dup()
1017   static void f32_gemm_3x8__sse2_dup(benchmark::State& state, const char* net) {  in f32_gemm_3x8__sse2_dup()  argument
1018     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_3x8__sse2_dup, 3, 8, 1, 1,  in f32_gemm_3x8__sse2_dup()
1021   static void f32_gemm_4x8__sse2_dup(benchmark::State& state, const char* net) {  in f32_gemm_4x8__sse2_dup()  argument
1022     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__sse2_dup, 4, 8, 1, 1,  in f32_gemm_4x8__sse2_dup()
1025   static void f32_gemm_5x8__sse2_dup(benchmark::State& state, const char* net) {  in f32_gemm_5x8__sse2_dup()  argument
1026     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_5x8__sse2_dup, 5, 8, 1, 1,  in f32_gemm_5x8__sse2_dup()
1030   static void f32_gemm_1x8__sse_load1(benchmark::State& state, const char* net) {  in f32_gemm_1x8__sse_load1()  argument
1031     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_1x8__sse_load1, 1, 8, 1, 1,  in f32_gemm_1x8__sse_load1()
1034   static void f32_gemm_3x8__sse_load1(benchmark::State& state, const char* net) {  in f32_gemm_3x8__sse_load1()  argument
1035     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_3x8__sse_load1, 3, 8, 1, 1,  in f32_gemm_3x8__sse_load1()
1038   static void f32_gemm_4x8__sse_load1(benchmark::State& state, const char* net) {  in f32_gemm_4x8__sse_load1()  argument
1039     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__sse_load1, 4, 8, 1, 1,  in f32_gemm_4x8__sse_load1()
1042   static void f32_gemm_5x8__sse_load1(benchmark::State& state, const char* net) {  in f32_gemm_5x8__sse_load1()  argument
1043     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_5x8__sse_load1, 5, 8, 1, 1,  in f32_gemm_5x8__sse_load1()
1047   static void f32_gemm_1x8__sse_dup(benchmark::State& state, const char* net) {  in f32_gemm_1x8__sse_dup()  argument
1048     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_1x8__sse_dup, 1, 8, 1, 1,  in f32_gemm_1x8__sse_dup()
1051   static void f32_gemm_3x8__sse_dup(benchmark::State& state, const char* net) {  in f32_gemm_3x8__sse_dup()  argument
1052     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_3x8__sse_dup, 3, 8, 1, 1,  in f32_gemm_3x8__sse_dup()
1055   static void f32_gemm_4x8__sse_dup(benchmark::State& state, const char* net) {  in f32_gemm_4x8__sse_dup()  argument
1056     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__sse_dup, 4, 8, 1, 1,  in f32_gemm_4x8__sse_dup()
1059   static void f32_gemm_5x8__sse_dup(benchmark::State& state, const char* net) {  in f32_gemm_5x8__sse_dup()  argument
1060     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_5x8__sse_dup, 5, 8, 1, 1,  in f32_gemm_5x8__sse_dup()
1064   static void f32_gemm_1x8s4__sse(benchmark::State& state, const char* net) {  in f32_gemm_1x8s4__sse()  argument
1065     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_1x8s4__sse, 1, 8, 1, 4,  in f32_gemm_1x8s4__sse()
1068   static void f32_gemm_3x8s4__sse(benchmark::State& state, const char* net) {  in f32_gemm_3x8s4__sse()  argument
1069     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_3x8s4__sse, 3, 8, 1, 4,  in f32_gemm_3x8s4__sse()
1072   static void f32_gemm_4x8s4__sse(benchmark::State& state, const char* net) {  in f32_gemm_4x8s4__sse()  argument
1073     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8s4__sse, 4, 8, 1, 4,  in f32_gemm_4x8s4__sse()
1076   static void f32_gemm_5x8s4__sse(benchmark::State& state, const char* net) {  in f32_gemm_5x8s4__sse()  argument
1077     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_5x8s4__sse, 5, 8, 1, 4,  in f32_gemm_5x8s4__sse()
1081   static void f32_ppmm_4x8_unipass__sse(benchmark::State& state, const char* net) {  in f32_ppmm_4x8_unipass__sse()  argument
1082 …  PPMM1PBenchmark(state, xnn_f32_ppmm_minmax_ukernel_4x8__sse, xnn_x32_packx_ukernel_4x__sse, 4, 8,  in f32_ppmm_4x8_unipass__sse()
1085   static void f32_ppmm_4x8_twopass__sse(benchmark::State& state, const char* net) {  in f32_ppmm_4x8_twopass__sse()  argument
1086 …  PPMM2PBenchmark(state, xnn_f32_ppmm_minmax_ukernel_4x8__sse, xnn_x32_packx_ukernel_4x__sse, 4, 8,  in f32_ppmm_4x8_twopass__sse()
1149   static void f32_gemm_3x8__wasmrelaxedsimd_loadsplat(benchmark::State& state, const char* net) {  in BENCHMARK_GEMM()
1150     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_3x8__wasmrelaxedsimd_loadsplat, 3, 8, 1, 1,  in BENCHMARK_GEMM()
1153   static void f32_gemm_4x8__wasmrelaxedsimd_loadsplat(benchmark::State& state, const char* net) {  in f32_gemm_4x8__wasmrelaxedsimd_loadsplat()  argument
1154     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__wasmrelaxedsimd_loadsplat, 4, 8, 1, 1,  in f32_gemm_4x8__wasmrelaxedsimd_loadsplat()
1157   static void f32_gemm_5x8__wasmrelaxedsimd_loadsplat(benchmark::State& state, const char* net) {  in f32_gemm_5x8__wasmrelaxedsimd_loadsplat()  argument
1158     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_5x8__wasmrelaxedsimd_loadsplat, 5, 8, 1, 1,  in f32_gemm_5x8__wasmrelaxedsimd_loadsplat()
1161   static void f32_gemm_6x8__wasmrelaxedsimd_loadsplat(benchmark::State& state, const char* net) {  in f32_gemm_6x8__wasmrelaxedsimd_loadsplat()  argument
1162     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8__wasmrelaxedsimd_loadsplat, 6, 8, 1, 1,  in f32_gemm_6x8__wasmrelaxedsimd_loadsplat()
1165 …static void f32_gemm_3x8__wasmrelaxedsimd_fma_loadsplat(benchmark::State& state, const char* net) {  in f32_gemm_3x8__wasmrelaxedsimd_fma_loadsplat()  argument
1166     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_3x8__wasmrelaxedsimd_fma_loadsplat, 3, 8, 1, 1,  in f32_gemm_3x8__wasmrelaxedsimd_fma_loadsplat()
1169 …static void f32_gemm_4x8__wasmrelaxedsimd_fma_loadsplat(benchmark::State& state, const char* net) {  in f32_gemm_4x8__wasmrelaxedsimd_fma_loadsplat()  argument
1170     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__wasmrelaxedsimd_fma_loadsplat, 4, 8, 1, 1,  in f32_gemm_4x8__wasmrelaxedsimd_fma_loadsplat()
1173 …static void f32_gemm_5x8__wasmrelaxedsimd_fma_loadsplat(benchmark::State& state, const char* net) {  in f32_gemm_5x8__wasmrelaxedsimd_fma_loadsplat()  argument
1174     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_5x8__wasmrelaxedsimd_fma_loadsplat, 5, 8, 1, 1,  in f32_gemm_5x8__wasmrelaxedsimd_fma_loadsplat()
1177 …static void f32_gemm_6x8__wasmrelaxedsimd_fma_loadsplat(benchmark::State& state, const char* net) {  in f32_gemm_6x8__wasmrelaxedsimd_fma_loadsplat()  argument
1178     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8__wasmrelaxedsimd_fma_loadsplat, 6, 8, 1, 1,  in f32_gemm_6x8__wasmrelaxedsimd_fma_loadsplat()
1181   static void f32_gemm_3x8__wasmrelaxedsimd_splat(benchmark::State& state, const char* net) {  in f32_gemm_3x8__wasmrelaxedsimd_splat()  argument
1182     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_3x8__wasmrelaxedsimd_splat, 3, 8, 1, 1,  in f32_gemm_3x8__wasmrelaxedsimd_splat()
1185   static void f32_gemm_4x8__wasmrelaxedsimd_splat(benchmark::State& state, const char* net) {  in f32_gemm_4x8__wasmrelaxedsimd_splat()  argument
1186     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__wasmrelaxedsimd_splat, 4, 8, 1, 1,  in f32_gemm_4x8__wasmrelaxedsimd_splat()
1189   static void f32_gemm_5x8__wasmrelaxedsimd_splat(benchmark::State& state, const char* net) {  in f32_gemm_5x8__wasmrelaxedsimd_splat()  argument
1190     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_5x8__wasmrelaxedsimd_splat, 5, 8, 1, 1,  in f32_gemm_5x8__wasmrelaxedsimd_splat()
1193   static void f32_gemm_6x8__wasmrelaxedsimd_splat(benchmark::State& state, const char* net) {  in f32_gemm_6x8__wasmrelaxedsimd_splat()  argument
1194     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8__wasmrelaxedsimd_splat, 6, 8, 1, 1,  in f32_gemm_6x8__wasmrelaxedsimd_splat()
1197   static void f32_gemm_3x8__wasmrelaxedsimd_fma_splat(benchmark::State& state, const char* net) {  in f32_gemm_3x8__wasmrelaxedsimd_fma_splat()  argument
1198     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_3x8__wasmrelaxedsimd_fma_splat, 3, 8, 1, 1,  in f32_gemm_3x8__wasmrelaxedsimd_fma_splat()
1201   static void f32_gemm_4x8__wasmrelaxedsimd_fma_splat(benchmark::State& state, const char* net) {  in f32_gemm_4x8__wasmrelaxedsimd_fma_splat()  argument
1202     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__wasmrelaxedsimd_fma_splat, 4, 8, 1, 1,  in f32_gemm_4x8__wasmrelaxedsimd_fma_splat()
1205   static void f32_gemm_5x8__wasmrelaxedsimd_fma_splat(benchmark::State& state, const char* net) {  in f32_gemm_5x8__wasmrelaxedsimd_fma_splat()  argument
1206     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_5x8__wasmrelaxedsimd_fma_splat, 5, 8, 1, 1,  in f32_gemm_5x8__wasmrelaxedsimd_fma_splat()
1209   static void f32_gemm_6x8__wasmrelaxedsimd_fma_splat(benchmark::State& state, const char* net) {  in f32_gemm_6x8__wasmrelaxedsimd_fma_splat()  argument
1210     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8__wasmrelaxedsimd_fma_splat, 6, 8, 1, 1,  in f32_gemm_6x8__wasmrelaxedsimd_fma_splat()
1213   static void f32_gemm_3x8s4__wasmrelaxedsimd(benchmark::State& state, const char* net) {  in f32_gemm_3x8s4__wasmrelaxedsimd()  argument
1214     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_3x8s4__wasmrelaxedsimd, 3, 8, 1, 4,  in f32_gemm_3x8s4__wasmrelaxedsimd()
1217   static void f32_gemm_4x8s4__wasmrelaxedsimd(benchmark::State& state, const char* net) {  in f32_gemm_4x8s4__wasmrelaxedsimd()  argument
1218     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8s4__wasmrelaxedsimd, 4, 8, 1, 4,  in f32_gemm_4x8s4__wasmrelaxedsimd()
1221   static void f32_gemm_5x8s4__wasmrelaxedsimd(benchmark::State& state, const char* net) {  in f32_gemm_5x8s4__wasmrelaxedsimd()  argument
1222     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_5x8s4__wasmrelaxedsimd, 5, 8, 1, 4,  in f32_gemm_5x8s4__wasmrelaxedsimd()
1225   static void f32_gemm_6x8s4__wasmrelaxedsimd(benchmark::State& state, const char* net) {  in f32_gemm_6x8s4__wasmrelaxedsimd()  argument
1226     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8s4__wasmrelaxedsimd, 6, 8, 1, 4,  in f32_gemm_6x8s4__wasmrelaxedsimd()
1229   static void f32_gemm_3x8s4__wasmrelaxedsimd_fma(benchmark::State& state, const char* net) {  in f32_gemm_3x8s4__wasmrelaxedsimd_fma()  argument
1230     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_3x8s4__wasmrelaxedsimd_fma, 3, 8, 1, 4,  in f32_gemm_3x8s4__wasmrelaxedsimd_fma()
1233   static void f32_gemm_4x8s4__wasmrelaxedsimd_fma(benchmark::State& state, const char* net) {  in f32_gemm_4x8s4__wasmrelaxedsimd_fma()  argument
1234     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8s4__wasmrelaxedsimd_fma, 4, 8, 1, 4,  in f32_gemm_4x8s4__wasmrelaxedsimd_fma()
1237   static void f32_gemm_5x8s4__wasmrelaxedsimd_fma(benchmark::State& state, const char* net) {  in f32_gemm_5x8s4__wasmrelaxedsimd_fma()  argument
1238     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_5x8s4__wasmrelaxedsimd_fma, 5, 8, 1, 4,  in f32_gemm_5x8s4__wasmrelaxedsimd_fma()
1241   static void f32_gemm_6x8s4__wasmrelaxedsimd_fma(benchmark::State& state, const char* net) {  in f32_gemm_6x8s4__wasmrelaxedsimd_fma()  argument
1242     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8s4__wasmrelaxedsimd_fma, 6, 8, 1, 4,  in f32_gemm_6x8s4__wasmrelaxedsimd_fma()
1279   static void f32_gemm_3x8__wasmsimd_arm_loadsplat(benchmark::State& state, const char* net) {  in BENCHMARK_GEMM()
1280     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_arm_loadsplat, 3, 8, 1, 1,  in BENCHMARK_GEMM()
1283   static void f32_gemm_4x8__wasmsimd_arm_loadsplat(benchmark::State& state, const char* net) {  in f32_gemm_4x8__wasmsimd_arm_loadsplat()  argument
1284     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_arm_loadsplat, 4, 8, 1, 1,  in f32_gemm_4x8__wasmsimd_arm_loadsplat()
1287   static void f32_gemm_5x8__wasmsimd_arm_loadsplat(benchmark::State& state, const char* net) {  in f32_gemm_5x8__wasmsimd_arm_loadsplat()  argument
1288     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_5x8__wasmsimd_arm_loadsplat, 5, 8, 1, 1,  in f32_gemm_5x8__wasmsimd_arm_loadsplat()
1291   static void f32_gemm_6x8__wasmsimd_arm_loadsplat(benchmark::State& state, const char* net) {  in f32_gemm_6x8__wasmsimd_arm_loadsplat()  argument
1292     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8__wasmsimd_arm_loadsplat, 6, 8, 1, 1,  in f32_gemm_6x8__wasmsimd_arm_loadsplat()
1295   static void f32_gemm_3x8__wasmsimd_x86_loadsplat(benchmark::State& state, const char* net) {  in f32_gemm_3x8__wasmsimd_x86_loadsplat()  argument
1296     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_x86_loadsplat, 3, 8, 1, 1,  in f32_gemm_3x8__wasmsimd_x86_loadsplat()
1299   static void f32_gemm_4x8__wasmsimd_x86_loadsplat(benchmark::State& state, const char* net) {  in f32_gemm_4x8__wasmsimd_x86_loadsplat()  argument
1300     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_x86_loadsplat, 4, 8, 1, 1,  in f32_gemm_4x8__wasmsimd_x86_loadsplat()
1303   static void f32_gemm_5x8__wasmsimd_x86_loadsplat(benchmark::State& state, const char* net) {  in f32_gemm_5x8__wasmsimd_x86_loadsplat()  argument
1304     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_5x8__wasmsimd_x86_loadsplat, 5, 8, 1, 1,  in f32_gemm_5x8__wasmsimd_x86_loadsplat()
1307   static void f32_gemm_6x8__wasmsimd_x86_loadsplat(benchmark::State& state, const char* net) {  in f32_gemm_6x8__wasmsimd_x86_loadsplat()  argument
1308     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8__wasmsimd_x86_loadsplat, 6, 8, 1, 1,  in f32_gemm_6x8__wasmsimd_x86_loadsplat()
1311   static void f32_gemm_3x8__wasmsimd_arm_splat(benchmark::State& state, const char* net) {  in f32_gemm_3x8__wasmsimd_arm_splat()  argument
1312     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_arm_splat, 3, 8, 1, 1,  in f32_gemm_3x8__wasmsimd_arm_splat()
1315   static void f32_gemm_4x8__wasmsimd_arm_splat(benchmark::State& state, const char* net) {  in f32_gemm_4x8__wasmsimd_arm_splat()  argument
1316     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_arm_splat, 4, 8, 1, 1,  in f32_gemm_4x8__wasmsimd_arm_splat()
1319   static void f32_gemm_5x8__wasmsimd_arm_splat(benchmark::State& state, const char* net) {  in f32_gemm_5x8__wasmsimd_arm_splat()  argument
1320     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_5x8__wasmsimd_arm_splat, 5, 8, 1, 1,  in f32_gemm_5x8__wasmsimd_arm_splat()
1323   static void f32_gemm_6x8__wasmsimd_arm_splat(benchmark::State& state, const char* net) {  in f32_gemm_6x8__wasmsimd_arm_splat()  argument
1324     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8__wasmsimd_arm_splat, 6, 8, 1, 1,  in f32_gemm_6x8__wasmsimd_arm_splat()
1327   static void f32_gemm_3x8__wasmsimd_x86_splat(benchmark::State& state, const char* net) {  in f32_gemm_3x8__wasmsimd_x86_splat()  argument
1328     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_x86_splat, 3, 8, 1, 1,  in f32_gemm_3x8__wasmsimd_x86_splat()
1331   static void f32_gemm_4x8__wasmsimd_x86_splat(benchmark::State& state, const char* net) {  in f32_gemm_4x8__wasmsimd_x86_splat()  argument
1332     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_x86_splat, 4, 8, 1, 1,  in f32_gemm_4x8__wasmsimd_x86_splat()
1335   static void f32_gemm_5x8__wasmsimd_x86_splat(benchmark::State& state, const char* net) {  in f32_gemm_5x8__wasmsimd_x86_splat()  argument
1336     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_5x8__wasmsimd_x86_splat, 5, 8, 1, 1,  in f32_gemm_5x8__wasmsimd_x86_splat()
1339   static void f32_gemm_6x8__wasmsimd_x86_splat(benchmark::State& state, const char* net) {  in f32_gemm_6x8__wasmsimd_x86_splat()  argument
1340     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8__wasmsimd_x86_splat, 6, 8, 1, 1,  in f32_gemm_6x8__wasmsimd_x86_splat()
1343   static void f32_gemm_3x8s4__wasmsimd_arm(benchmark::State& state, const char* net) {  in f32_gemm_3x8s4__wasmsimd_arm()  argument
1344     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_3x8s4__wasmsimd_arm, 3, 8, 1, 4,  in f32_gemm_3x8s4__wasmsimd_arm()
1347   static void f32_gemm_4x8s4__wasmsimd_arm(benchmark::State& state, const char* net) {  in f32_gemm_4x8s4__wasmsimd_arm()  argument
1348     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8s4__wasmsimd_arm, 4, 8, 1, 4,  in f32_gemm_4x8s4__wasmsimd_arm()
1351   static void f32_gemm_5x8s4__wasmsimd_arm(benchmark::State& state, const char* net) {  in f32_gemm_5x8s4__wasmsimd_arm()  argument
1352     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_5x8s4__wasmsimd_arm, 5, 8, 1, 4,  in f32_gemm_5x8s4__wasmsimd_arm()
1355   static void f32_gemm_6x8s4__wasmsimd_arm(benchmark::State& state, const char* net) {  in f32_gemm_6x8s4__wasmsimd_arm()  argument
1356     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8s4__wasmsimd_arm, 6, 8, 1, 4,  in f32_gemm_6x8s4__wasmsimd_arm()
1359   static void f32_gemm_3x8s4__wasmsimd_x86(benchmark::State& state, const char* net) {  in f32_gemm_3x8s4__wasmsimd_x86()  argument
1360     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_3x8s4__wasmsimd_x86, 3, 8, 1, 4,  in f32_gemm_3x8s4__wasmsimd_x86()
1363   static void f32_gemm_4x8s4__wasmsimd_x86(benchmark::State& state, const char* net) {  in f32_gemm_4x8s4__wasmsimd_x86()  argument
1364     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x8s4__wasmsimd_x86, 4, 8, 1, 4,  in f32_gemm_4x8s4__wasmsimd_x86()
1367   static void f32_gemm_5x8s4__wasmsimd_x86(benchmark::State& state, const char* net) {  in f32_gemm_5x8s4__wasmsimd_x86()  argument
1368     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_5x8s4__wasmsimd_x86, 5, 8, 1, 4,  in f32_gemm_5x8s4__wasmsimd_x86()
1371   static void f32_gemm_6x8s4__wasmsimd_x86(benchmark::State& state, const char* net) {  in f32_gemm_6x8s4__wasmsimd_x86()  argument
1372     GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_6x8s4__wasmsimd_x86, 6, 8, 1, 4,  in f32_gemm_6x8s4__wasmsimd_x86()
1376   static void f32_ppmm_4x8_unipass__wasmsimd_arm_splat(benchmark::State& state, const char* net) {  in f32_ppmm_4x8_unipass__wasmsimd_arm_splat()  argument
1377 …PPMM1PBenchmark(state, xnn_f32_ppmm_minmax_ukernel_4x8__wasmsimd_arm_splat, xnn_x32_packx_ukernel_…  in f32_ppmm_4x8_unipass__wasmsimd_arm_splat()
1380   static void f32_ppmm_4x8_unipass__wasmsimd_x86_splat(benchmark::State& state, const char* net) {  in f32_ppmm_4x8_unipass__wasmsimd_x86_splat()  argument
1381 …PPMM1PBenchmark(state, xnn_f32_ppmm_minmax_ukernel_4x8__wasmsimd_x86_splat, xnn_x32_packx_ukernel_…  in f32_ppmm_4x8_unipass__wasmsimd_x86_splat()
1385   static void f32_ppmm_4x8_twopass__wasmsimd_arm_splat(benchmark::State& state, const char* net) {  in f32_ppmm_4x8_twopass__wasmsimd_arm_splat()  argument
1386 …PPMM2PBenchmark(state, xnn_f32_ppmm_minmax_ukernel_4x8__wasmsimd_arm_splat, xnn_x32_packx_ukernel_…  in f32_ppmm_4x8_twopass__wasmsimd_arm_splat()
1389   static void f32_ppmm_4x8_twopass__wasmsimd_x86_splat(benchmark::State& state, const char* net) {  in f32_ppmm_4x8_twopass__wasmsimd_x86_splat()  argument
1390 …PPMM2PBenchmark(state, xnn_f32_ppmm_minmax_ukernel_4x8__wasmsimd_x86_splat, xnn_x32_packx_ukernel_…  in f32_ppmm_4x8_twopass__wasmsimd_x86_splat()
1432 static void f32_gemm_1x4__scalar(benchmark::State& state, const char* net) {  in BENCHMARK_GEMM()
1433   GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_1x4__scalar, 1, 4, 1, 1,  in BENCHMARK_GEMM()
1436 static void f32_gemm_2x4__scalar(benchmark::State& state, const char* net) {  in f32_gemm_2x4__scalar()  argument
1437   GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_2x4__scalar, 2, 4, 1, 1,  in f32_gemm_2x4__scalar()
1440 static void f32_gemm_4x4__scalar(benchmark::State& state, const char* net) {  in f32_gemm_4x4__scalar()  argument
1441   GEMMBenchmark(state, xnn_f32_gemm_minmax_ukernel_4x4__scalar, 4, 4, 1, 1,  in f32_gemm_4x4__scalar()
1445 static void f32_ppmm_2x4_unipass__scalar(benchmark::State& state, const char* net) {  in f32_ppmm_2x4_unipass__scalar()  argument
1446 …PPMM1PBenchmark(state, xnn_f32_ppmm_minmax_ukernel_2x4__scalar, xnn_x32_packx_ukernel_2x__scalar, …  in f32_ppmm_2x4_unipass__scalar()
1449 static void f32_ppmm_4x2_unipass__scalar(benchmark::State& state, const char* net) {  in f32_ppmm_4x2_unipass__scalar()  argument
1450 …PPMM1PBenchmark(state, xnn_f32_ppmm_minmax_ukernel_4x2__scalar, xnn_x32_packx_ukernel_4x__scalar, …  in f32_ppmm_4x2_unipass__scalar()
1453 static void f32_ppmm_4x4_unipass__scalar(benchmark::State& state, const char* net) {  in f32_ppmm_4x4_unipass__scalar()  argument
1454 …PPMM1PBenchmark(state, xnn_f32_ppmm_minmax_ukernel_4x4__scalar, xnn_x32_packx_ukernel_4x__scalar, …  in f32_ppmm_4x4_unipass__scalar()
1457 static void f32_ppmm_3x3_unipass__scalar(benchmark::State& state, const char* net) {  in f32_ppmm_3x3_unipass__scalar()  argument
1458 …PPMM1PBenchmark(state, xnn_f32_ppmm_minmax_ukernel_3x3__scalar, xnn_x32_packx_ukernel_3x__scalar, …  in f32_ppmm_3x3_unipass__scalar()
1462 static void f32_ppmm_2x4_twopass__scalar(benchmark::State& state, const char* net) {  in f32_ppmm_2x4_twopass__scalar()  argument
1463 …PPMM2PBenchmark(state, xnn_f32_ppmm_minmax_ukernel_2x4__scalar, xnn_x32_packx_ukernel_2x__scalar, …  in f32_ppmm_2x4_twopass__scalar()
1466 static void f32_ppmm_4x2_twopass__scalar(benchmark::State& state, const char* net) {  in f32_ppmm_4x2_twopass__scalar()  argument
1467 …PPMM2PBenchmark(state, xnn_f32_ppmm_minmax_ukernel_4x2__scalar, xnn_x32_packx_ukernel_4x__scalar, …  in f32_ppmm_4x2_twopass__scalar()
1470 static void f32_ppmm_4x4_twopass__scalar(benchmark::State& state, const char* net) {  in f32_ppmm_4x4_twopass__scalar()  argument
1471 …PPMM2PBenchmark(state, xnn_f32_ppmm_minmax_ukernel_4x4__scalar, xnn_x32_packx_ukernel_4x__scalar, …  in f32_ppmm_4x4_twopass__scalar()
1474 static void f32_ppmm_3x3_twopass__scalar(benchmark::State& state, const char* net) {  in f32_ppmm_3x3_twopass__scalar()  argument
1475 …PPMM2PBenchmark(state, xnn_f32_ppmm_minmax_ukernel_3x3__scalar, xnn_x32_packx_ukernel_3x__scalar, …  in f32_ppmm_3x3_twopass__scalar()