[ { "50.00 percentile latency (ns)": 8768696, "90.00 percentile latency (ns)": 9794228, "95.00 percentile latency (ns)": 10145834, "97.00 percentile latency (ns)": 10489375, "99.00 percentile latency (ns)": 14728511, "99.90 percentile latency (ns)": 38535158, "Completed samples per second": 801241.88, "Max latency (ns)": 59167541, "Mean latency (ns)": 8944883, "Min duration satisfied": "Yes", "Min latency (ns)": 3871287, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "Performance constraints satisfied": "Yes", "Result is": "VALID", "SUT name": "LWIS_Server", "Scenario": "server", "Scheduled samples per second": 801252.02, "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "40GB", "accelerator_memory_configuration": "HBM2", "accelerator_model_name": "NVIDIA A100-SXM4-40GB", "accelerator_on-chip_memories": "", "accelerators_per_node": 8, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.accuracy": 75.738, "characteristics.good": 37869, "characteristics.scheduled_queries_per_second": 801252.02, "characteristics.scheduled_queries_per_second.normalized_per_core": 100156.5025, "characteristics.scheduled_queries_per_second.normalized_per_processor": 100156.5025, "characteristics.total": 50000, "ck_system": "DGX-A100_A100-SXM4-40GBx8_TRT", "ck_used": true, "cooling": "", "dataset": "ImageNet 2012", "dataset_link": "https://github.com/ctuning/ck/blob/master/docs/mlperf-automation/datasets/imagenet2012.md", "dim_x_default": "characteristics.scheduled_queries_per_second", "dim_y_default": "characteristics.accuracy", "dim_y_maximize": true, "division": "open", "formal_model": "resnet50", "formal_model_accuracy": 99.0, "formal_model_link": "https://github.com/mlcommons/ck-mlops/tree/main/package", "framework": "TensorRT 7.2.3, CUDA 11.1", "host_memory_capacity": "768 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 120, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "3.6 TB", "host_storage_type": "NVMe SSD", "hw_notes": "", "informal_model": "resnet50", "input_data_types": "int8", "key.accuracy": "characteristics.accuracy", "max_async_queries": 0, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 270336, "mlperf_version": 1.0, "normalize_cores": 8, "normalize_processors": 8, "note_code": "https://github.com/mlcommons/inference_results_v1.0/tree/master/open/Alibaba/code", "note_details": "https://github.com/mlcommons/inference_results_v1.0/tree/master/open/Alibaba/results/DGX-A100_A100-SXM4-40GBx8_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "TensorRT 7.2.3, CUDA 11.1, cuDNN 8.1.1, Driver 460.32.03, DALI 0.30.0; GCC 7.5.0; Python 3.7.10", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 2048, "print_timestamps": 0, "problem": false, "qsl_rng_seed": 7322528924094909334, "retraining": "N", "sample_index_rng_seed": 1570999273408051088, "samples_per_query": 1, "schedule_rng_seed": 3507442325620259414, "starting_weights_filename": "code/resnet50/tensorrt/ofa_autosinian_is176.onnx", "status": "available", "submitter": "Alibaba", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Alibaba", "sw_notes": "Powered by CK v2.5.8 (https://github.com/ctuning/ck)", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/DGX-A100_A100-SXM4-40GBx8_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 15000000, "target_qps": 801000, "task": "image classification", "task2": "image classification", "total_cores": 240, "uid": "7f7843c154b40937", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 6010023, "90.00 percentile latency (ns)": 6740510, "95.00 percentile latency (ns)": 6863466, "97.00 percentile latency (ns)": 6939303, "99.00 percentile latency (ns)": 7108750, "99.90 percentile latency (ns)": 7880924, "Completed samples per second": 120086.84, "Max latency (ns)": 46545758, "Mean latency (ns)": 6021071, "Min duration satisfied": "Yes", "Min latency (ns)": 3033829, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "Performance constraints satisfied": "Yes", "Result is": "VALID", "SUT name": "LWIS_Server", "Scenario": "server", "Scheduled samples per second": 120088.09, "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "22GB", "accelerator_memory_configuration": "HBM2", "accelerator_model_name": "NVIDIA A10", "accelerator_on-chip_memories": "", "accelerators_per_node": 2, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.accuracy": 75.718, "characteristics.good": 37859, "characteristics.scheduled_queries_per_second": 120088.09, "characteristics.scheduled_queries_per_second.normalized_per_core": 60044.045, "characteristics.scheduled_queries_per_second.normalized_per_processor": 60044.045, "characteristics.total": 50000, "ck_system": "A10x2_TRT", "ck_used": true, "cooling": "", "dataset": "ImageNet 2012", "dataset_link": "https://github.com/ctuning/ck/blob/master/docs/mlperf-automation/datasets/imagenet2012.md", "dim_x_default": "characteristics.scheduled_queries_per_second", "dim_y_default": "characteristics.accuracy", "dim_y_maximize": true, "division": "open", "formal_model": "resnet50", "formal_model_accuracy": 99.0, "formal_model_link": "https://github.com/mlcommons/ck-mlops/tree/main/package", "framework": "TensorRT 7.2.3, CUDA 11.1", "host_memory_capacity": "768 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 120, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "3.5 TB", "host_storage_type": "NVMe SSD", "hw_notes": "", "informal_model": "resnet50", "input_data_types": "int8", "key.accuracy": "characteristics.accuracy", "max_async_queries": 0, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 270336, "mlperf_version": 1.0, "normalize_cores": 2, "normalize_processors": 2, "note_code": "https://github.com/mlcommons/inference_results_v1.0/tree/master/open/Alibaba/code", "note_details": "https://github.com/mlcommons/inference_results_v1.0/tree/master/open/Alibaba/results/A10x2_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "TensorRT 7.2.3, CUDA 11.1, cuDNN 8.1.1, Driver 460.32.03, DALI 0.30.0; GCC 7.5.0; Python 3.7.10", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 2048, "print_timestamps": 0, "problem": false, "qsl_rng_seed": 7322528924094909334, "retraining": "N", "sample_index_rng_seed": 1570999273408051088, "samples_per_query": 1, "schedule_rng_seed": 3507442325620259414, "starting_weights_filename": "code/resnet50/tensorrt/ofa_autosinian_is176.onnx", "status": "Preview", "submitter": "Alibaba", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Alibaba", "sw_notes": "Powered by CK v2.5.8 (https://github.com/ctuning/ck)", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/A10x2_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 15000000, "target_qps": 120100, "task": "image classification", "task2": "image classification", "total_cores": 240, "uid": "6e2ab8aead49b780", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 5520979, "90.00 percentile latency (ns)": 5864834, "95.00 percentile latency (ns)": 5934313, "97.00 percentile latency (ns)": 5975688, "99.00 percentile latency (ns)": 6051052, "99.90 percentile latency (ns)": 6204125, "Completed samples per second": 176081.04, "Max latency (ns)": 17925809, "Mean latency (ns)": 5456794, "Min duration satisfied": "Yes", "Min latency (ns)": 3659007, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "Performance constraints satisfied": "Yes", "Result is": "VALID", "SUT name": "LWIS_Server", "Scenario": "server", "Scheduled samples per second": 176082.87, "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "16 GB", "accelerator_memory_configuration": "GDDR6", "accelerator_model_name": "Nvidia Tesla T4", "accelerator_on-chip_memories": "", "accelerators_per_node": 8, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.accuracy": 75.74, "characteristics.good": 37870, "characteristics.scheduled_queries_per_second": 176082.87, "characteristics.scheduled_queries_per_second.normalized_per_core": 22010.35875, "characteristics.scheduled_queries_per_second.normalized_per_processor": 22010.35875, "characteristics.total": 50000, "ck_system": "T4x8_TRT", "ck_used": true, "cooling": "", "dataset": "ImageNet 2012", "dataset_link": "https://github.com/ctuning/ck/blob/master/docs/mlperf-automation/datasets/imagenet2012.md", "dim_x_default": "characteristics.scheduled_queries_per_second", "dim_y_default": "characteristics.accuracy", "dim_y_maximize": true, "division": "open", "formal_model": "resnet50", "formal_model_accuracy": 99.0, "formal_model_link": "https://github.com/mlcommons/ck-mlops/tree/main/package", "framework": "TensorRT 7.2.2.3, CUDA 11.2", "host_memory_capacity": "512 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 120, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "7 TB", "host_storage_type": "NVMe SSD", "hw_notes": "", "informal_model": "resnet50", "input_data_types": "fp32", "key.accuracy": "characteristics.accuracy", "max_async_queries": 0, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 270336, "mlperf_version": 1.0, "normalize_cores": 8, "normalize_processors": 8, "note_code": "https://github.com/mlcommons/inference_results_v1.0/tree/master/open/Alibaba/code", "note_details": "https://github.com/mlcommons/inference_results_v1.0/tree/master/open/Alibaba/results/T4x8_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "TensorRT 7.2.3, CUDA 11.1, cuDNN 8.1.1, Driver 460.32.03, DALI 0.30.0; GCC 7.5.0; Python 3.7.10", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 2048, "print_timestamps": 0, "problem": false, "qsl_rng_seed": 7322528924094909334, "retraining": "N", "sample_index_rng_seed": 1570999273408051088, "samples_per_query": 1, "schedule_rng_seed": 3507442325620259414, "starting_weights_filename": "code/resnet50/tensorrt/ofa_autosinian_is176.onnx", "status": "available", "submitter": "Alibaba", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Alibaba", "sw_notes": "Powered by CK v2.5.8 (https://github.com/ctuning/ck)", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/T4x8_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 15000000, "target_qps": 176100, "task": "image classification", "task2": "image classification", "total_cores": 240, "uid": "f6f254b7ca6c3107", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 5463922, "90.00 percentile latency (ns)": 5761012, "95.00 percentile latency (ns)": 5842154, "97.00 percentile latency (ns)": 5898360, "99.00 percentile latency (ns)": 6063214, "99.90 percentile latency (ns)": 11002440, "Completed samples per second": 244990.0, "Max latency (ns)": 30752419, "Mean latency (ns)": 5480553, "Min duration satisfied": "Yes", "Min latency (ns)": 4430329, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "Performance constraints satisfied": "Yes", "Result is": "VALID", "SUT name": "LWIS_Server", "Scenario": "server", "Scheduled samples per second": 244992.37, "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "40GB", "accelerator_memory_configuration": "HBM2", "accelerator_model_name": "NVIDIA A100-SXM4-40GB", "accelerator_on-chip_memories": "", "accelerators_per_node": 8, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.accuracy": 76.028, "characteristics.good": 38014, "characteristics.scheduled_queries_per_second": 244992.37, "characteristics.scheduled_queries_per_second.normalized_per_core": 30624.04625, "characteristics.scheduled_queries_per_second.normalized_per_processor": 30624.04625, "characteristics.total": 50000, "ck_system": "DGX-A100_A100-SXM4-40GBx8_TRT", "ck_used": false, "cooling": "", "dataset": "ImageNet 2012", "dataset_link": "https://github.com/ctuning/ck/blob/master/docs/mlperf-automation/datasets/imagenet2012.md", "dim_x_default": "characteristics.scheduled_queries_per_second", "dim_y_default": "characteristics.accuracy", "dim_y_maximize": true, "division": "closed", "formal_model": "resnet50-v1.5", "formal_model_accuracy": 99.0, "formal_model_link": "https://github.com/mlcommons/ck-mlops/tree/main/package", "framework": "TensorRT 7.2.3, CUDA 11.1", "host_memory_capacity": "768 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 26, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "Intel(R) Xeon(R) Platinum 8269CY CPU @ 2.50GHz", "host_processors_per_node": 2, "host_storage_capacity": "3.6 TB", "host_storage_type": "NVMe SSD", "hw_notes": "", "informal_model": "resnet50", "input_data_types": "int8", "key.accuracy": "characteristics.accuracy", "max_async_queries": 0, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 270336, "mlperf_version": 1.0, "normalize_cores": 8, "normalize_processors": 8, "note_code": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/Alibaba/code", "note_details": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/Alibaba/results/DGX-A100_A100-SXM4-40GBx8_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.4", "other_software_stack": "TensorRT 7.2.3, CUDA 11.1, cuDNN 8.1.1, Driver 460.32.03, DALI 0.30.0", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 2048, "print_timestamps": 0, "problem": false, "qsl_rng_seed": 7322528924094909334, "retraining": "N", "sample_index_rng_seed": 1570999273408051088, "samples_per_query": 1, "schedule_rng_seed": 3507442325620259414, "starting_weights_filename": "resnet50_v1.onnx", "status": "available", "submitter": "Alibaba", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Alibaba", "sw_notes": "", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/DGX-A100_A100-SXM4-40GBx8_TRT", "system_name": "Alibaba Cloud Sinian Platform A100 (8x A100-SXM4-40GB)", "system_type": "datacenter", "target_latency (ns)": 15000000, "target_qps": 245000, "task": "image classification", "task2": "image classification", "total_cores": 52, "uid": "77723e6cb2f44d48", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 5630590, "90.00 percentile latency (ns)": 7992180, "95.00 percentile latency (ns)": 8435174, "97.00 percentile latency (ns)": 8720517, "99.00 percentile latency (ns)": 9248438, "99.90 percentile latency (ns)": 10443559, "Completed samples per second": 21991.16, "Max latency (ns)": 16967476, "Mean latency (ns)": 5928762, "Min duration satisfied": "Yes", "Min latency (ns)": 1608951, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "Performance constraints satisfied": "Yes", "Result is": "VALID", "SUT name": "LWIS_Server", "Scenario": "server", "Scheduled samples per second": 21991.28, "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "22GB", "accelerator_memory_configuration": "HBM2", "accelerator_model_name": "NVIDIA A10", "accelerator_on-chip_memories": "", "accelerators_per_node": 2, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.accuracy": 76.028, "characteristics.good": 38014, "characteristics.scheduled_queries_per_second": 21991.28, "characteristics.scheduled_queries_per_second.normalized_per_core": 10995.64, "characteristics.scheduled_queries_per_second.normalized_per_processor": 10995.64, "characteristics.total": 50000, "ck_system": "A10x2_TRT", "ck_used": false, "cooling": "", "dataset": "ImageNet 2012", "dataset_link": "https://github.com/ctuning/ck/blob/master/docs/mlperf-automation/datasets/imagenet2012.md", "dim_x_default": "characteristics.scheduled_queries_per_second", "dim_y_default": "characteristics.accuracy", "dim_y_maximize": true, "division": "closed", "formal_model": "resnet50-v1.5", "formal_model_accuracy": 99.0, "formal_model_link": "https://github.com/mlcommons/ck-mlops/tree/main/package", "framework": "TensorRT 7.2.3, CUDA 11.1", "host_memory_capacity": "768 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 32, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "Intel(R) Xeon(R) Platinum 8357C CPU @ 2.70GHz", "host_processors_per_node": 2, "host_storage_capacity": "3.5 TB", "host_storage_type": "NVMe SSD", "hw_notes": "", "informal_model": "resnet50", "input_data_types": "int8", "key.accuracy": "characteristics.accuracy", "max_async_queries": 0, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 270336, "mlperf_version": 1.0, "normalize_cores": 2, "normalize_processors": 2, "note_code": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/Alibaba/code", "note_details": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/Alibaba/results/A10x2_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.4", "other_software_stack": "TensorRT 7.2.3, CUDA 11.1, cuDNN 8.1.1, Driver 460.32.03, DALI 0.30.0", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 2048, "print_timestamps": 0, "problem": false, "qsl_rng_seed": 7322528924094909334, "retraining": "N", "sample_index_rng_seed": 1570999273408051088, "samples_per_query": 1, "schedule_rng_seed": 3507442325620259414, "starting_weights_filename": "resnet50_v1.onnx", "status": "Preview", "submitter": "Alibaba", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Alibaba", "sw_notes": "", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/A10x2_TRT", "system_name": "Alibaba Cloud Sinian Platform A10 (2x A10)", "system_type": "datacenter", "target_latency (ns)": 15000000, "target_qps": 22000, "task": "image classification", "task2": "image classification", "total_cores": 64, "uid": "4567ad841326f0ce", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 357755, "90.00 percentile latency (ns)": 950166, "95.00 percentile latency (ns)": 1510435, "97.00 percentile latency (ns)": 2694032, "99.00 percentile latency (ns)": 10161237, "99.90 percentile latency (ns)": 25331487, "Completed samples per second": 45167.59, "Max latency (ns)": 26338387, "Mean latency (ns)": 715484, "Min duration satisfied": "Yes", "Min latency (ns)": 125761, "Min queries satisfied": "Yes", "Mode": "Performance", "Performance constraints satisfied": "Yes", "Result is": "VALID", "SUT name": "", "Scenario": "server", "Scheduled samples per second": 45169.48, "accelerator_memory_capacity": 0, "accelerator_memory_configuration": "n/a", "accelerator_model_name": "HanGuang 800", "accelerators_per_node": 1, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "characteristics.accuracy": 76.038, "characteristics.good": 38019, "characteristics.scheduled_queries_per_second": 45169.48, "characteristics.scheduled_queries_per_second.normalized_per_core": 45169.48, "characteristics.scheduled_queries_per_second.normalized_per_processor": 45169.48, "characteristics.total": 50000, "ck_system": "alibaba_HanGuang", "ck_used": true, "dataset": "ImageNet 2012", "dataset_link": "https://github.com/ctuning/ck/blob/master/docs/mlperf-automation/datasets/imagenet2012.md", "dim_x_default": "characteristics.scheduled_queries_per_second", "dim_y_default": "characteristics.accuracy", "dim_y_maximize": true, "division": "closed", "formal_model": "resnet50-v1.5", "formal_model_accuracy": 99.0, "formal_model_link": "https://github.com/mlcommons/ck-mlops/tree/main/package", "framework": "TensorFlow 1.13.1 6612da89516247503f03ef76e974b51a434fb52e", "host_memory_capacity": "512GB", "host_processor_core_count": 120, "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "12TB", "host_storage_type": "SSD", "informal_model": "resnet", "input_data_types": "int8", "key.accuracy": "characteristics.accuracy", "max_async_queries": 10000, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 2500000, "mlperf_version": 0.5, "normalize_cores": 1, "normalize_processors": 1, "note_code": "https://github.com/mlcommons/inference_results_v0.5/tree/master/closed/Alibaba/code", "note_details": "https://github.com/mlcommons/inference_results_v0.5/tree/master/closed/Alibaba/results/alibaba_HanGuang", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "HanGuangAI 1.0.3, python 3.5.2; GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 1600, "print_timestamps": true, "problem": false, "qsl_rng_seed": 3133965575612453542, "retraining": "none", "sample_index_rng_seed": 665484352860916858, "samples_per_query": 1, "schedule_rng_seed": 3622009729038561421, "starting_weights_filename": "https://zenodo.org/record/2535873/files/resnet50_v1.pb", "status": "preview", "submitter": "Alibaba", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Alibaba", "sw_notes": "Powered by CK v2.5.8 (https://github.com/ctuning/ck)", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/alibaba_HanGuang", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "target_latency (ns)": 15000000, "target_qps": 45120, "task": "image classification", "task2": "image classification", "total_cores": 240, "uid": "4480d966a6a250aa", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization" } ]