[ { "50.00 percentile latency (ns)": 48030158, "90.00 percentile latency (ns)": 67777239, "95.00 percentile latency (ns)": 77125349, "97.00 percentile latency (ns)": 83455063, "99.00 percentile latency (ns)": 98372749, "99.90 percentile latency (ns)": 127458439, "Completed samples per second": 49.99, "Max latency (ns)": 182999982, "Mean latency (ns)": 52743012, "Min duration satisfied": "Yes", "Min latency (ns)": 37459566, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "Performance constraints satisfied": "Yes", "Result is": "VALID", "SUT name": "SUT", "Scenario": "server", "Scheduled samples per second": 49.99, "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "N/A", "accelerator_memory_configuration": "", "accelerator_model_name": "N/A", "accelerator_on-chip_memories": "", "accelerators_per_node": 0, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.mAP": 19.876, "characteristics.scheduled_queries_per_second": 49.99, "characteristics.scheduled_queries_per_second.normalized_per_core": 0.6248750000000001, "characteristics.scheduled_queries_per_second.normalized_per_processor": 24.995, "ck_system": "1-node-2S-ICX-OpenVINO-INT8", "ck_used": false, "cooling": "Air", "dataset": "COCO 2017 (300x300)", "dataset_link": "https://github.com/ctuning/ck/blob/master/docs/mlperf-automation/datasets/coco2017.md", "dim_x_default": "characteristics.scheduled_queries_per_second", "dim_y_default": "characteristics.mAP", "dim_y_maximize": true, "division": "closed", "formal_model": "ssd-mobilenet", "formal_model_accuracy": 99.0, "formal_model_link": "https://github.com/mlcommons/ck-mlops/tree/main/package", "framework": "OpenVINO", "host_memory_capacity": "1024GB", "host_memory_configuration": "8 slots / 64GB each / 3200 MT/s per socket", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 40, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "Intel(R) Xeon(R) Platinum 8380 CPU @ 2.30GHz", "host_processors_per_node": 2, "host_storage_capacity": "", "host_storage_type": "", "hw_notes": "Intel Devlopment Platform (2U Server)", "informal_model": "ssd-resnet34", "input_data_types": "No change.", "key.accuracy": "characteristics.mAP", "max_async_queries": 0, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 270336, "mlperf_version": 1.1, "normalize_cores": 80, "normalize_processors": 2, "note_code": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Intel/code", "note_details": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Intel/results/1-node-2S-ICX-OpenVINO-INT8", "number_of_nodes": 1, "operating_system": "Ubuntu 20.04.1 LTS", "other_software_stack": "5.4.0-45-generic", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 64, "print_timestamps": 0, "problem": false, "qsl_rng_seed": 1624344308455410291, "retraining": "We don’t retrain the model weight.", "sample_index_rng_seed": 517984244576520566, "samples_per_query": 1, "schedule_rng_seed": 10051496985653635065, "starting_weights_filename": "https://zenodo.org/record/3228411/files/resnet34-ssd1200.onnx", "status": "available", "submitter": "Intel", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Intel", "sw_notes": "", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/1-node-2S-ICX-OpenVINO-INT8", "system_name": "1-node-2S-ICX-OpenVINO-INT8", "system_type": "datacenter", "target_latency (ns)": 100000000, "target_qps": 50, "task": "object detection", "task2": "object detection", "total_cores": 80, "uid": "8229603e64f60ad1", "use_accelerator": false, "weight_data_types": "No change.", "weight_transformations": "We transfer the weight from int8 datatype in ONNX file to int8 datatype in OpenVino IR file." }, { "50.00 percentile latency (ns)": 52492960, "90.00 percentile latency (ns)": 68185610, "95.00 percentile latency (ns)": 79071890, "97.00 percentile latency (ns)": 85715898, "99.00 percentile latency (ns)": 98928283, "99.90 percentile latency (ns)": 128151673, "Completed samples per second": 40.09, "Max latency (ns)": 179200056, "Mean latency (ns)": 56201071, "Min duration satisfied": "Yes", "Min latency (ns)": 40411353, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "Performance constraints satisfied": "Yes", "Result is": "VALID", "SUT name": "SUT", "Scenario": "server", "Scheduled samples per second": 40.09, "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "N/A", "accelerator_memory_configuration": "", "accelerator_model_name": "N/A", "accelerator_on-chip_memories": "", "accelerators_per_node": 0, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.mAP": 19.886, "characteristics.scheduled_queries_per_second": 40.09, "characteristics.scheduled_queries_per_second.normalized_per_core": 0.501125, "characteristics.scheduled_queries_per_second.normalized_per_processor": 20.045, "ck_system": "1-node-2S-ICX-OpenVINO-INT8", "ck_used": false, "cooling": "", "dataset": "COCO 2017 (300x300)", "dataset_link": "https://github.com/ctuning/ck/blob/master/docs/mlperf-automation/datasets/coco2017.md", "dim_x_default": "characteristics.scheduled_queries_per_second", "dim_y_default": "characteristics.mAP", "dim_y_maximize": true, "division": "closed", "formal_model": "ssd-mobilenet", "formal_model_accuracy": 99.0, "formal_model_link": "https://github.com/mlcommons/ck-mlops/tree/main/package", "framework": "OpenVINO", "host_memory_capacity": "1024GB", "host_memory_configuration": "8 slots / 64GB each / 3200 MT/s per socket", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 40, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "Intel(R) Xeon(R) Platinum 8380 CPU @ 2.30GHz", "host_processors_per_node": 2, "host_storage_capacity": "", "host_storage_type": "", "hw_notes": "", "informal_model": "ssd-resnet34", "input_data_types": "No change.", "key.accuracy": "characteristics.mAP", "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": 80, "normalize_processors": 2, "note_code": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/Intel/code", "note_details": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/Intel/results/1-node-2S-ICX-OpenVINO-INT8", "number_of_nodes": 1, "operating_system": "Ubuntu 20.04.1 LTS", "other_software_stack": "5.4.0-45-generic", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 64, "print_timestamps": 0, "problem": false, "qsl_rng_seed": 7322528924094909334, "retraining": "We don’t retrain the model weight.", "sample_index_rng_seed": 1570999273408051088, "samples_per_query": 1, "schedule_rng_seed": 3507442325620259414, "starting_weights_filename": "https://zenodo.org/record/3228411/files/resnet34-ssd1200.onnx", "status": "preview", "submitter": "Intel", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Intel", "sw_notes": "", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/1-node-2S-ICX-OpenVINO-INT8", "system_name": "1-node-2S-ICX-OpenVINO-INT8", "system_type": "datacenter", "target_latency (ns)": 100000000, "target_qps": 40, "task": "object detection", "task2": "object detection", "total_cores": 80, "uid": "eb66c6574697a8e8", "use_accelerator": false, "weight_data_types": "No change.", "weight_transformations": "We transfer the weight from int8 datatype in ONNX file to int8 datatype in OpenVino IR file." }, { "50.00 percentile latency (ns)": 67241789, "90.00 percentile latency (ns)": 70527281, "95.00 percentile latency (ns)": 71201347, "97.00 percentile latency (ns)": 71665934, "99.00 percentile latency (ns)": 75519895, "99.90 percentile latency (ns)": 89996806, "Completed samples per second": 130.29, "Max latency (ns)": 146719516, "Mean latency (ns)": 67173893, "Min duration satisfied": "Yes", "Min latency (ns)": 51174332, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "Performance constraints satisfied": "Yes", "Result is": "VALID", "SUT name": "SUT\u0000", "Scenario": "server", "Scheduled samples per second": 130.3, "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "N/A", "accelerator_memory_configuration": "", "accelerator_model_name": "N/A", "accelerator_on-chip_memories": "", "accelerators_per_node": 0, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.mAP": 19.886, "characteristics.scheduled_queries_per_second": 130.3, "characteristics.scheduled_queries_per_second.normalized_per_core": 0.13572916666666668, "characteristics.scheduled_queries_per_second.normalized_per_processor": 16.2875, "ck_system": "1-node-8S-CPX-OpenVINO-INT8", "ck_used": true, "cooling": "", "dataset": "COCO 2017 (300x300)", "dataset_link": "https://github.com/ctuning/ck/blob/master/docs/mlperf-automation/datasets/coco2017.md", "dim_x_default": "characteristics.scheduled_queries_per_second", "dim_y_default": "characteristics.mAP", "dim_y_maximize": true, "division": "closed", "formal_model": "ssd-mobilenet", "formal_model_accuracy": 99.0, "formal_model_link": "https://github.com/mlcommons/ck-mlops/tree/main/package", "framework": "OpenVINO", "host_memory_capacity": "3072GB", "host_memory_configuration": "6 slots / 64GB each / 3200 MT/s per socket", "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": 8, "host_storage_capacity": "", "host_storage_type": "", "hw_notes": "", "informal_model": "ssd-resnet34", "input_data_types": "No change.", "key.accuracy": "characteristics.mAP", "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": 960, "normalize_processors": 8, "note_code": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/Intel/code", "note_details": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/Intel/results/1-node-8S-CPX-OpenVINO-INT8", "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": "5.4.0-66-generic; GCC 7.5.0; Python 3.7.10", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 64, "print_timestamps": 0, "problem": false, "qsl_rng_seed": 7322528924094909334, "retraining": "We don’t retrain the model weight.", "sample_index_rng_seed": 1570999273408051088, "samples_per_query": 1, "schedule_rng_seed": 3507442325620259414, "starting_weights_filename": "The original weight filename: https://zenodo.org/record/3928973/files/224_224_160.onnx", "status": "available", "submitter": "Intel", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Intel", "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/1-node-8S-CPX-OpenVINO-INT8", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 100000000, "target_qps": 130, "task": "object detection", "task2": "object detection", "total_cores": 960, "uid": "c189f94872875f54", "use_accelerator": false, "weight_data_types": "No change.", "weight_transformations": "We transfer the weight from int8 datatype in ONNX file to int8 datatype in OpenVino IR file." }, { "50.00 percentile latency (ns)": 67284106, "90.00 percentile latency (ns)": 71509919, "95.00 percentile latency (ns)": 82325438, "97.00 percentile latency (ns)": 89191458, "99.00 percentile latency (ns)": 99699451, "99.90 percentile latency (ns)": 121087684, "Completed samples per second": 68.15, "Max latency (ns)": 263000417, "Mean latency (ns)": 68417383, "Min duration satisfied": "Yes", "Min latency (ns)": 60849240, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "Performance constraints satisfied": "Yes", "Result is": "VALID", "SUT name": "SUT\u0000", "Scenario": "server", "Scheduled samples per second": 68.16, "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "N/A", "accelerator_memory_configuration": "", "accelerator_model_name": "N/A", "accelerator_on-chip_memories": "", "accelerators_per_node": 0, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.mAP": 19.886, "characteristics.scheduled_queries_per_second": 68.16, "characteristics.scheduled_queries_per_second.normalized_per_core": 0.142, "characteristics.scheduled_queries_per_second.normalized_per_processor": 17.04, "ck_system": "1-node-4S-CPX-OpenVINO-INT8", "ck_used": true, "cooling": "", "dataset": "COCO 2017 (300x300)", "dataset_link": "https://github.com/ctuning/ck/blob/master/docs/mlperf-automation/datasets/coco2017.md", "dim_x_default": "characteristics.scheduled_queries_per_second", "dim_y_default": "characteristics.mAP", "dim_y_maximize": true, "division": "closed", "formal_model": "ssd-mobilenet", "formal_model_accuracy": 99.0, "formal_model_link": "https://github.com/mlcommons/ck-mlops/tree/main/package", "framework": "OpenVINO", "host_memory_capacity": "1536GB", "host_memory_configuration": "6 slots / 64GB each / 3200 MT/s per socket", "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": 4, "host_storage_capacity": "", "host_storage_type": "", "hw_notes": "", "informal_model": "ssd-resnet34", "input_data_types": "No change.", "key.accuracy": "characteristics.mAP", "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": 480, "normalize_processors": 4, "note_code": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/Intel/code", "note_details": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/Intel/results/1-node-4S-CPX-OpenVINO-INT8", "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": "5.4.0-65-generic; GCC 7.5.0; Python 3.7.10", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 64, "print_timestamps": 0, "problem": false, "qsl_rng_seed": 7322528924094909334, "retraining": "We don’t retrain the model weight.", "sample_index_rng_seed": 1570999273408051088, "samples_per_query": 1, "schedule_rng_seed": 3507442325620259414, "starting_weights_filename": "https://zenodo.org/record/2535873/files/resnet50_v1.pb", "status": "available", "submitter": "Intel", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Intel", "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/1-node-4S-CPX-OpenVINO-INT8", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 100000000, "target_qps": 68, "task": "object detection", "task2": "object detection", "total_cores": 480, "uid": "765c2e2aa464c87b", "use_accelerator": false, "weight_data_types": "No change.", "weight_transformations": "https://docs.openvinotoolkit.org/latest/openvino_docs_IE_DG_Bfloat16Inference.html" }, { "50.00 percentile latency (ns)": 2363285, "90.00 percentile latency (ns)": 3480321, "95.00 percentile latency (ns)": 3961457, "97.00 percentile latency (ns)": 4371236, "99.00 percentile latency (ns)": 7416184, "99.90 percentile latency (ns)": 54726120, "Completed samples per second": 5262.7, "Max latency (ns)": 70242530, "Mean latency (ns)": 2937884, "Min duration satisfied": "Yes", "Min latency (ns)": 1984286, "Min queries satisfied": "Yes", "Mode": "Performance", "Performance constraints satisfied": "Yes", "Result is": "VALID", "SUT name": "SUT\u0000", "Scenario": "server", "Scheduled samples per second": 5262.92, "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "N/A", "accelerator_memory_configuration": "", "accelerator_model_name": "N/A", "accelerator_on-chip_memories": "", "accelerators_per_node": 0, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "characteristics.mAP": 22.627, "characteristics.scheduled_queries_per_second": 5262.92, "characteristics.scheduled_queries_per_second.normalized_per_core": 21.928833333333333, "characteristics.scheduled_queries_per_second.normalized_per_processor": 2631.46, "ck_system": "clx_9282-2s_openvino-linux", "ck_used": true, "cooling": "", "dataset": "COCO 2017 (300x300)", "dataset_link": "https://github.com/ctuning/ck/blob/master/docs/mlperf-automation/datasets/coco2017.md", "dim_x_default": "characteristics.scheduled_queries_per_second", "dim_y_default": "characteristics.mAP", "dim_y_maximize": true, "division": "closed", "formal_model": "ssd-mobilenet", "formal_model_accuracy": 99.0, "formal_model_link": "https://github.com/mlcommons/ck-mlops/tree/main/package", "framework": "OpenVino", "host_memory_capacity": "768GB", "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": "422G", "host_storage_type": "SSD", "hw_notes": "", "informal_model": "ssd-small", "input_data_types": "We transform the fp32 input data to int8 data.", "key.accuracy": "characteristics.mAP", "max_async_queries": 0, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 270336, "mlperf_version": 0.5, "normalize_cores": 240, "normalize_processors": 2, "note_code": "https://github.com/mlcommons/inference_results_v0.5/tree/master/closed/Intel/code", "note_details": "https://github.com/mlcommons/inference_results_v0.5/tree/master/closed/Intel/results/clx_9282-2s_openvino-linux", "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": "MKL (2019.4.243); GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 256, "print_timestamps": true, "problem": false, "qsl_rng_seed": 3133965575612453542, "retraining": "We don’t retrain the model weight.", "sample_index_rng_seed": 665484352860916858, "samples_per_query": 1, "schedule_rng_seed": 3622009729038561421, "starting_weights_filename": "The original weight filename: https://zenodo.org/record/3401714/files/ssd_mobilenet_v1_quant_ft_no_zero_point_frozen_inference_graph.pb and the transformed int8 weight named init_net_int8_euler.pb", "status": "available", "submitter": "Intel", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Intel", "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/clx_9282-2s_openvino-linux", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "target_latency (ns)": 10000000, "target_qps": 5260, "task": "object detection", "task2": "object detection", "total_cores": 240, "uid": "1d4a52873b062560", "use_accelerator": false, "weight_data_types": "We transform the original fp32 weight to int8 weight.", "weight_transformations": "We transform the weight from fp32 datatype in onnx file to int8 datatype in pb file." }, { "50.00 percentile latency (ns)": 77879377, "90.00 percentile latency (ns)": 84327571, "95.00 percentile latency (ns)": 87707228, "97.00 percentile latency (ns)": 90105959, "99.00 percentile latency (ns)": 94402041, "99.90 percentile latency (ns)": 113059433, "Completed samples per second": 129.88, "Max latency (ns)": 263435494, "Mean latency (ns)": 78501653, "Min duration satisfied": "Yes", "Min latency (ns)": 63116455, "Min queries satisfied": "Yes", "Mode": "Performance", "Performance constraints satisfied": "Yes", "Result is": "VALID", "SUT name": "SUT\u0000", "Scenario": "server", "Scheduled samples per second": 129.88, "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "N/A", "accelerator_memory_configuration": "", "accelerator_model_name": "N/A", "accelerator_on-chip_memories": "", "accelerators_per_node": 0, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.scheduled_queries_per_second": 129.88, "characteristics.scheduled_queries_per_second.normalized_per_core": 0.13529166666666667, "characteristics.scheduled_queries_per_second.normalized_per_processor": 16.235, "ck_system": "1-node-8s-cpx-ov-int8", "ck_used": true, "cooling": "", "dataset": "COCO 2017 (300x300)", "dataset_link": "https://github.com/ctuning/ck/blob/master/docs/mlperf-automation/datasets/coco2017.md", "dim_x_default": "characteristics.scheduled_queries_per_second", "dim_y_default": "characteristics.mAP", "dim_y_maximize": true, "division": "closed", "formal_model": "ssd-mobilenet", "formal_model_accuracy": 99.0, "formal_model_link": "https://github.com/mlcommons/ck-mlops/tree/main/package", "framework": "OpenVINO", "host_memory_capacity": "1536GB", "host_memory_configuration": "6 slots / 32GB each / 3200 MT/s per socket", "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": 8, "host_storage_capacity": "", "host_storage_type": "", "hw_notes": "", "informal_model": "ssd-resnet34", "input_data_types": "No change.", "max_async_queries": 0, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 270336, "mlperf_version": 0.7, "normalize_cores": 960, "normalize_processors": 8, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/Intel/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/Intel/results/1-node-8s-cpx-ov-int8", "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": "5.4.0-45-generic; GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 64, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "We don’t retrain the model weight.", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 1, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "https://zenodo.org/record/3228411/files/resnet34-ssd1200.onnx", "status": "preview", "submitter": "Intel", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Intel", "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/1-node-8s-cpx-ov-int8", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 100000000, "target_qps": 130, "task": "object detection", "task2": "object detection", "total_cores": 960, "uid": "5629f79411ffd643", "use_accelerator": false, "weight_data_types": "No change.", "weight_transformations": "We transfer the weight from int8 datatype in ONNX file to int8 datatype in OpenVino IR file." }, { "50.00 percentile latency (ns)": 42558513, "90.00 percentile latency (ns)": 66354414, "95.00 percentile latency (ns)": 75710479, "97.00 percentile latency (ns)": 83104972, "99.00 percentile latency (ns)": 98745190, "99.90 percentile latency (ns)": 133455397, "Completed samples per second": 64.94, "Max latency (ns)": 236608952, "Mean latency (ns)": 48951846, "Min duration satisfied": "Yes", "Min latency (ns)": 36054433, "Min queries satisfied": "Yes", "Mode": "Performance", "Performance constraints satisfied": "Yes", "Result is": "VALID", "SUT name": "SUT\u0000", "Scenario": "server", "Scheduled samples per second": 64.94, "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "N/A", "accelerator_memory_configuration": "", "accelerator_model_name": "N/A", "accelerator_on-chip_memories": "", "accelerators_per_node": 0, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.scheduled_queries_per_second": 64.94, "characteristics.scheduled_queries_per_second.normalized_per_core": 0.13529166666666667, "characteristics.scheduled_queries_per_second.normalized_per_processor": 16.235, "ck_system": "1-node-4s-cpx-ov-int8", "ck_used": true, "cooling": "", "dataset": "COCO 2017 (300x300)", "dataset_link": "https://github.com/ctuning/ck/blob/master/docs/mlperf-automation/datasets/coco2017.md", "dim_x_default": "characteristics.scheduled_queries_per_second", "dim_y_default": "characteristics.mAP", "dim_y_maximize": true, "division": "closed", "formal_model": "ssd-mobilenet", "formal_model_accuracy": 99.0, "formal_model_link": "https://github.com/mlcommons/ck-mlops/tree/main/package", "framework": "OpenVINO", "host_memory_capacity": "768GB", "host_memory_configuration": "6 slots / 32GB each / 3200 MT/s per socket", "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": 4, "host_storage_capacity": "", "host_storage_type": "", "hw_notes": "", "informal_model": "ssd-resnet34", "input_data_types": "No change.", "max_async_queries": 0, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 270336, "mlperf_version": 0.7, "normalize_cores": 480, "normalize_processors": 4, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/Intel/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/Intel/results/1-node-4s-cpx-ov-int8", "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": "5.4.0-45-generic; GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 64, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "We don’t retrain the model weight.", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 1, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "https://zenodo.org/record/3228411/files/resnet34-ssd1200.onnx", "status": "preview", "submitter": "Intel", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Intel", "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/1-node-4s-cpx-ov-int8", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 100000000, "target_qps": 65, "task": "object detection", "task2": "object detection", "total_cores": 480, "uid": "ee1f48bc44ff59fe", "use_accelerator": false, "weight_data_types": "No change.", "weight_transformations": "We transfer the weight from int8 datatype in ONNX file to int8 datatype in OpenVino IR file." } ]