[ { "50.00 percentile": 1, "50.00 percentile latency (ns)": 27172601, "90.00 percentile": 1, "90.00 percentile latency (ns)": 47029522, "95.00 percentile": 1, "95.00 percentile latency (ns)": 48055474, "97.00 percentile": 1, "97.00 percentile latency (ns)": 48566192, "99.00 percentile": 1, "99.00 percentile latency (ns)": 49119566, "99.90 percentile": 2, "99.90 percentile latency (ns)": 49946423, "Max latency (ns)": 70732428, "Mean latency (ns)": 30258926, "Min duration satisfied": "Yes", "Min latency (ns)": 10919940, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "Per-sample latency": "", "Performance constraints satisfied": "Yes", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per query": 344, "Scenario": "multistream", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "16 GB", "accelerator_memory_configuration": "GDDR6", "accelerator_model_name": "NVIDIA T4", "accelerator_on-chip_memories": "", "accelerators_per_node": 1, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.mAP": 22.911, "characteristics.samples_per_query": 344, "characteristics.samples_per_query.normalized_per_core": 344.0, "characteristics.samples_per_query.normalized_per_processor": 344.0, "ck_system": "T4x1_TRT", "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.samples_per_query", "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": "TensorRT 7.2.3, CUDA 11.1", "host_memory_capacity": "128 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 4, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "Intel(R) Xeon(R) D-2123IT CPU @ 2.20GHz", "host_processors_per_node": 1, "host_storage_capacity": "2 TB", "host_storage_type": "M.2", "hw_notes": "ECC on", "informal_model": "ssd-mobilenet", "input_data_types": "int8", "key.accuracy": "characteristics.mAP", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 270336, "mlperf_version": 1.0, "normalize_cores": 1, "normalize_processors": 1, "note_code": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/Lenovo/code", "note_details": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/Lenovo/results/T4x1_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": 1024, "print_timestamps": 0, "problem": false, "qsl_rng_seed": 7322528924094909334, "retraining": "N", "sample_index_rng_seed": 1570999273408051088, "samples_per_query": 344, "schedule_rng_seed": 3507442325620259414, "starting_weights_filename": "ssd_mobilenet_v1_coco_2018_01_28/frozen_inference_graph.pb", "status": "available", "submitter": "Lenovo", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Lenovo", "sw_notes": "", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/T4x1_TRT", "system_name": "Lenovo ThinkSystem SE350 server", "system_type": "edge", "target_latency (ns)": 50000000, "target_qps": 20, "task": "object detection", "task2": "object detection", "total_cores": 4, "uid": "98d325934d248ebb", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile": 1, "50.00 percentile latency (ns)": 46653620, "90.00 percentile": 1, "90.00 percentile latency (ns)": 63654635, "95.00 percentile": 1, "95.00 percentile latency (ns)": 63743044, "97.00 percentile": 1, "97.00 percentile latency (ns)": 63837211, "99.00 percentile": 1, "99.00 percentile latency (ns)": 63949232, "99.90 percentile": 1, "99.90 percentile latency (ns)": 64213496, "Max latency (ns)": 64911725, "Mean latency (ns)": 47455339, "Min duration satisfied": "Yes", "Min latency (ns)": 21911040, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "Per-sample latency": "", "Performance constraints satisfied": "Yes", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per query": 8, "Scenario": "multistream", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "16 GB", "accelerator_memory_configuration": "GDDR6", "accelerator_model_name": "NVIDIA T4", "accelerator_on-chip_memories": "", "accelerators_per_node": 1, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.mAP": 20.111, "characteristics.samples_per_query": 8, "characteristics.samples_per_query.normalized_per_core": 8.0, "characteristics.samples_per_query.normalized_per_processor": 8.0, "ck_system": "T4x1_TRT", "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.samples_per_query", "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": "TensorRT 7.2.3, CUDA 11.1", "host_memory_capacity": "128 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 4, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "Intel(R) Xeon(R) D-2123IT CPU @ 2.20GHz", "host_processors_per_node": 1, "host_storage_capacity": "2 TB", "host_storage_type": "M.2", "hw_notes": "ECC on", "informal_model": "ssd-resnet34", "input_data_types": "int8", "key.accuracy": "characteristics.mAP", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 270336, "mlperf_version": 1.0, "normalize_cores": 1, "normalize_processors": 1, "note_code": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/Lenovo/code", "note_details": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/Lenovo/results/T4x1_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": 64, "print_timestamps": 0, "problem": false, "qsl_rng_seed": 7322528924094909334, "retraining": "N", "sample_index_rng_seed": 1570999273408051088, "samples_per_query": 8, "schedule_rng_seed": 3507442325620259414, "starting_weights_filename": "resnet34-ssd1200.pytorch", "status": "available", "submitter": "Lenovo", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Lenovo", "sw_notes": "", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/T4x1_TRT", "system_name": "Lenovo ThinkSystem SE350 server", "system_type": "edge", "target_latency (ns)": 66666666, "target_qps": 15, "task": "object detection", "task2": "object detection", "total_cores": 4, "uid": "4f6292a5f472bb73", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile": 1, "50.00 percentile latency (ns)": 26783675, "90.00 percentile": 1, "90.00 percentile latency (ns)": 46869632, "95.00 percentile": 1, "95.00 percentile latency (ns)": 48002393, "97.00 percentile": 1, "97.00 percentile latency (ns)": 48231904, "99.00 percentile": 1, "99.00 percentile latency (ns)": 48611710, "99.90 percentile": 1, "99.90 percentile latency (ns)": 48897875, "Max latency (ns)": 49480069, "Mean latency (ns)": 30442560, "Min duration satisfied": "Yes", "Min latency (ns)": 10765111, "Min queries satisfied": "Yes", "Mode": "Performance", "Per-sample latency": "", "Performance constraints satisfied": "Yes", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per query": 368, "Scenario": "multistream", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "16 GB", "accelerator_memory_configuration": "GDDR6", "accelerator_model_name": "NVIDIA T4", "accelerator_on-chip_memories": "", "accelerators_per_node": 1, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.mAP": 22.911, "characteristics.samples_per_query": 368, "characteristics.samples_per_query.normalized_per_core": 368.0, "characteristics.samples_per_query.normalized_per_processor": 368.0, "ck_system": "T4x1_TRT", "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.samples_per_query", "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": "TensorRT 7.2, CUDA 11.0 Update 1", "host_memory_capacity": "128 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 4, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "Intel(R) Xeon(R) D-2123IT CPU @ 2.20GHz", "host_processors_per_node": 1, "host_storage_capacity": "2 TB", "host_storage_type": "M.2", "hw_notes": "ECC off", "informal_model": "ssd-mobilenet", "input_data_types": "int8", "key.accuracy": "characteristics.mAP", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 270336, "mlperf_version": 0.7, "normalize_cores": 1, "normalize_processors": 1, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/Lenovo/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/Lenovo/results/T4x1_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.4", "other_software_stack": "TensorRT 7.2, CUDA 11.0 Update 1, cuDNN 8.0.2, DALI 0.25.0", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 1024, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 368, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "ssd_mobilenet_v1_coco_2018_01_28/frozen_inference_graph.pb", "status": "available", "submitter": "Lenovo", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Lenovo", "sw_notes": "", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/T4x1_TRT", "system_name": "LenovoSE350 (1x T4)", "system_type": "edge", "target_latency (ns)": 50000000, "target_qps": 20, "task": "object detection", "task2": "object detection", "total_cores": 4, "uid": "1b19b74b6b8e6418", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile": 1, "50.00 percentile latency (ns)": 46129604, "90.00 percentile": 1, "90.00 percentile latency (ns)": 58415443, "95.00 percentile": 1, "95.00 percentile latency (ns)": 58834581, "97.00 percentile": 1, "97.00 percentile latency (ns)": 58971545, "99.00 percentile": 1, "99.00 percentile latency (ns)": 59335388, "99.90 percentile": 1, "99.90 percentile latency (ns)": 59518054, "Max latency (ns)": 60130327, "Mean latency (ns)": 43257660, "Min duration satisfied": "Yes", "Min latency (ns)": 21591213, "Min queries satisfied": "Yes", "Mode": "Performance", "Per-sample latency": "", "Performance constraints satisfied": "Yes", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per query": 8, "Scenario": "multistream", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "16 GB", "accelerator_memory_configuration": "GDDR6", "accelerator_model_name": "NVIDIA T4", "accelerator_on-chip_memories": "", "accelerators_per_node": 1, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.mAP": 20.111, "characteristics.samples_per_query": 8, "characteristics.samples_per_query.normalized_per_core": 8.0, "characteristics.samples_per_query.normalized_per_processor": 8.0, "ck_system": "T4x1_TRT", "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.samples_per_query", "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": "TensorRT 7.2, CUDA 11.0 Update 1", "host_memory_capacity": "128 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 4, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "Intel(R) Xeon(R) D-2123IT CPU @ 2.20GHz", "host_processors_per_node": 1, "host_storage_capacity": "2 TB", "host_storage_type": "M.2", "hw_notes": "ECC off", "informal_model": "ssd-resnet34", "input_data_types": "int8", "key.accuracy": "characteristics.mAP", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 270336, "mlperf_version": 0.7, "normalize_cores": 1, "normalize_processors": 1, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/Lenovo/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/Lenovo/results/T4x1_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.4", "other_software_stack": "TensorRT 7.2, CUDA 11.0 Update 1, cuDNN 8.0.2, DALI 0.25.0", "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": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 8, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "resnet34-ssd1200.pytorch", "status": "available", "submitter": "Lenovo", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Lenovo", "sw_notes": "", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/T4x1_TRT", "system_name": "LenovoSE350 (1x T4)", "system_type": "edge", "target_latency (ns)": 66666666, "target_qps": 15, "task": "object detection", "task2": "object detection", "total_cores": 4, "uid": "23f14332bff50d8a", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" } ]