[ { "50.00 percentile latency (ns)": 8653189, "90.00 percentile latency (ns)": 8729209, "90th percentile latency (ns)": 8729209, "95.00 percentile latency (ns)": 8745035, "97.00 percentile latency (ns)": 8760578, "99.00 percentile latency (ns)": 8837594, "99.90 percentile latency (ns)": 8897330, "Max latency (ns)": 9965701, "Mean latency (ns)": 8632268, "Min duration satisfied": "Yes", "Min latency (ns)": 6187126, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "QPS w/ loadgen overhead": 115.05, "QPS w/o loadgen overhead": 115.84, "Result is": "VALID", "SUT name": "LWIS_Server", "Scenario": "singlestream", "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.90th_percentile_latency_ms": 8.729209, "characteristics.90th_percentile_latency_ns": 8729209.0, "characteristics.90th_percentile_latency_s": 0.008729209, "characteristics.90th_percentile_latency_us": 8729.209, "characteristics.mAP": 20.111, "ck_system": "XE2420_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.90th_percentile_latency_ms", "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/octoml/mlops/tree/main/package, https://github.com/ctuning/ai/tree/main/package", "framework": "TensorRT 7.2.3, CUDA 11.1", "host_memory_capacity": "255 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 22, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "Intel(R) Xeon(R) Gold 6238 CPU @ 2.10GHz", "host_processors_per_node": 2, "host_storage_capacity": "1 TB", "host_storage_type": "NVMe SSD", "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": 1024, "mlperf_version": "v1.0", "normalize_cores": 1, "normalize_processors": 1, "note_code": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/DellEMC/results/XE2420_T4x1_TRT", "number_of_nodes": 1, "operating_system": "CentOS 8.2.2004", "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": 1, "schedule_rng_seed": 3507442325620259414, "starting_weights_filename": "resnet34-ssd1200.pytorch", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "sw_notes": "", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/XE2420_T4x1_TRT", "system_name": "Dell EMC PowerEdge XE2420 (1x T4)", "system_type": "edge", "target_latency (ns)": 0, "target_qps": 128.916, "task": "object detection", "task2": "object detection", "total_cores": 44, "uid": "d48169859b587602", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 6760214, "90.00 percentile latency (ns)": 6795877, "90th percentile latency (ns)": 6795877, "95.00 percentile latency (ns)": 6804555, "97.00 percentile latency (ns)": 6809317, "99.00 percentile latency (ns)": 6819674, "99.90 percentile latency (ns)": 6863932, "Max latency (ns)": 11421624, "Mean latency (ns)": 6756063, "Min duration satisfied": "Yes", "Min latency (ns)": 6522523, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "QPS w/ loadgen overhead": 146.55, "QPS w/o loadgen overhead": 148.02, "Result is": "VALID", "SUT name": "LWIS_Server", "Scenario": "singlestream", "accelerator_cooling_type": "passive", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "16GB", "accelerator_memory_configuration": "GDDR6", "accelerator_model_name": "NVIDIA T4", "accelerator_on-chip_memories": "", "accelerators_per_node": 4, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "boot_firmware_version": "1.1.3", "characteristics.90th_percentile_latency_ms": 6.795877, "characteristics.90th_percentile_latency_ns": 6795877.0, "characteristics.90th_percentile_latency_s": 0.006795877, "characteristics.90th_percentile_latency_us": 6795.877, "characteristics.mAP": 20.111, "characteristics.power": 4.807602814037434, "characteristics.power.normalized_per_core": 1.2019007035093585, "characteristics.power.normalized_per_processor": 1.2019007035093585, "ck_system": "XE2420_edge_T4x4_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.90th_percentile_latency_ms", "dim_y_default": "characteristics.mAP", "dim_y_maximize": true, "division": "closed", "filesystem": "ext3/ext4", "formal_model": "ssd-mobilenet", "formal_model_accuracy": 99.0, "formal_model_link": "https://github.com/octoml/mlops/tree/main/package, https://github.com/ctuning/ai/tree/main/package", "framework": "TensorRT 7.2.3, CUDA 11.1", "host_cooling_type": "passive", "host_memory_capacity": "384 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 24, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "Intel(R) Xeon(R) Gold 6252 CPU @ 2.10GHz", "host_processors_per_node": 2, "host_storage_capacity": "4 TB", "host_storage_type": "NVMe SSD", "hw_notes": "ECC on", "informal_model": "ssd-resnet34", "input_data_types": "int8", "key.accuracy": "characteristics.mAP", "management_firmware_version": "4.40.10.00", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1024, "mlperf_version": "v1.0", "normalize_cores": 4, "normalize_processors": 4, "note_code": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/DellEMC/results/XE2420_edge_T4x4_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, "power_settings": "closed/DellEMC/measurements/XE2420_edge_T4x4_TRT/XE2420_T4x4_power_settings.adoc", "print_timestamps": 0, "problem": false, "psu_details": "2x2000W", "qsl_rng_seed": 7322528924094909334, "retraining": "N", "sample_index_rng_seed": 1570999273408051088, "samples_per_query": 1, "schedule_rng_seed": 3507442325620259414, "starting_weights_filename": "resnet34-ssd1200.pytorch", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "sw_notes": "", "system_cooling_type": "air", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/XE2420_edge_T4x4_TRT", "system_name": "Dell EMC PowerEdge XE2420 (4x T4)", "system_type": "edge", "target_latency (ns)": 0, "target_qps": 128.916, "task": "object detection", "task2": "object detection", "total_cores": 48, "uid": "524bfca4280e67e0", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 1260203, "90.00 percentile latency (ns)": 1284188, "90th percentile latency (ns)": 1284188, "95.00 percentile latency (ns)": 1293533, "97.00 percentile latency (ns)": 1299284, "99.00 percentile latency (ns)": 1313758, "99.90 percentile latency (ns)": 1336337, "Max latency (ns)": 4318232, "Mean latency (ns)": 1263461, "Min duration satisfied": "Yes", "Min latency (ns)": 1226566, "Min queries satisfied": "Yes", "Mode": "Performance", "QPS w/ loadgen overhead": 789.47, "QPS w/o loadgen overhead": 791.48, "Result is": "VALID", "SUT name": "XLNX_AI", "Scenario": "singlestream", "accelerator_memory_capacity": "16 GB per accelerator", "accelerator_memory_configuration": "HBM", "accelerator_model_name": "xilinx_u280_v3me", "accelerators_per_node": 1, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.90th_percentile_latency_ms": 1.284188, "characteristics.90th_percentile_latency_ns": 1284188.0, "characteristics.90th_percentile_latency_s": 0.001284188, "characteristics.90th_percentile_latency_us": 1284.188, "characteristics.mAP": 23.412, "ck_system": "R740xd_u280_v3me", "ck_used": false, "dataset": "COCO 2017 (300x300)", "dataset_link": "https://github.com/ctuning/ck/blob/master/docs/mlperf-automation/datasets/coco2017.md", "dim_x_default": "characteristics.90th_percentile_latency_ms", "dim_y_default": "characteristics.mAP", "dim_y_maximize": true, "division": "open", "formal_model": "ssd-mobilenet", "formal_model_accuracy": 99.0, "formal_model_link": "https://github.com/octoml/mlops/tree/main/package, https://github.com/ctuning/ai/tree/main/package", "framework": "TF", "host_memory_capacity": "7680 GB", "host_processor_core_count": 28, "host_processor_model_name": "Intel Xeon Platinum 8276", "host_processors_per_node": 2, "host_storage_capacity": "21.42 GB", "host_storage_type": "HDD", "informal_model": "ssd-mobilenet", "input_data_types": "", "key.accuracy": "characteristics.mAP", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 1024, "mlperf_version": "v0.7", "normalize_cores": 1, "normalize_processors": 1, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/open/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/open/DellEMC/results/R740xd_u280_v3me", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04", "other_software_stack": "Xilinx Vitis Runtime", "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": 12786827339337101903, "retraining": "", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 1, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/R740xd_u280_v3me", "system_name": "R740xd_u280_v3me", "system_type": "edge", "target_latency (ns)": 0, "target_qps": 1000, "task": "object detection", "task2": "object detection", "total_cores": 56, "uid": "fef74ab44d384e55", "use_accelerator": true, "weight_data_types": "", "weight_transformations": "" } ]