[ { "50.00 percentile latency (ns)": 33556738, "90.00 percentile latency (ns)": 33702439, "90th percentile latency (ns)": 33702439, "95.00 percentile latency (ns)": 33711105, "97.00 percentile latency (ns)": 33716829, "99.00 percentile latency (ns)": 33729361, "99.90 percentile latency (ns)": 34351113, "Max latency (ns)": 46436260, "Mean latency (ns)": 33597571, "Min duration satisfied": "Yes", "Min latency (ns)": 33489737, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "QPS w/ loadgen overhead": 29.75, "QPS w/o loadgen overhead": 29.76, "Result is": "VALID", "SUT name": "LWIS_Server", "Scenario": "singlestream", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "40GB", "accelerator_memory_configuration": "", "accelerator_model_name": "NVIDIA A100-PCIe", "accelerator_on-chip_memories": "", "accelerators_per_node": 2, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "boot_firmware_version": "", "characteristics.90th_percentile_latency_ms": 33.702439, "characteristics.90th_percentile_latency_ns": 33702439.0, "characteristics.90th_percentile_latency_s": 0.033702439, "characteristics.90th_percentile_latency_us": 33702.439, "characteristics.enhancing tumor": 0.7802, "characteristics.mean": 0.85437, "characteristics.power": 18.442999187743336, "characteristics.power.normalized_per_core": 9.221499593871668, "characteristics.power.normalized_per_processor": 9.221499593871668, "characteristics.tumor core": 0.8697, "characteristics.whole tumor": 0.9132, "ck_system": "NE5260M5_A100-PCIex2_TRT_MaxQ", "ck_used": false, "cooling": "", "dataset": "BraTS 2019", "dataset_link": "https://www.med.upenn.edu/cbica/brats2019/data.html", "dim_x_default": "characteristics.90th_percentile_latency_ms", "dim_y_default": "characteristics.mean", "dim_y_maximize": true, "disk_controllers": "", "disk_drives": "", "division": "closed", "filesystem": "", "formal_model": "3d-unet", "formal_model_accuracy": 99.0, "formal_model_link": "", "framework": "TensorRT 8.0.1, CUDA 11.3", "host_memory_capacity": "1 TB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 28, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "Intel(R) Xeon(R) Gold 6258R", "host_processors_per_node": 2, "host_storage_capacity": "2.9 TB * 2", "host_storage_type": "NVMe SSD", "hw_notes": "", "informal_model": "3d-unet-99", "input_data_types": "fp16", "key.accuracy": "characteristics.mean", "management_firmware_version": "", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1024, "mlperf_version": 1.1, "network_speed_mbit": "", "nics_enabled_connected": "", "nics_enabled_firmware": "", "nics_enabled_os": "", "normalize_cores": 2, "normalize_processors": 2, "note_code": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Inspur/code", "note_details": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Inspur/results/NE5260M5_A100-PCIex2_TRT_MaxQ", "number_of_nodes": 1, "number_of_type_nics_installed": "", "operating_system": "Ubuntu 20.04.2", "other_hardware": "", "other_software_stack": "TensorRT 8.0.1, CUDA 11.3, cuDNN 8.2.1, Driver 470.57.02, DALI 0.31.0", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 16, "power_management": "", "power_supply_details": "", "power_supply_quantity_and_rating_watts": "", "print_timestamps": 0, "problem": false, "qsl_rng_seed": 1624344308455410291, "retraining": "N", "sample_index_rng_seed": 517984244576520566, "samples_per_query": 1, "schedule_rng_seed": 10051496985653635065, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "Inspur", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Inspur", "sw_notes": "", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/NE5260M5_A100-PCIex2_TRT_MaxQ", "system_name": "NE5260M5 (2x A100-PCIe, TensorRT)", "system_type": "edge", "target_latency (ns)": 0, "target_qps": 23.8095, "task": "image segmentation", "task2": "image segmentation", "total_cores": 56, "uid": "47d7439640e39d5f", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 33556738, "90.00 percentile latency (ns)": 33702439, "90th percentile latency (ns)": 33702439, "95.00 percentile latency (ns)": 33711105, "97.00 percentile latency (ns)": 33716829, "99.00 percentile latency (ns)": 33729361, "99.90 percentile latency (ns)": 34351113, "Max latency (ns)": 46436260, "Mean latency (ns)": 33597571, "Min duration satisfied": "Yes", "Min latency (ns)": 33489737, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "QPS w/ loadgen overhead": 29.75, "QPS w/o loadgen overhead": 29.76, "Result is": "VALID", "SUT name": "LWIS_Server", "Scenario": "singlestream", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "40GB", "accelerator_memory_configuration": "", "accelerator_model_name": "NVIDIA A100-PCIe", "accelerator_on-chip_memories": "", "accelerators_per_node": 2, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "boot_firmware_version": "", "characteristics.90th_percentile_latency_ms": 33.702439, "characteristics.90th_percentile_latency_ns": 33702439.0, "characteristics.90th_percentile_latency_s": 0.033702439, "characteristics.90th_percentile_latency_us": 33702.439, "characteristics.enhancing tumor": 0.7802, "characteristics.mean": 0.85437, "characteristics.power": 18.442999187743336, "characteristics.power.normalized_per_core": 9.221499593871668, "characteristics.power.normalized_per_processor": 9.221499593871668, "characteristics.tumor core": 0.8697, "characteristics.whole tumor": 0.9132, "ck_system": "NE5260M5_A100-PCIex2_TRT_MaxQ", "ck_used": false, "cooling": "", "dataset": "BraTS 2019", "dataset_link": "https://www.med.upenn.edu/cbica/brats2019/data.html", "dim_x_default": "characteristics.90th_percentile_latency_ms", "dim_y_default": "characteristics.mean", "dim_y_maximize": true, "disk_controllers": "", "disk_drives": "", "division": "closed", "filesystem": "", "formal_model": "3d-unet", "formal_model_accuracy": 99.9, "formal_model_link": "", "framework": "TensorRT 8.0.1, CUDA 11.3", "host_memory_capacity": "1 TB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 28, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "Intel(R) Xeon(R) Gold 6258R", "host_processors_per_node": 2, "host_storage_capacity": "2.9 TB * 2", "host_storage_type": "NVMe SSD", "hw_notes": "", "informal_model": "3d-unet-99.9", "input_data_types": "fp16", "key.accuracy": "characteristics.mean", "management_firmware_version": "", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1024, "mlperf_version": 1.1, "network_speed_mbit": "", "nics_enabled_connected": "", "nics_enabled_firmware": "", "nics_enabled_os": "", "normalize_cores": 2, "normalize_processors": 2, "note_code": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Inspur/code", "note_details": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Inspur/results/NE5260M5_A100-PCIex2_TRT_MaxQ", "number_of_nodes": 1, "number_of_type_nics_installed": "", "operating_system": "Ubuntu 20.04.2", "other_hardware": "", "other_software_stack": "TensorRT 8.0.1, CUDA 11.3, cuDNN 8.2.1, Driver 470.57.02, DALI 0.31.0", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 16, "power_management": "", "power_supply_details": "", "power_supply_quantity_and_rating_watts": "", "print_timestamps": 0, "problem": false, "qsl_rng_seed": 1624344308455410291, "retraining": "N", "sample_index_rng_seed": 517984244576520566, "samples_per_query": 1, "schedule_rng_seed": 10051496985653635065, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "Inspur", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Inspur", "sw_notes": "", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/NE5260M5_A100-PCIex2_TRT_MaxQ", "system_name": "NE5260M5 (2x A100-PCIe, TensorRT)", "system_type": "edge", "target_latency (ns)": 0, "target_qps": 23.8095, "task": "image segmentation", "task2": "image segmentation", "total_cores": 56, "uid": "5934fa9979043d54", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 33527195, "90.00 percentile latency (ns)": 33647437, "90th percentile latency (ns)": 33647437, "95.00 percentile latency (ns)": 33675510, "97.00 percentile latency (ns)": 33684424, "99.00 percentile latency (ns)": 33697550, "99.90 percentile latency (ns)": 33972305, "Max latency (ns)": 35684968, "Mean latency (ns)": 33537905, "Min duration satisfied": "Yes", "Min latency (ns)": 33461391, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "QPS w/ loadgen overhead": 29.81, "QPS w/o loadgen overhead": 29.82, "Result is": "VALID", "SUT name": "LWIS_Server", "Scenario": "singlestream", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "40GB", "accelerator_memory_configuration": "", "accelerator_model_name": "NVIDIA A100-PCIe", "accelerator_on-chip_memories": "", "accelerators_per_node": 2, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "boot_firmware_version": "", "characteristics.90th_percentile_latency_ms": 33.647437, "characteristics.90th_percentile_latency_ns": 33647437.0, "characteristics.90th_percentile_latency_s": 0.033647437, "characteristics.90th_percentile_latency_us": 33647.437, "characteristics.enhancing tumor": 0.7802, "characteristics.mean": 0.85437, "characteristics.tumor core": 0.8697, "characteristics.whole tumor": 0.9132, "ck_system": "NE5260M5_A100-PCIex2_TRT", "ck_used": false, "cooling": "", "dataset": "BraTS 2019", "dataset_link": "https://www.med.upenn.edu/cbica/brats2019/data.html", "dim_x_default": "characteristics.90th_percentile_latency_ms", "dim_y_default": "characteristics.mean", "dim_y_maximize": true, "disk_controllers": "", "disk_drives": "", "division": "closed", "filesystem": "", "formal_model": "3d-unet", "formal_model_accuracy": 99.0, "formal_model_link": "", "framework": "TensorRT 8.0.1, CUDA 11.3", "host_memory_capacity": "1 TB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 28, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "Intel(R) Xeon(R) Gold 6258R", "host_processors_per_node": 2, "host_storage_capacity": "2.9 TB * 2", "host_storage_type": "NVMe SSD", "hw_notes": "", "informal_model": "3d-unet-99", "input_data_types": "fp16", "key.accuracy": "characteristics.mean", "management_firmware_version": "", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1024, "mlperf_version": 1.1, "network_speed_mbit": "", "nics_enabled_connected": "", "nics_enabled_firmware": "", "nics_enabled_os": "", "normalize_cores": 2, "normalize_processors": 2, "note_code": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Inspur/code", "note_details": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Inspur/results/NE5260M5_A100-PCIex2_TRT", "number_of_nodes": 1, "number_of_type_nics_installed": "", "operating_system": "Ubuntu 20.04.2", "other_hardware": "", "other_software_stack": "TensorRT 8.0.1, CUDA 11.3, cuDNN 8.2.1, Driver 470.57.02, DALI 0.31.0", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 16, "power_management": "", "power_supply_details": "", "power_supply_quantity_and_rating_watts": "", "print_timestamps": 0, "problem": false, "qsl_rng_seed": 1624344308455410291, "retraining": "N", "sample_index_rng_seed": 517984244576520566, "samples_per_query": 1, "schedule_rng_seed": 10051496985653635065, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "Inspur", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Inspur", "sw_notes": "", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/NE5260M5_A100-PCIex2_TRT", "system_name": "NE5260M5 (2x A100-PCIe, TensorRT)", "system_type": "edge", "target_latency (ns)": 0, "target_qps": 23.8095, "task": "image segmentation", "task2": "image segmentation", "total_cores": 56, "uid": "6288686cbaeaca5e", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 33527195, "90.00 percentile latency (ns)": 33647437, "90th percentile latency (ns)": 33647437, "95.00 percentile latency (ns)": 33675510, "97.00 percentile latency (ns)": 33684424, "99.00 percentile latency (ns)": 33697550, "99.90 percentile latency (ns)": 33972305, "Max latency (ns)": 35684968, "Mean latency (ns)": 33537905, "Min duration satisfied": "Yes", "Min latency (ns)": 33461391, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "QPS w/ loadgen overhead": 29.81, "QPS w/o loadgen overhead": 29.82, "Result is": "VALID", "SUT name": "LWIS_Server", "Scenario": "singlestream", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "40GB", "accelerator_memory_configuration": "", "accelerator_model_name": "NVIDIA A100-PCIe", "accelerator_on-chip_memories": "", "accelerators_per_node": 2, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "boot_firmware_version": "", "characteristics.90th_percentile_latency_ms": 33.647437, "characteristics.90th_percentile_latency_ns": 33647437.0, "characteristics.90th_percentile_latency_s": 0.033647437, "characteristics.90th_percentile_latency_us": 33647.437, "characteristics.enhancing tumor": 0.7802, "characteristics.mean": 0.85437, "characteristics.tumor core": 0.8697, "characteristics.whole tumor": 0.9132, "ck_system": "NE5260M5_A100-PCIex2_TRT", "ck_used": false, "cooling": "", "dataset": "BraTS 2019", "dataset_link": "https://www.med.upenn.edu/cbica/brats2019/data.html", "dim_x_default": "characteristics.90th_percentile_latency_ms", "dim_y_default": "characteristics.mean", "dim_y_maximize": true, "disk_controllers": "", "disk_drives": "", "division": "closed", "filesystem": "", "formal_model": "3d-unet", "formal_model_accuracy": 99.9, "formal_model_link": "", "framework": "TensorRT 8.0.1, CUDA 11.3", "host_memory_capacity": "1 TB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 28, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "Intel(R) Xeon(R) Gold 6258R", "host_processors_per_node": 2, "host_storage_capacity": "2.9 TB * 2", "host_storage_type": "NVMe SSD", "hw_notes": "", "informal_model": "3d-unet-99.9", "input_data_types": "fp16", "key.accuracy": "characteristics.mean", "management_firmware_version": "", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1024, "mlperf_version": 1.1, "network_speed_mbit": "", "nics_enabled_connected": "", "nics_enabled_firmware": "", "nics_enabled_os": "", "normalize_cores": 2, "normalize_processors": 2, "note_code": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Inspur/code", "note_details": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Inspur/results/NE5260M5_A100-PCIex2_TRT", "number_of_nodes": 1, "number_of_type_nics_installed": "", "operating_system": "Ubuntu 20.04.2", "other_hardware": "", "other_software_stack": "TensorRT 8.0.1, CUDA 11.3, cuDNN 8.2.1, Driver 470.57.02, DALI 0.31.0", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 16, "power_management": "", "power_supply_details": "", "power_supply_quantity_and_rating_watts": "", "print_timestamps": 0, "problem": false, "qsl_rng_seed": 1624344308455410291, "retraining": "N", "sample_index_rng_seed": 517984244576520566, "samples_per_query": 1, "schedule_rng_seed": 10051496985653635065, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "Inspur", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Inspur", "sw_notes": "", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/NE5260M5_A100-PCIex2_TRT", "system_name": "NE5260M5 (2x A100-PCIe, TensorRT)", "system_type": "edge", "target_latency (ns)": 0, "target_qps": 23.8095, "task": "image segmentation", "task2": "image segmentation", "total_cores": 56, "uid": "43a7bebd44751a27", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 33493200, "90.00 percentile latency (ns)": 33548654, "90th percentile latency (ns)": 33548654, "95.00 percentile latency (ns)": 33617634, "97.00 percentile latency (ns)": 33629861, "99.00 percentile latency (ns)": 33661500, "99.90 percentile latency (ns)": 33946699, "Max latency (ns)": 38206503, "Mean latency (ns)": 33504843, "Min duration satisfied": "Yes", "Min latency (ns)": 33434008, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "QPS w/ loadgen overhead": 29.84, "QPS w/o loadgen overhead": 29.85, "Result is": "VALID", "SUT name": "Triton_Server", "Scenario": "singlestream", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "40GB", "accelerator_memory_configuration": "", "accelerator_model_name": "NVIDIA A100-PCIe", "accelerator_on-chip_memories": "", "accelerators_per_node": 2, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "boot_firmware_version": "", "characteristics.90th_percentile_latency_ms": 33.548654, "characteristics.90th_percentile_latency_ns": 33548654.0, "characteristics.90th_percentile_latency_s": 0.033548654, "characteristics.90th_percentile_latency_us": 33548.654, "characteristics.enhancing tumor": 0.7802, "characteristics.mean": 0.85437, "characteristics.tumor core": 0.8697, "characteristics.whole tumor": 0.9132, "ck_system": "NE5260M5_A100-PCIex2_TRT_Triton", "ck_used": false, "cooling": "", "dataset": "BraTS 2019", "dataset_link": "https://www.med.upenn.edu/cbica/brats2019/data.html", "dim_x_default": "characteristics.90th_percentile_latency_ms", "dim_y_default": "characteristics.mean", "dim_y_maximize": true, "disk_controllers": "", "disk_drives": "", "division": "closed", "filesystem": "", "formal_model": "3d-unet", "formal_model_accuracy": 99.0, "formal_model_link": "", "framework": "TensorRT 8.0.1, CUDA 11.3", "host_memory_capacity": "1 TB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 28, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "Intel(R) Xeon(R) Gold 6258R", "host_processors_per_node": 2, "host_storage_capacity": "2.9 TB * 2", "host_storage_type": "NVMe SSD", "hw_notes": "", "informal_model": "3d-unet-99", "input_data_types": "fp16", "key.accuracy": "characteristics.mean", "management_firmware_version": "", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1024, "mlperf_version": 1.1, "network_speed_mbit": "", "nics_enabled_connected": "", "nics_enabled_firmware": "", "nics_enabled_os": "", "normalize_cores": 2, "normalize_processors": 2, "note_code": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Inspur/code", "note_details": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Inspur/results/NE5260M5_A100-PCIex2_TRT_Triton", "number_of_nodes": 1, "number_of_type_nics_installed": "", "operating_system": "Ubuntu 20.04.2", "other_hardware": "", "other_software_stack": "TensorRT 8.0.1, CUDA 11.3, cuDNN 8.2.1, Driver 470.57.02, DALI 0.31.0", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 16, "power_management": "", "power_supply_details": "", "power_supply_quantity_and_rating_watts": "", "print_timestamps": 0, "problem": false, "qsl_rng_seed": 1624344308455410291, "retraining": "N", "sample_index_rng_seed": 517984244576520566, "samples_per_query": 1, "schedule_rng_seed": 10051496985653635065, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "Inspur", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Inspur", "sw_notes": "", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/NE5260M5_A100-PCIex2_TRT_Triton", "system_name": "NE5260M5 (2x A100-PCIe, TensorRT)", "system_type": "edge", "target_latency (ns)": 0, "target_qps": 23.8095, "task": "image segmentation", "task2": "image segmentation", "total_cores": 56, "uid": "e179b4099b3adae8", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 33493200, "90.00 percentile latency (ns)": 33548654, "90th percentile latency (ns)": 33548654, "95.00 percentile latency (ns)": 33617634, "97.00 percentile latency (ns)": 33629861, "99.00 percentile latency (ns)": 33661500, "99.90 percentile latency (ns)": 33946699, "Max latency (ns)": 38206503, "Mean latency (ns)": 33504843, "Min duration satisfied": "Yes", "Min latency (ns)": 33434008, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "QPS w/ loadgen overhead": 29.84, "QPS w/o loadgen overhead": 29.85, "Result is": "VALID", "SUT name": "Triton_Server", "Scenario": "singlestream", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "40GB", "accelerator_memory_configuration": "", "accelerator_model_name": "NVIDIA A100-PCIe", "accelerator_on-chip_memories": "", "accelerators_per_node": 2, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "boot_firmware_version": "", "characteristics.90th_percentile_latency_ms": 33.548654, "characteristics.90th_percentile_latency_ns": 33548654.0, "characteristics.90th_percentile_latency_s": 0.033548654, "characteristics.90th_percentile_latency_us": 33548.654, "characteristics.enhancing tumor": 0.7802, "characteristics.mean": 0.85437, "characteristics.tumor core": 0.8697, "characteristics.whole tumor": 0.9132, "ck_system": "NE5260M5_A100-PCIex2_TRT_Triton", "ck_used": false, "cooling": "", "dataset": "BraTS 2019", "dataset_link": "https://www.med.upenn.edu/cbica/brats2019/data.html", "dim_x_default": "characteristics.90th_percentile_latency_ms", "dim_y_default": "characteristics.mean", "dim_y_maximize": true, "disk_controllers": "", "disk_drives": "", "division": "closed", "filesystem": "", "formal_model": "3d-unet", "formal_model_accuracy": 99.9, "formal_model_link": "", "framework": "TensorRT 8.0.1, CUDA 11.3", "host_memory_capacity": "1 TB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 28, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "Intel(R) Xeon(R) Gold 6258R", "host_processors_per_node": 2, "host_storage_capacity": "2.9 TB * 2", "host_storage_type": "NVMe SSD", "hw_notes": "", "informal_model": "3d-unet-99.9", "input_data_types": "fp16", "key.accuracy": "characteristics.mean", "management_firmware_version": "", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1024, "mlperf_version": 1.1, "network_speed_mbit": "", "nics_enabled_connected": "", "nics_enabled_firmware": "", "nics_enabled_os": "", "normalize_cores": 2, "normalize_processors": 2, "note_code": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Inspur/code", "note_details": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Inspur/results/NE5260M5_A100-PCIex2_TRT_Triton", "number_of_nodes": 1, "number_of_type_nics_installed": "", "operating_system": "Ubuntu 20.04.2", "other_hardware": "", "other_software_stack": "TensorRT 8.0.1, CUDA 11.3, cuDNN 8.2.1, Driver 470.57.02, DALI 0.31.0", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 16, "power_management": "", "power_supply_details": "", "power_supply_quantity_and_rating_watts": "", "print_timestamps": 0, "problem": false, "qsl_rng_seed": 1624344308455410291, "retraining": "N", "sample_index_rng_seed": 517984244576520566, "samples_per_query": 1, "schedule_rng_seed": 10051496985653635065, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "Inspur", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Inspur", "sw_notes": "", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/NE5260M5_A100-PCIex2_TRT_Triton", "system_name": "NE5260M5 (2x A100-PCIe, TensorRT)", "system_type": "edge", "target_latency (ns)": 0, "target_qps": 23.8095, "task": "image segmentation", "task2": "image segmentation", "total_cores": 56, "uid": "4242a4c8a97595c8", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 16665036, "90.00 percentile latency (ns)": 16680964, "90th percentile latency (ns)": 16680964, "95.00 percentile latency (ns)": 16686097, "97.00 percentile latency (ns)": 16689558, "99.00 percentile latency (ns)": 16701804, "99.90 percentile latency (ns)": 17264158, "Max latency (ns)": 21257495, "Mean latency (ns)": 16668673, "Min duration satisfied": "Yes", "Min latency (ns)": 16628425, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "QPS w/ loadgen overhead": 59.98, "QPS w/o loadgen overhead": 59.99, "Result is": "VALID", "SUT name": "LWIS_Server", "Scenario": "singlestream", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "80 GB", "accelerator_memory_configuration": "", "accelerator_model_name": "NVIDIA A100-SXM-80GB", "accelerator_on-chip_memories": "", "accelerators_per_node": 1, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "boot_firmware_version": "", "characteristics.90th_percentile_latency_ms": 16.680964, "characteristics.90th_percentile_latency_ns": 16680964.0, "characteristics.90th_percentile_latency_s": 0.016680964, "characteristics.90th_percentile_latency_us": 16680.964, "characteristics.enhancing tumor": 0.7789, "characteristics.mean": 0.85403, "characteristics.tumor core": 0.869, "characteristics.whole tumor": 0.9142, "ck_system": "NF5688M6_A100-SXM-80GBx1_TRT", "ck_used": false, "cooling": "", "dataset": "BraTS 2019", "dataset_link": "https://www.med.upenn.edu/cbica/brats2019/data.html", "dim_x_default": "characteristics.90th_percentile_latency_ms", "dim_y_default": "characteristics.mean", "dim_y_maximize": true, "disk_controllers": "", "disk_drives": "", "division": "closed", "filesystem": "", "formal_model": "3d-unet", "formal_model_accuracy": 99.0, "formal_model_link": "", "framework": "TensorRT 8.0.1, CUDA 11.3", "host_memory_capacity": "2 TB", "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 8358", "host_processors_per_node": 2, "host_storage_capacity": "28 TB + 448GB*2", "host_storage_type": "NVMe SSD + M.2", "hw_notes": "NVIDIA A100-SXM-80GB-500W", "informal_model": "3d-unet-99", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "management_firmware_version": "", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1024, "mlperf_version": 1.1, "network_speed_mbit": "", "nics_enabled_connected": "", "nics_enabled_firmware": "", "nics_enabled_os": "", "normalize_cores": 1, "normalize_processors": 1, "note_code": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Inspur/code", "note_details": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Inspur/results/NF5688M6_A100-SXM-80GBx1_TRT", "number_of_nodes": 1, "number_of_type_nics_installed": "", "operating_system": "Ubuntu 20.04.2", "other_hardware": "", "other_software_stack": "TensorRT 8.0.1, CUDA 11.3, cuDNN 8.2.1, Driver 470.57.02, DALI 1.0.0", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 16, "power_management": "", "power_supply_details": "", "power_supply_quantity_and_rating_watts": "", "print_timestamps": 0, "problem": false, "qsl_rng_seed": 1624344308455410291, "retraining": "N", "sample_index_rng_seed": 517984244576520566, "samples_per_query": 1, "schedule_rng_seed": 10051496985653635065, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "Inspur", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Inspur", "sw_notes": "", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/NF5688M6_A100-SXM-80GBx1_TRT", "system_name": "NF5688M6 (8x A100-SXM-80GB, TensorRT)", "system_type": "edge", "target_latency (ns)": 0, "target_qps": 40, "task": "image segmentation", "task2": "image segmentation", "total_cores": 64, "uid": "b353c5deb4075a50", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 16665036, "90.00 percentile latency (ns)": 16680964, "90th percentile latency (ns)": 16680964, "95.00 percentile latency (ns)": 16686097, "97.00 percentile latency (ns)": 16689558, "99.00 percentile latency (ns)": 16701804, "99.90 percentile latency (ns)": 17264158, "Max latency (ns)": 21257495, "Mean latency (ns)": 16668673, "Min duration satisfied": "Yes", "Min latency (ns)": 16628425, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "QPS w/ loadgen overhead": 59.98, "QPS w/o loadgen overhead": 59.99, "Result is": "VALID", "SUT name": "LWIS_Server", "Scenario": "singlestream", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "80 GB", "accelerator_memory_configuration": "", "accelerator_model_name": "NVIDIA A100-SXM-80GB", "accelerator_on-chip_memories": "", "accelerators_per_node": 1, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "boot_firmware_version": "", "characteristics.90th_percentile_latency_ms": 16.680964, "characteristics.90th_percentile_latency_ns": 16680964.0, "characteristics.90th_percentile_latency_s": 0.016680964, "characteristics.90th_percentile_latency_us": 16680.964, "characteristics.enhancing tumor": 0.7789, "characteristics.mean": 0.85403, "characteristics.tumor core": 0.869, "characteristics.whole tumor": 0.9142, "ck_system": "NF5688M6_A100-SXM-80GBx1_TRT", "ck_used": false, "cooling": "", "dataset": "BraTS 2019", "dataset_link": "https://www.med.upenn.edu/cbica/brats2019/data.html", "dim_x_default": "characteristics.90th_percentile_latency_ms", "dim_y_default": "characteristics.mean", "dim_y_maximize": true, "disk_controllers": "", "disk_drives": "", "division": "closed", "filesystem": "", "formal_model": "3d-unet", "formal_model_accuracy": 99.9, "formal_model_link": "", "framework": "TensorRT 8.0.1, CUDA 11.3", "host_memory_capacity": "2 TB", "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 8358", "host_processors_per_node": 2, "host_storage_capacity": "28 TB + 448GB*2", "host_storage_type": "NVMe SSD + M.2", "hw_notes": "NVIDIA A100-SXM-80GB-500W", "informal_model": "3d-unet-99.9", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "management_firmware_version": "", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1024, "mlperf_version": 1.1, "network_speed_mbit": "", "nics_enabled_connected": "", "nics_enabled_firmware": "", "nics_enabled_os": "", "normalize_cores": 1, "normalize_processors": 1, "note_code": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Inspur/code", "note_details": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Inspur/results/NF5688M6_A100-SXM-80GBx1_TRT", "number_of_nodes": 1, "number_of_type_nics_installed": "", "operating_system": "Ubuntu 20.04.2", "other_hardware": "", "other_software_stack": "TensorRT 8.0.1, CUDA 11.3, cuDNN 8.2.1, Driver 470.57.02, DALI 1.0.0", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 16, "power_management": "", "power_supply_details": "", "power_supply_quantity_and_rating_watts": "", "print_timestamps": 0, "problem": false, "qsl_rng_seed": 1624344308455410291, "retraining": "N", "sample_index_rng_seed": 517984244576520566, "samples_per_query": 1, "schedule_rng_seed": 10051496985653635065, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "Inspur", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Inspur", "sw_notes": "", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/NF5688M6_A100-SXM-80GBx1_TRT", "system_name": "NF5688M6 (8x A100-SXM-80GB, TensorRT)", "system_type": "edge", "target_latency (ns)": 0, "target_qps": 40, "task": "image segmentation", "task2": "image segmentation", "total_cores": 64, "uid": "4d85bc42cd88b993", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 16816287, "90.00 percentile latency (ns)": 16864805, "90th percentile latency (ns)": 16864805, "95.00 percentile latency (ns)": 17087849, "97.00 percentile latency (ns)": 17155303, "99.00 percentile latency (ns)": 17289343, "99.90 percentile latency (ns)": 18139269, "Max latency (ns)": 21344258, "Mean latency (ns)": 16843354, "Min duration satisfied": "Yes", "Min latency (ns)": 16725188, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "QPS w/ loadgen overhead": 59.25, "QPS w/o loadgen overhead": 59.37, "Result is": "VALID", "SUT name": "LWIS_Server", "Scenario": "singlestream", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "80 GB", "accelerator_memory_configuration": "", "accelerator_model_name": "NVIDIA A100-SXM-80GB", "accelerator_on-chip_memories": "", "accelerators_per_node": 1, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "boot_firmware_version": "", "characteristics.90th_percentile_latency_ms": 16.864805, "characteristics.90th_percentile_latency_ns": 16864805.0, "characteristics.90th_percentile_latency_s": 0.016864805, "characteristics.90th_percentile_latency_us": 16864.805, "characteristics.enhancing tumor": 0.7789, "characteristics.mean": 0.85403, "characteristics.tumor core": 0.869, "characteristics.whole tumor": 0.9142, "ck_system": "NF5488A5_A100-SXM-80GBx1_TRT", "ck_used": false, "cooling": "Liquid", "dataset": "BraTS 2019", "dataset_link": "https://www.med.upenn.edu/cbica/brats2019/data.html", "dim_x_default": "characteristics.90th_percentile_latency_ms", "dim_y_default": "characteristics.mean", "dim_y_maximize": true, "disk_controllers": "", "disk_drives": "", "division": "closed", "filesystem": "", "formal_model": "3d-unet", "formal_model_accuracy": 99.0, "formal_model_link": "", "framework": "TensorRT 8.0.1, CUDA 11.3", "host_memory_capacity": "2 TB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 64, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "AMD EPYC 7742", "host_processors_per_node": 2, "host_storage_capacity": "11.7 TB + 448GB*2", "host_storage_type": "NVMe SSD + M.2", "hw_notes": "NVIDIA A100-SXM-80GB-500W", "informal_model": "3d-unet-99", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "management_firmware_version": "", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1024, "mlperf_version": 1.1, "network_speed_mbit": "", "nics_enabled_connected": "", "nics_enabled_firmware": "", "nics_enabled_os": "", "normalize_cores": 1, "normalize_processors": 1, "note_code": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Inspur/code", "note_details": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Inspur/results/NF5488A5_A100-SXM-80GBx1_TRT", "number_of_nodes": 1, "number_of_type_nics_installed": "", "operating_system": "Ubuntu 20.04.2", "other_hardware": "", "other_software_stack": "TensorRT 8.0.1, CUDA 11.3, cuDNN 8.2.1, Driver 470.57.02, DALI 0.31.0", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 16, "power_management": "", "power_supply_details": "", "power_supply_quantity_and_rating_watts": "", "print_timestamps": 0, "problem": false, "qsl_rng_seed": 1624344308455410291, "retraining": "N", "sample_index_rng_seed": 517984244576520566, "samples_per_query": 1, "schedule_rng_seed": 10051496985653635065, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "Inspur", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Inspur", "sw_notes": "", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/NF5488A5_A100-SXM-80GBx1_TRT", "system_name": "NF5488A5 (1x A100-SXM-80GB, TensorRT)", "system_type": "edge", "target_latency (ns)": 0, "target_qps": 40, "task": "image segmentation", "task2": "image segmentation", "total_cores": 128, "uid": "2611248edbef75c1", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 16816287, "90.00 percentile latency (ns)": 16864805, "90th percentile latency (ns)": 16864805, "95.00 percentile latency (ns)": 17087849, "97.00 percentile latency (ns)": 17155303, "99.00 percentile latency (ns)": 17289343, "99.90 percentile latency (ns)": 18139269, "Max latency (ns)": 21344258, "Mean latency (ns)": 16843354, "Min duration satisfied": "Yes", "Min latency (ns)": 16725188, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "QPS w/ loadgen overhead": 59.25, "QPS w/o loadgen overhead": 59.37, "Result is": "VALID", "SUT name": "LWIS_Server", "Scenario": "singlestream", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "80 GB", "accelerator_memory_configuration": "", "accelerator_model_name": "NVIDIA A100-SXM-80GB", "accelerator_on-chip_memories": "", "accelerators_per_node": 1, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "boot_firmware_version": "", "characteristics.90th_percentile_latency_ms": 16.864805, "characteristics.90th_percentile_latency_ns": 16864805.0, "characteristics.90th_percentile_latency_s": 0.016864805, "characteristics.90th_percentile_latency_us": 16864.805, "characteristics.enhancing tumor": 0.7789, "characteristics.mean": 0.85403, "characteristics.tumor core": 0.869, "characteristics.whole tumor": 0.9142, "ck_system": "NF5488A5_A100-SXM-80GBx1_TRT", "ck_used": false, "cooling": "Liquid", "dataset": "BraTS 2019", "dataset_link": "https://www.med.upenn.edu/cbica/brats2019/data.html", "dim_x_default": "characteristics.90th_percentile_latency_ms", "dim_y_default": "characteristics.mean", "dim_y_maximize": true, "disk_controllers": "", "disk_drives": "", "division": "closed", "filesystem": "", "formal_model": "3d-unet", "formal_model_accuracy": 99.9, "formal_model_link": "", "framework": "TensorRT 8.0.1, CUDA 11.3", "host_memory_capacity": "2 TB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 64, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "AMD EPYC 7742", "host_processors_per_node": 2, "host_storage_capacity": "11.7 TB + 448GB*2", "host_storage_type": "NVMe SSD + M.2", "hw_notes": "NVIDIA A100-SXM-80GB-500W", "informal_model": "3d-unet-99.9", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "management_firmware_version": "", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1024, "mlperf_version": 1.1, "network_speed_mbit": "", "nics_enabled_connected": "", "nics_enabled_firmware": "", "nics_enabled_os": "", "normalize_cores": 1, "normalize_processors": 1, "note_code": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Inspur/code", "note_details": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Inspur/results/NF5488A5_A100-SXM-80GBx1_TRT", "number_of_nodes": 1, "number_of_type_nics_installed": "", "operating_system": "Ubuntu 20.04.2", "other_hardware": "", "other_software_stack": "TensorRT 8.0.1, CUDA 11.3, cuDNN 8.2.1, Driver 470.57.02, DALI 0.31.0", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 16, "power_management": "", "power_supply_details": "", "power_supply_quantity_and_rating_watts": "", "print_timestamps": 0, "problem": false, "qsl_rng_seed": 1624344308455410291, "retraining": "N", "sample_index_rng_seed": 517984244576520566, "samples_per_query": 1, "schedule_rng_seed": 10051496985653635065, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "Inspur", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Inspur", "sw_notes": "", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/NF5488A5_A100-SXM-80GBx1_TRT", "system_name": "NF5488A5 (1x A100-SXM-80GB, TensorRT)", "system_type": "edge", "target_latency (ns)": 0, "target_qps": 40, "task": "image segmentation", "task2": "image segmentation", "total_cores": 128, "uid": "1c2412489e9ca7e3", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 34875525, "90.00 percentile latency (ns)": 35036544, "90th percentile latency (ns)": 35036544, "95.00 percentile latency (ns)": 35057076, "97.00 percentile latency (ns)": 35064954, "99.00 percentile latency (ns)": 35077447, "99.90 percentile latency (ns)": 35105830, "Max latency (ns)": 35780367, "Mean latency (ns)": 34883111, "Min duration satisfied": "Yes", "Min latency (ns)": 34738007, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "QPS w/ loadgen overhead": 28.66, "QPS w/o loadgen overhead": 28.67, "Result is": "VALID", "SUT name": "LWIS_Server", "Scenario": "singlestream", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "40GB", "accelerator_memory_configuration": "HBM2", "accelerator_model_name": "NVIDIA A100-PCIe", "accelerator_on-chip_memories": "", "accelerators_per_node": 2, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.90th_percentile_latency_ms": 35.036544, "characteristics.90th_percentile_latency_ns": 35036544.0, "characteristics.90th_percentile_latency_s": 0.035036544, "characteristics.90th_percentile_latency_us": 35036.544, "characteristics.enhancing tumor": 0.7802, "characteristics.mean": 0.85387, "characteristics.tumor core": 0.8683, "characteristics.whole tumor": 0.9131, "ck_system": "NE5260M5_A100-PCIex2_TRT", "ck_used": false, "cooling": "", "dataset": "BraTS 2019", "dataset_link": "https://www.med.upenn.edu/cbica/brats2019/data.html", "dim_x_default": "characteristics.90th_percentile_latency_ms", "dim_y_default": "characteristics.mean", "dim_y_maximize": true, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.0, "formal_model_link": "", "framework": "TensorRT 7.2, CUDA Version: 11.2", "host_memory_capacity": "1 TB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 28, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "Intel(R) Xeon(R) Gold 6258R CPU @ 2.70GHz", "host_processors_per_node": 2, "host_storage_capacity": "5 TB", "host_storage_type": "NVMe SSD", "hw_notes": "", "informal_model": "3d-unet-99", "input_data_types": "fp16", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1024, "mlperf_version": 1.0, "normalize_cores": 2, "normalize_processors": 2, "note_code": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/Inspur/code", "note_details": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/Inspur/results/NE5260M5_A100-PCIex2_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.2", "other_software_stack": "TensorRT 7.2, CUDA 11.2, cuDNN 8.0.2, DALI 0.25.0", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 16, "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": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "Inspur", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Inspur", "sw_notes": "", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/NE5260M5_A100-PCIex2_TRT", "system_name": "NE5260M5", "system_type": "edge", "target_latency (ns)": 0, "target_qps": 23.8095, "task": "image segmentation", "task2": "image segmentation", "total_cores": 56, "uid": "7624aae544c8da6e", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 34875525, "90.00 percentile latency (ns)": 35036544, "90th percentile latency (ns)": 35036544, "95.00 percentile latency (ns)": 35057076, "97.00 percentile latency (ns)": 35064954, "99.00 percentile latency (ns)": 35077447, "99.90 percentile latency (ns)": 35105830, "Max latency (ns)": 35780367, "Mean latency (ns)": 34883111, "Min duration satisfied": "Yes", "Min latency (ns)": 34738007, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "QPS w/ loadgen overhead": 28.66, "QPS w/o loadgen overhead": 28.67, "Result is": "VALID", "SUT name": "LWIS_Server", "Scenario": "singlestream", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "40GB", "accelerator_memory_configuration": "HBM2", "accelerator_model_name": "NVIDIA A100-PCIe", "accelerator_on-chip_memories": "", "accelerators_per_node": 2, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.90th_percentile_latency_ms": 35.036544, "characteristics.90th_percentile_latency_ns": 35036544.0, "characteristics.90th_percentile_latency_s": 0.035036544, "characteristics.90th_percentile_latency_us": 35036.544, "characteristics.enhancing tumor": 0.7802, "characteristics.mean": 0.85387, "characteristics.tumor core": 0.8683, "characteristics.whole tumor": 0.9131, "ck_system": "NE5260M5_A100-PCIex2_TRT", "ck_used": false, "cooling": "", "dataset": "BraTS 2019", "dataset_link": "https://www.med.upenn.edu/cbica/brats2019/data.html", "dim_x_default": "characteristics.90th_percentile_latency_ms", "dim_y_default": "characteristics.mean", "dim_y_maximize": true, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.9, "formal_model_link": "", "framework": "TensorRT 7.2, CUDA Version: 11.2", "host_memory_capacity": "1 TB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 28, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "Intel(R) Xeon(R) Gold 6258R CPU @ 2.70GHz", "host_processors_per_node": 2, "host_storage_capacity": "5 TB", "host_storage_type": "NVMe SSD", "hw_notes": "", "informal_model": "3d-unet-99.9", "input_data_types": "fp16", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1024, "mlperf_version": 1.0, "normalize_cores": 2, "normalize_processors": 2, "note_code": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/Inspur/code", "note_details": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/Inspur/results/NE5260M5_A100-PCIex2_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.2", "other_software_stack": "TensorRT 7.2, CUDA 11.2, cuDNN 8.0.2, DALI 0.25.0", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 16, "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": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "Inspur", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Inspur", "sw_notes": "", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/NE5260M5_A100-PCIex2_TRT", "system_name": "NE5260M5", "system_type": "edge", "target_latency (ns)": 0, "target_qps": 23.8095, "task": "image segmentation", "task2": "image segmentation", "total_cores": 56, "uid": "3e94692ade67ada4", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" } ]