[
  {
    "50.00 percentile latency (ns)": 367430046108,
    "90.00 percentile latency (ns)": 661745946290,
    "95.00 percentile latency (ns)": 698683950949,
    "97.00 percentile latency (ns)": 713535995182,
    "99.00 percentile latency (ns)": 728161362704,
    "99.90 percentile latency (ns)": 734727538670,
    "Max latency (ns)": 735455206945,
    "Mean latency (ns)": 367283572012,
    "Min duration satisfied": "Yes",
    "Min latency (ns)": 74978274,
    "Min queries satisfied": "Yes",
    "Mode": "PerformanceOnly",
    "Result is": "VALID",
    "SUT name": "Triton_Server",
    "Samples per second": 215.377,
    "Scenario": "offline",
    "accelerator_frequency": "",
    "accelerator_host_interconnect": "",
    "accelerator_interconnect": "",
    "accelerator_interconnect_topology": "",
    "accelerator_memory_capacity": "40 GB",
    "accelerator_memory_configuration": "HBM2e",
    "accelerator_model_name": "NVIDIA A100-SXM4-40GB",
    "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": "",
    "characteristics.enhancing tumor": 0.7798,
    "characteristics.mean": 0.8543,
    "characteristics.samples_per_second": 215.377,
    "characteristics.samples_per_second.normalized_per_core": 53.84425,
    "characteristics.samples_per_second.normalized_per_processor": 53.84425,
    "characteristics.tumor core": 0.8691,
    "characteristics.whole tumor": 0.914,
    "ck_system": "HPE-ProLiant-XL645d-Gen10-Plus_A100-SXM4-40GBx4_TRT_Triton",
    "ck_used": true,
    "cooling": "",
    "dataset": "BraTS 2019",
    "dataset_link": "https://www.med.upenn.edu/cbica/brats2019/data.html",
    "dim_x_default": "characteristics.samples_per_second",
    "dim_x_maximize": true,
    "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 7.2.3, CUDA 11.1",
    "host_memory_capacity": "512 GB",
    "host_memory_configuration": "",
    "host_networking": "",
    "host_networking_topology": "",
    "host_processor_caches": "",
    "host_processor_core_count": 120,
    "host_processor_frequency": "",
    "host_processor_interconnect": "",
    "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor",
    "host_processors_per_node": 1,
    "host_storage_capacity": "3.2 TB",
    "host_storage_type": "SSD",
    "hw_notes": "",
    "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": 1,
    "mlperf_version": 1.0,
    "network_speed_mbit": "",
    "nics_enabled_connected": "",
    "nics_enabled_firmware": "",
    "nics_enabled_os": "",
    "normalize_cores": 4,
    "normalize_processors": 4,
    "note_code": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/HPE/code",
    "note_details": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/HPE/results/HPE-ProLiant-XL645d-Gen10-Plus_A100-SXM4-40GBx4_TRT_Triton",
    "number_of_nodes": 1,
    "number_of_type_nics_installed": "",
    "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)",
    "other_hardware": "",
    "other_software_stack": "TensorRT 7.2.3, CUDA 11.1, cuDNN 8.1.1, Driver 460.27.04, DALI 0.30.0, Triton 21.02; GCC 7.5.0; Python 3.7.10",
    "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": 7322528924094909334,
    "retraining": "N",
    "sample_index_rng_seed": 1570999273408051088,
    "samples_per_query": 158400,
    "schedule_rng_seed": 3507442325620259414,
    "starting_weights_filename": "224_224_160_dyanmic_bs.onnx",
    "status": "available",
    "submitter": "HPE",
    "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/HPE",
    "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/HPE-ProLiant-XL645d-Gen10-Plus_A100-SXM4-40GBx4_TRT_Triton",
    "system_name": "Microsoft Corporation 7.0 (Virtual Machine)",
    "system_type": "datacenter",
    "target_latency (ns)": 0,
    "target_qps": 240,
    "task": "image segmentation",
    "task2": "image segmentation",
    "total_cores": 120,
    "uid": "e7d7924abe54c9fb",
    "use_accelerator": true,
    "weight_data_types": "int8",
    "weight_transformations": "quantization, affine fusion"
  },
  {
    "50.00 percentile latency (ns)": 367430046108,
    "90.00 percentile latency (ns)": 661745946290,
    "95.00 percentile latency (ns)": 698683950949,
    "97.00 percentile latency (ns)": 713535995182,
    "99.00 percentile latency (ns)": 728161362704,
    "99.90 percentile latency (ns)": 734727538670,
    "Max latency (ns)": 735455206945,
    "Mean latency (ns)": 367283572012,
    "Min duration satisfied": "Yes",
    "Min latency (ns)": 74978274,
    "Min queries satisfied": "Yes",
    "Mode": "PerformanceOnly",
    "Result is": "VALID",
    "SUT name": "Triton_Server",
    "Samples per second": 215.377,
    "Scenario": "offline",
    "accelerator_frequency": "",
    "accelerator_host_interconnect": "",
    "accelerator_interconnect": "",
    "accelerator_interconnect_topology": "",
    "accelerator_memory_capacity": "40 GB",
    "accelerator_memory_configuration": "HBM2e",
    "accelerator_model_name": "NVIDIA A100-SXM4-40GB",
    "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": "",
    "characteristics.enhancing tumor": 0.7798,
    "characteristics.mean": 0.8543,
    "characteristics.samples_per_second": 215.377,
    "characteristics.samples_per_second.normalized_per_core": 53.84425,
    "characteristics.samples_per_second.normalized_per_processor": 53.84425,
    "characteristics.tumor core": 0.8691,
    "characteristics.whole tumor": 0.914,
    "ck_system": "HPE-ProLiant-XL645d-Gen10-Plus_A100-SXM4-40GBx4_TRT_Triton",
    "ck_used": true,
    "cooling": "",
    "dataset": "BraTS 2019",
    "dataset_link": "https://www.med.upenn.edu/cbica/brats2019/data.html",
    "dim_x_default": "characteristics.samples_per_second",
    "dim_x_maximize": true,
    "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 7.2.3, CUDA 11.1",
    "host_memory_capacity": "512 GB",
    "host_memory_configuration": "",
    "host_networking": "",
    "host_networking_topology": "",
    "host_processor_caches": "",
    "host_processor_core_count": 120,
    "host_processor_frequency": "",
    "host_processor_interconnect": "",
    "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor",
    "host_processors_per_node": 1,
    "host_storage_capacity": "3.2 TB",
    "host_storage_type": "SSD",
    "hw_notes": "",
    "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": 1,
    "mlperf_version": 1.0,
    "network_speed_mbit": "",
    "nics_enabled_connected": "",
    "nics_enabled_firmware": "",
    "nics_enabled_os": "",
    "normalize_cores": 4,
    "normalize_processors": 4,
    "note_code": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/HPE/code",
    "note_details": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/HPE/results/HPE-ProLiant-XL645d-Gen10-Plus_A100-SXM4-40GBx4_TRT_Triton",
    "number_of_nodes": 1,
    "number_of_type_nics_installed": "",
    "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)",
    "other_hardware": "",
    "other_software_stack": "TensorRT 7.2.3, CUDA 11.1, cuDNN 8.1.1, Driver 460.27.04, DALI 0.30.0, Triton 21.02; GCC 7.5.0; Python 3.7.10",
    "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": 7322528924094909334,
    "retraining": "N",
    "sample_index_rng_seed": 1570999273408051088,
    "samples_per_query": 158400,
    "schedule_rng_seed": 3507442325620259414,
    "starting_weights_filename": "224_224_160_dyanmic_bs.onnx",
    "status": "available",
    "submitter": "HPE",
    "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/HPE",
    "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/HPE-ProLiant-XL645d-Gen10-Plus_A100-SXM4-40GBx4_TRT_Triton",
    "system_name": "Microsoft Corporation 7.0 (Virtual Machine)",
    "system_type": "datacenter",
    "target_latency (ns)": 0,
    "target_qps": 240,
    "task": "image segmentation",
    "task2": "image segmentation",
    "total_cores": 120,
    "uid": "87a41d19d8739ff9",
    "use_accelerator": true,
    "weight_data_types": "int8",
    "weight_transformations": "quantization, affine fusion"
  },
  {
    "50.00 percentile latency (ns)": 339279746144,
    "90.00 percentile latency (ns)": 611082393722,
    "95.00 percentile latency (ns)": 645065752591,
    "97.00 percentile latency (ns)": 658658972097,
    "99.00 percentile latency (ns)": 672244303871,
    "99.90 percentile latency (ns)": 678356975718,
    "Max latency (ns)": 679025961567,
    "Mean latency (ns)": 339330097812,
    "Min duration satisfied": "Yes",
    "Min latency (ns)": 48505915,
    "Min queries satisfied": "Yes",
    "Mode": "PerformanceOnly",
    "Result is": "VALID",
    "SUT name": "LWIS_Server",
    "Samples per second": 233.275,
    "Scenario": "offline",
    "accelerator_frequency": "",
    "accelerator_host_interconnect": "",
    "accelerator_interconnect": "",
    "accelerator_interconnect_topology": "",
    "accelerator_memory_capacity": "40 GB",
    "accelerator_memory_configuration": "HBM2e",
    "accelerator_model_name": "NVIDIA A100-SXM4-40GB",
    "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": "",
    "characteristics.enhancing tumor": 0.7798,
    "characteristics.mean": 0.8543,
    "characteristics.samples_per_second": 233.275,
    "characteristics.samples_per_second.normalized_per_core": 58.31875,
    "characteristics.samples_per_second.normalized_per_processor": 58.31875,
    "characteristics.tumor core": 0.8691,
    "characteristics.whole tumor": 0.914,
    "ck_system": "HPE-ProLiant-XL645d-Gen10-Plus_A100-SXM4-40GBx4_TRT",
    "ck_used": true,
    "cooling": "",
    "dataset": "BraTS 2019",
    "dataset_link": "https://www.med.upenn.edu/cbica/brats2019/data.html",
    "dim_x_default": "characteristics.samples_per_second",
    "dim_x_maximize": true,
    "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 7.2.3, CUDA 11.1",
    "host_memory_capacity": "512 GB",
    "host_memory_configuration": "",
    "host_networking": "",
    "host_networking_topology": "",
    "host_processor_caches": "",
    "host_processor_core_count": 120,
    "host_processor_frequency": "",
    "host_processor_interconnect": "",
    "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor",
    "host_processors_per_node": 1,
    "host_storage_capacity": "3.2 TB",
    "host_storage_type": "SSD",
    "hw_notes": "",
    "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": 1,
    "mlperf_version": 1.0,
    "network_speed_mbit": "",
    "nics_enabled_connected": "",
    "nics_enabled_firmware": "",
    "nics_enabled_os": "",
    "normalize_cores": 4,
    "normalize_processors": 4,
    "note_code": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/HPE/code",
    "note_details": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/HPE/results/HPE-ProLiant-XL645d-Gen10-Plus_A100-SXM4-40GBx4_TRT",
    "number_of_nodes": 1,
    "number_of_type_nics_installed": "",
    "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)",
    "other_hardware": "",
    "other_software_stack": "TensorRT 7.2.3, CUDA 11.1, cuDNN 8.1.1, Driver 460.27.04, DALI 0.30.0; GCC 7.5.0; Python 3.7.10",
    "performance_issue_same": 0,
    "performance_issue_same_index": 0,
    "performance_issue_unique": 0,
    "performance_sample_count": 16,
    "power_management": "",
    "power_supply_details": "",
    "power_supply_quantity_and_rating_watts": "",
    "print_timestamps": 0,
    "problem": false,
    "qsl_rng_seed": 7322528924094909334,
    "retraining": "N",
    "sample_index_rng_seed": 1570999273408051088,
    "samples_per_query": 158400,
    "schedule_rng_seed": 3507442325620259414,
    "starting_weights_filename": "224_224_160_dyanmic_bs.onnx",
    "status": "available",
    "submitter": "HPE",
    "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/HPE",
    "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/HPE-ProLiant-XL645d-Gen10-Plus_A100-SXM4-40GBx4_TRT",
    "system_name": "Microsoft Corporation 7.0 (Virtual Machine)",
    "system_type": "datacenter",
    "target_latency (ns)": 0,
    "target_qps": 240,
    "task": "image segmentation",
    "task2": "image segmentation",
    "total_cores": 120,
    "uid": "d5dee06180404e7b",
    "use_accelerator": true,
    "weight_data_types": "int8",
    "weight_transformations": "quantization, affine fusion"
  },
  {
    "50.00 percentile latency (ns)": 339279746144,
    "90.00 percentile latency (ns)": 611082393722,
    "95.00 percentile latency (ns)": 645065752591,
    "97.00 percentile latency (ns)": 658658972097,
    "99.00 percentile latency (ns)": 672244303871,
    "99.90 percentile latency (ns)": 678356975718,
    "Max latency (ns)": 679025961567,
    "Mean latency (ns)": 339330097812,
    "Min duration satisfied": "Yes",
    "Min latency (ns)": 48505915,
    "Min queries satisfied": "Yes",
    "Mode": "PerformanceOnly",
    "Result is": "VALID",
    "SUT name": "LWIS_Server",
    "Samples per second": 233.275,
    "Scenario": "offline",
    "accelerator_frequency": "",
    "accelerator_host_interconnect": "",
    "accelerator_interconnect": "",
    "accelerator_interconnect_topology": "",
    "accelerator_memory_capacity": "40 GB",
    "accelerator_memory_configuration": "HBM2e",
    "accelerator_model_name": "NVIDIA A100-SXM4-40GB",
    "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": "",
    "characteristics.enhancing tumor": 0.7798,
    "characteristics.mean": 0.8543,
    "characteristics.samples_per_second": 233.275,
    "characteristics.samples_per_second.normalized_per_core": 58.31875,
    "characteristics.samples_per_second.normalized_per_processor": 58.31875,
    "characteristics.tumor core": 0.8691,
    "characteristics.whole tumor": 0.914,
    "ck_system": "HPE-ProLiant-XL645d-Gen10-Plus_A100-SXM4-40GBx4_TRT",
    "ck_used": true,
    "cooling": "",
    "dataset": "BraTS 2019",
    "dataset_link": "https://www.med.upenn.edu/cbica/brats2019/data.html",
    "dim_x_default": "characteristics.samples_per_second",
    "dim_x_maximize": true,
    "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 7.2.3, CUDA 11.1",
    "host_memory_capacity": "512 GB",
    "host_memory_configuration": "",
    "host_networking": "",
    "host_networking_topology": "",
    "host_processor_caches": "",
    "host_processor_core_count": 120,
    "host_processor_frequency": "",
    "host_processor_interconnect": "",
    "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor",
    "host_processors_per_node": 1,
    "host_storage_capacity": "3.2 TB",
    "host_storage_type": "SSD",
    "hw_notes": "",
    "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": 1,
    "mlperf_version": 1.0,
    "network_speed_mbit": "",
    "nics_enabled_connected": "",
    "nics_enabled_firmware": "",
    "nics_enabled_os": "",
    "normalize_cores": 4,
    "normalize_processors": 4,
    "note_code": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/HPE/code",
    "note_details": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/HPE/results/HPE-ProLiant-XL645d-Gen10-Plus_A100-SXM4-40GBx4_TRT",
    "number_of_nodes": 1,
    "number_of_type_nics_installed": "",
    "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)",
    "other_hardware": "",
    "other_software_stack": "TensorRT 7.2.3, CUDA 11.1, cuDNN 8.1.1, Driver 460.27.04, DALI 0.30.0; GCC 7.5.0; Python 3.7.10",
    "performance_issue_same": 0,
    "performance_issue_same_index": 0,
    "performance_issue_unique": 0,
    "performance_sample_count": 16,
    "power_management": "",
    "power_supply_details": "",
    "power_supply_quantity_and_rating_watts": "",
    "print_timestamps": 0,
    "problem": false,
    "qsl_rng_seed": 7322528924094909334,
    "retraining": "N",
    "sample_index_rng_seed": 1570999273408051088,
    "samples_per_query": 158400,
    "schedule_rng_seed": 3507442325620259414,
    "starting_weights_filename": "224_224_160_dyanmic_bs.onnx",
    "status": "available",
    "submitter": "HPE",
    "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/HPE",
    "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/HPE-ProLiant-XL645d-Gen10-Plus_A100-SXM4-40GBx4_TRT",
    "system_name": "Microsoft Corporation 7.0 (Virtual Machine)",
    "system_type": "datacenter",
    "target_latency (ns)": 0,
    "target_qps": 240,
    "task": "image segmentation",
    "task2": "image segmentation",
    "total_cores": 120,
    "uid": "4f6483f33e50b74a",
    "use_accelerator": true,
    "weight_data_types": "int8",
    "weight_transformations": "quantization, affine fusion"
  }
]