[ { "50.00 percentile latency (ns)": 393528986314, "90.00 percentile latency (ns)": 708464350449, "95.00 percentile latency (ns)": 747796152762, "97.00 percentile latency (ns)": 763668449394, "99.00 percentile latency (ns)": 779388276209, "99.90 percentile latency (ns)": 786432459643, "Max latency (ns)": 787197283707, "Mean latency (ns)": 393544222294, "Min duration satisfied": "Yes", "Min latency (ns)": 345912865, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 31.2196, "Scenario": "offline", "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.enhancing tumor": 0.7798, "characteristics.mean": 0.85427, "characteristics.samples_per_second": 31.2196, "characteristics.samples_per_second.normalized_per_core": 7.8049, "characteristics.samples_per_second.normalized_per_processor": 7.8049, "characteristics.tumor core": 0.869, "characteristics.whole tumor": 0.914, "ck_system": "XE2420_datacenter_T4x4_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, "division": "closed", "filesystem": "ext3/ext4", "formal_model": "3d-unet", "formal_model_accuracy": 99.0, "formal_model_link": "", "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": 120, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "4 TB", "host_storage_type": "NVMe SSD", "hw_notes": "ECC on", "informal_model": "3d-unet-99", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "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": 1, "mlperf_version": 1.0, "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_datacenter_T4x4_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "TensorRT 7.2.3, CUDA 11.1, cuDNN 8.1.1, Driver 460.32.03, 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_settings": "closed/DellEMC/measurements/XE2420_datacenter_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": 24576, "schedule_rng_seed": 3507442325620259414, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "sw_notes": "Powered by CK v2.5.8 (https://github.com/ctuning/ck)", "system_cooling_type": "air", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/XE2420_datacenter_T4x4_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 8, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "92566bcd81004045", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 393528986314, "90.00 percentile latency (ns)": 708464350449, "95.00 percentile latency (ns)": 747796152762, "97.00 percentile latency (ns)": 763668449394, "99.00 percentile latency (ns)": 779388276209, "99.90 percentile latency (ns)": 786432459643, "Max latency (ns)": 787197283707, "Mean latency (ns)": 393544222294, "Min duration satisfied": "Yes", "Min latency (ns)": 345912865, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 31.2196, "Scenario": "offline", "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.enhancing tumor": 0.7798, "characteristics.mean": 0.85427, "characteristics.samples_per_second": 31.2196, "characteristics.samples_per_second.normalized_per_core": 7.8049, "characteristics.samples_per_second.normalized_per_processor": 7.8049, "characteristics.tumor core": 0.869, "characteristics.whole tumor": 0.914, "ck_system": "XE2420_datacenter_T4x4_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, "division": "closed", "filesystem": "ext3/ext4", "formal_model": "3d-unet", "formal_model_accuracy": 99.9, "formal_model_link": "", "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": 120, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "4 TB", "host_storage_type": "NVMe SSD", "hw_notes": "ECC on", "informal_model": "3d-unet-99.9", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "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": 1, "mlperf_version": 1.0, "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_datacenter_T4x4_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "TensorRT 7.2.3, CUDA 11.1, cuDNN 8.1.1, Driver 460.32.03, 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_settings": "closed/DellEMC/measurements/XE2420_datacenter_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": 24576, "schedule_rng_seed": 3507442325620259414, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "sw_notes": "Powered by CK v2.5.8 (https://github.com/ctuning/ck)", "system_cooling_type": "air", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/XE2420_datacenter_T4x4_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 8, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "8a846a4d4b983034", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 314201803541, "90.00 percentile latency (ns)": 565391117437, "95.00 percentile latency (ns)": 596802706348, "97.00 percentile latency (ns)": 609353895749, "99.00 percentile latency (ns)": 621934589632, "99.90 percentile latency (ns)": 627574281161, "Max latency (ns)": 628198883953, "Mean latency (ns)": 314138376642, "Min duration satisfied": "Yes", "Min latency (ns)": 92453052, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 244.795, "Scenario": "offline", "accelerator_cooling_type": "", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "80GB", "accelerator_memory_configuration": "HBM2", "accelerator_model_name": "NVIDIA A100-SXM-80GB", "accelerator_on-chip_memories": "", "accelerators_per_node": 4, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7798, "characteristics.mean": 0.8543, "characteristics.samples_per_second": 244.795, "characteristics.samples_per_second.normalized_per_core": 61.19875, "characteristics.samples_per_second.normalized_per_processor": 61.19875, "characteristics.tumor core": 0.8691, "characteristics.whole tumor": 0.914, "ck_system": "XE8545_7763_A100-SXM4-80GBx4_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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.0, "formal_model_link": "", "framework": "TensorRT 7.2.3, CUDA 11.1", "host_cooling_type": "", "host_memory_capacity": "1 TB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 120, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "3 TB", "host_storage_type": "NVMe SSD", "hw_notes": "500W A100-SXM-80GB", "informal_model": "3d-unet-99", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1, "mlperf_version": 1.0, "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/XE8545_7763_A100-SXM4-80GBx4_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "TensorRT 7.2.3, CUDA 11.1, cuDNN 8.1.1, Driver 460.32.03, 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": "", "print_timestamps": 0, "problem": false, "qsl_rng_seed": 7322528924094909334, "retraining": "N", "sample_index_rng_seed": 1570999273408051088, "samples_per_query": 153780, "schedule_rng_seed": 3507442325620259414, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/XE8545_7763_A100-SXM4-80GBx4_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 233, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "3dd7bbb479ea32c0", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 314201803541, "90.00 percentile latency (ns)": 565391117437, "95.00 percentile latency (ns)": 596802706348, "97.00 percentile latency (ns)": 609353895749, "99.00 percentile latency (ns)": 621934589632, "99.90 percentile latency (ns)": 627574281161, "Max latency (ns)": 628198883953, "Mean latency (ns)": 314138376642, "Min duration satisfied": "Yes", "Min latency (ns)": 92453052, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 244.795, "Scenario": "offline", "accelerator_cooling_type": "", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "80GB", "accelerator_memory_configuration": "HBM2", "accelerator_model_name": "NVIDIA A100-SXM-80GB", "accelerator_on-chip_memories": "", "accelerators_per_node": 4, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7798, "characteristics.mean": 0.8543, "characteristics.samples_per_second": 244.795, "characteristics.samples_per_second.normalized_per_core": 61.19875, "characteristics.samples_per_second.normalized_per_processor": 61.19875, "characteristics.tumor core": 0.8691, "characteristics.whole tumor": 0.914, "ck_system": "XE8545_7763_A100-SXM4-80GBx4_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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.9, "formal_model_link": "", "framework": "TensorRT 7.2.3, CUDA 11.1", "host_cooling_type": "", "host_memory_capacity": "1 TB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 120, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "3 TB", "host_storage_type": "NVMe SSD", "hw_notes": "500W A100-SXM-80GB", "informal_model": "3d-unet-99.9", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1, "mlperf_version": 1.0, "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/XE8545_7763_A100-SXM4-80GBx4_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "TensorRT 7.2.3, CUDA 11.1, cuDNN 8.1.1, Driver 460.32.03, 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": "", "print_timestamps": 0, "problem": false, "qsl_rng_seed": 7322528924094909334, "retraining": "N", "sample_index_rng_seed": 1570999273408051088, "samples_per_query": 153780, "schedule_rng_seed": 3507442325620259414, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/XE8545_7763_A100-SXM4-80GBx4_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 233, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "4f2b9d3408fc567f", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 344377639195, "90.00 percentile latency (ns)": 619721400616, "95.00 percentile latency (ns)": 654153457934, "97.00 percentile latency (ns)": 667904430893, "99.00 percentile latency (ns)": 681622065651, "99.90 percentile latency (ns)": 687776662737, "Max latency (ns)": 688450848138, "Mean latency (ns)": 344373630075, "Min duration satisfied": "Yes", "Min latency (ns)": 71169978, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 223.371, "Scenario": "offline", "accelerator_cooling_type": "", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "40GB", "accelerator_memory_configuration": "HBM2", "accelerator_model_name": "NVIDIA A100-SXM-40GB", "accelerator_on-chip_memories": "", "accelerators_per_node": 4, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7798, "characteristics.mean": 0.8543, "characteristics.samples_per_second": 223.371, "characteristics.samples_per_second.normalized_per_core": 55.84275, "characteristics.samples_per_second.normalized_per_processor": 55.84275, "characteristics.tumor core": 0.8691, "characteristics.whole tumor": 0.914, "ck_system": "XE8545_7713_A100-SXM4-40GBx4", "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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.0, "formal_model_link": "", "framework": "TensorRT 7.2.3, CUDA 11.1", "host_cooling_type": "", "host_memory_capacity": "1 TB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 120, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "3 TB", "host_storage_type": "NVMe SSD", "hw_notes": "", "informal_model": "3d-unet-99", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1, "mlperf_version": 1.0, "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/XE8545_7713_A100-SXM4-40GBx4", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "TensorRT 7.2.3, CUDA 11.1, cuDNN 8.1.1, Driver 460.32.03, 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": "", "print_timestamps": 0, "problem": false, "qsl_rng_seed": 7322528924094909334, "retraining": "N", "sample_index_rng_seed": 1570999273408051088, "samples_per_query": 153780, "schedule_rng_seed": 3507442325620259414, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/XE8545_7713_A100-SXM4-40GBx4", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 233, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "2149e1a0d2e1ef65", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 344377639195, "90.00 percentile latency (ns)": 619721400616, "95.00 percentile latency (ns)": 654153457934, "97.00 percentile latency (ns)": 667904430893, "99.00 percentile latency (ns)": 681622065651, "99.90 percentile latency (ns)": 687776662737, "Max latency (ns)": 688450848138, "Mean latency (ns)": 344373630075, "Min duration satisfied": "Yes", "Min latency (ns)": 71169978, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 223.371, "Scenario": "offline", "accelerator_cooling_type": "", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "40GB", "accelerator_memory_configuration": "HBM2", "accelerator_model_name": "NVIDIA A100-SXM-40GB", "accelerator_on-chip_memories": "", "accelerators_per_node": 4, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7798, "characteristics.mean": 0.8543, "characteristics.samples_per_second": 223.371, "characteristics.samples_per_second.normalized_per_core": 55.84275, "characteristics.samples_per_second.normalized_per_processor": 55.84275, "characteristics.tumor core": 0.8691, "characteristics.whole tumor": 0.914, "ck_system": "XE8545_7713_A100-SXM4-40GBx4", "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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.9, "formal_model_link": "", "framework": "TensorRT 7.2.3, CUDA 11.1", "host_cooling_type": "", "host_memory_capacity": "1 TB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 120, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "3 TB", "host_storage_type": "NVMe SSD", "hw_notes": "", "informal_model": "3d-unet-99.9", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1, "mlperf_version": 1.0, "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/XE8545_7713_A100-SXM4-40GBx4", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "TensorRT 7.2.3, CUDA 11.1, cuDNN 8.1.1, Driver 460.32.03, 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": "", "print_timestamps": 0, "problem": false, "qsl_rng_seed": 7322528924094909334, "retraining": "N", "sample_index_rng_seed": 1570999273408051088, "samples_per_query": 153780, "schedule_rng_seed": 3507442325620259414, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/XE8545_7713_A100-SXM4-40GBx4", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 233, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "604d5e7444e9b69e", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 329064872851, "90.00 percentile latency (ns)": 592139258652, "95.00 percentile latency (ns)": 625018608015, "97.00 percentile latency (ns)": 638190817713, "99.00 percentile latency (ns)": 651332056727, "99.90 percentile latency (ns)": 657249164415, "Max latency (ns)": 657902552866, "Mean latency (ns)": 329039981108, "Min duration satisfied": "Yes", "Min latency (ns)": 63748952, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 233.743, "Scenario": "offline", "accelerator_cooling_type": "", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "40GB", "accelerator_memory_configuration": "HBM2", "accelerator_model_name": "NVIDIA A100-SXM-40GB", "accelerator_on-chip_memories": "", "accelerators_per_node": 4, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7798, "characteristics.mean": 0.8543, "characteristics.samples_per_second": 233.743, "characteristics.samples_per_second.normalized_per_core": 58.43575, "characteristics.samples_per_second.normalized_per_processor": 58.43575, "characteristics.tumor core": 0.8691, "characteristics.whole tumor": 0.914, "ck_system": "XE8545_7763_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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.0, "formal_model_link": "", "framework": "TensorRT 7.2.3, CUDA 11.1", "host_cooling_type": "", "host_memory_capacity": "1 TB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 120, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "3 TB", "host_storage_type": "NVMe SSD", "hw_notes": "", "informal_model": "3d-unet-99", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1, "mlperf_version": 1.0, "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/XE8545_7763_A100-SXM4-40GBx4_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "TensorRT 7.2.3, CUDA 11.1, cuDNN 8.1.1, Driver 460.32.03, 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": "", "print_timestamps": 0, "problem": false, "qsl_rng_seed": 7322528924094909334, "retraining": "N", "sample_index_rng_seed": 1570999273408051088, "samples_per_query": 153780, "schedule_rng_seed": 3507442325620259414, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/XE8545_7763_A100-SXM4-40GBx4_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 233, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "7ea684bafaf00a1e", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 329064872851, "90.00 percentile latency (ns)": 592139258652, "95.00 percentile latency (ns)": 625018608015, "97.00 percentile latency (ns)": 638190817713, "99.00 percentile latency (ns)": 651332056727, "99.90 percentile latency (ns)": 657249164415, "Max latency (ns)": 657902552866, "Mean latency (ns)": 329039981108, "Min duration satisfied": "Yes", "Min latency (ns)": 63748952, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 233.743, "Scenario": "offline", "accelerator_cooling_type": "", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "40GB", "accelerator_memory_configuration": "HBM2", "accelerator_model_name": "NVIDIA A100-SXM-40GB", "accelerator_on-chip_memories": "", "accelerators_per_node": 4, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7798, "characteristics.mean": 0.8543, "characteristics.samples_per_second": 233.743, "characteristics.samples_per_second.normalized_per_core": 58.43575, "characteristics.samples_per_second.normalized_per_processor": 58.43575, "characteristics.tumor core": 0.8691, "characteristics.whole tumor": 0.914, "ck_system": "XE8545_7763_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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.9, "formal_model_link": "", "framework": "TensorRT 7.2.3, CUDA 11.1", "host_cooling_type": "", "host_memory_capacity": "1 TB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 120, "host_processor_frequency": "", "host_processor_interconnect": "", "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "3 TB", "host_storage_type": "NVMe SSD", "hw_notes": "", "informal_model": "3d-unet-99.9", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1, "mlperf_version": 1.0, "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/XE8545_7763_A100-SXM4-40GBx4_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "TensorRT 7.2.3, CUDA 11.1, cuDNN 8.1.1, Driver 460.32.03, 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": "", "print_timestamps": 0, "problem": false, "qsl_rng_seed": 7322528924094909334, "retraining": "N", "sample_index_rng_seed": 1570999273408051088, "samples_per_query": 153780, "schedule_rng_seed": 3507442325620259414, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/XE8545_7763_A100-SXM4-40GBx4_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 233, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "45084f22a621fa6e", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 319688810554, "90.00 percentile latency (ns)": 575598005546, "95.00 percentile latency (ns)": 607575218504, "97.00 percentile latency (ns)": 620356864516, "99.00 percentile latency (ns)": 633142097309, "99.90 percentile latency (ns)": 638915693356, "Max latency (ns)": 639532345716, "Mean latency (ns)": 319730580613, "Min duration satisfied": "Yes", "Min latency (ns)": 158403211, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 55.7282, "Scenario": "offline", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "48 GB", "accelerator_memory_configuration": "GDDR6", "accelerator_model_name": "NVIDIA Quadro RTX 8000", "accelerator_on-chip_memories": "", "accelerators_per_node": 3, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7798, "characteristics.mean": 0.85427, "characteristics.samples_per_second": 55.7282, "characteristics.samples_per_second.normalized_per_core": 18.576066666666666, "characteristics.samples_per_second.normalized_per_processor": 18.576066666666666, "characteristics.tumor core": 0.869, "characteristics.whole tumor": 0.914, "ck_system": "R7525_QuadroRTX8000x3_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, "division": "closed", "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": 2, "host_storage_capacity": "1.8 TB", "host_storage_type": "NVMe SSD", "hw_notes": "ECC on", "informal_model": "3d-unet-99", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1, "mlperf_version": 1.0, "normalize_cores": 3, "normalize_processors": 3, "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/R7525_QuadroRTX8000x3_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "TensorRT 7.2.3, CUDA 11.1, cuDNN 8.1.1, Driver 460.32.03, 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, "print_timestamps": 0, "problem": false, "qsl_rng_seed": 7322528924094909334, "retraining": "N", "sample_index_rng_seed": 1570999273408051088, "samples_per_query": 35640, "schedule_rng_seed": 3507442325620259414, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/R7525_QuadroRTX8000x3_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 54, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "126c694f57340098", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 319688810554, "90.00 percentile latency (ns)": 575598005546, "95.00 percentile latency (ns)": 607575218504, "97.00 percentile latency (ns)": 620356864516, "99.00 percentile latency (ns)": 633142097309, "99.90 percentile latency (ns)": 638915693356, "Max latency (ns)": 639532345716, "Mean latency (ns)": 319730580613, "Min duration satisfied": "Yes", "Min latency (ns)": 158403211, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 55.7282, "Scenario": "offline", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "48 GB", "accelerator_memory_configuration": "GDDR6", "accelerator_model_name": "NVIDIA Quadro RTX 8000", "accelerator_on-chip_memories": "", "accelerators_per_node": 3, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7798, "characteristics.mean": 0.85427, "characteristics.samples_per_second": 55.7282, "characteristics.samples_per_second.normalized_per_core": 18.576066666666666, "characteristics.samples_per_second.normalized_per_processor": 18.576066666666666, "characteristics.tumor core": 0.869, "characteristics.whole tumor": 0.914, "ck_system": "R7525_QuadroRTX8000x3_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, "division": "closed", "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": 2, "host_storage_capacity": "1.8 TB", "host_storage_type": "NVMe SSD", "hw_notes": "ECC on", "informal_model": "3d-unet-99.9", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1, "mlperf_version": 1.0, "normalize_cores": 3, "normalize_processors": 3, "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/R7525_QuadroRTX8000x3_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "TensorRT 7.2.3, CUDA 11.1, cuDNN 8.1.1, Driver 460.32.03, 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, "print_timestamps": 0, "problem": false, "qsl_rng_seed": 7322528924094909334, "retraining": "N", "sample_index_rng_seed": 1570999273408051088, "samples_per_query": 35640, "schedule_rng_seed": 3507442325620259414, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/R7525_QuadroRTX8000x3_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 54, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "de11b74ec7b6ee65", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 209978448066, "90.00 percentile latency (ns)": 378594336418, "95.00 percentile latency (ns)": 399669113040, "97.00 percentile latency (ns)": 408073321600, "99.00 percentile latency (ns)": 416538975700, "99.90 percentile latency (ns)": 420274949324, "Max latency (ns)": 420729017221, "Mean latency (ns)": 210079885227, "Min duration satisfied": "Yes", "Min latency (ns)": 312653483, "Min queries satisfied": "Yes", "Mode": "Performance", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 58.4129, "Scenario": "offline", "accelerator_frequency": "1590MHz", "accelerator_host_interconnect": "PCI Express 3.0", "accelerator_interconnect": "PCI Express 3.0", "accelerator_interconnect_topology": "4 Accelerators per CPU", "accelerator_memory_capacity": "16 GB", "accelerator_memory_configuration": "GDDR6", "accelerator_model_name": "NVIDIA Tesla T4", "accelerator_on-chip_memories": "", "accelerators_per_node": 8, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7793, "characteristics.mean": 0.85367, "characteristics.samples_per_second": 58.4129, "characteristics.samples_per_second.normalized_per_core": 7.3016125, "characteristics.samples_per_second.normalized_per_processor": 7.3016125, "characteristics.tumor core": 0.8692, "characteristics.whole tumor": 0.9125, "ck_system": "R7525_T4x8_TRT", "ck_used": true, "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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.0, "formal_model_link": "", "framework": "TensorRT 7.2, CUDA 11.0, cuDNN 8.0.2, cuBLAS 11.2.0, libjemalloc2, cub 1.8.0, tensorrt-laboratory mlperf branch", "host_memory_capacity": "1 TB", "host_memory_configuration": "8x64GB DDR4-3200 HMAA8GR7AJR4N-XN RDIMM ECC", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "2MB+16MB+128MB", "host_processor_core_count": 120, "host_processor_frequency": "2.35GHz", "host_processor_interconnect": "Infinity Fabric", "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "3 TB (5x800GB WUSTR6480ASS200 in RAID5)", "host_storage_type": "3D-TLC Solid State with 12Gbps SAS", "hw_notes": "ECC off", "informal_model": "3d-unet-99", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 1, "mlperf_version": 0.7, "normalize_cores": 8, "normalize_processors": 8, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/results/R7525_T4x8_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "docker 19.03.12, python 3.6.8, gcc 5.5.0, onnx 1.3.0, tensorflow 1.13.1, pytorch 1.1.0, torchvision 0.3.0, pycuda 2019.1, sacrebleu 1.3.3, SimpleJSON, OpenCV 4.1.1; GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 16, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 24576, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/R7525_T4x8_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 80, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "c03543848a791137", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 209978448066, "90.00 percentile latency (ns)": 378594336418, "95.00 percentile latency (ns)": 399669113040, "97.00 percentile latency (ns)": 408073321600, "99.00 percentile latency (ns)": 416538975700, "99.90 percentile latency (ns)": 420274949324, "Max latency (ns)": 420729017221, "Mean latency (ns)": 210079885227, "Min duration satisfied": "Yes", "Min latency (ns)": 312653483, "Min queries satisfied": "Yes", "Mode": "Performance", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 58.4129, "Scenario": "offline", "accelerator_frequency": "1590MHz", "accelerator_host_interconnect": "PCI Express 3.0", "accelerator_interconnect": "PCI Express 3.0", "accelerator_interconnect_topology": "4 Accelerators per CPU", "accelerator_memory_capacity": "16 GB", "accelerator_memory_configuration": "GDDR6", "accelerator_model_name": "NVIDIA Tesla T4", "accelerator_on-chip_memories": "", "accelerators_per_node": 8, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7793, "characteristics.mean": 0.85367, "characteristics.samples_per_second": 58.4129, "characteristics.samples_per_second.normalized_per_core": 7.3016125, "characteristics.samples_per_second.normalized_per_processor": 7.3016125, "characteristics.tumor core": 0.8692, "characteristics.whole tumor": 0.9125, "ck_system": "R7525_T4x8_TRT", "ck_used": true, "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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.9, "formal_model_link": "", "framework": "TensorRT 7.2, CUDA 11.0, cuDNN 8.0.2, cuBLAS 11.2.0, libjemalloc2, cub 1.8.0, tensorrt-laboratory mlperf branch", "host_memory_capacity": "1 TB", "host_memory_configuration": "8x64GB DDR4-3200 HMAA8GR7AJR4N-XN RDIMM ECC", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "2MB+16MB+128MB", "host_processor_core_count": 120, "host_processor_frequency": "2.35GHz", "host_processor_interconnect": "Infinity Fabric", "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "3 TB (5x800GB WUSTR6480ASS200 in RAID5)", "host_storage_type": "3D-TLC Solid State with 12Gbps SAS", "hw_notes": "ECC off", "informal_model": "3d-unet-99.9", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 1, "mlperf_version": 0.7, "normalize_cores": 8, "normalize_processors": 8, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/results/R7525_T4x8_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "docker 19.03.12, python 3.6.8, gcc 5.5.0, onnx 1.3.0, tensorflow 1.13.1, pytorch 1.1.0, torchvision 0.3.0, pycuda 2019.1, sacrebleu 1.3.3, SimpleJSON, OpenCV 4.1.1; GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 16, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 24576, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/R7525_T4x8_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 80, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "919b12a4b54b5da1", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 405526821390, "90.00 percentile latency (ns)": 729628876566, "95.00 percentile latency (ns)": 770197783361, "97.00 percentile latency (ns)": 786368424316, "99.00 percentile latency (ns)": 802549179977, "99.90 percentile latency (ns)": 809852927405, "Max latency (ns)": 810635193242, "Mean latency (ns)": 405459058226, "Min duration satisfied": "Yes", "Min latency (ns)": 435427301, "Min queries satisfied": "Yes", "Mode": "Performance", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 30.317, "Scenario": "offline", "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": 4, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7793, "characteristics.mean": 0.85367, "characteristics.samples_per_second": 30.317, "characteristics.samples_per_second.normalized_per_core": 7.57925, "characteristics.samples_per_second.normalized_per_processor": 7.57925, "characteristics.tumor core": 0.8692, "characteristics.whole tumor": 0.9125, "ck_system": "XE2420_T4x4_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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.0, "formal_model_link": "", "framework": "TensorRT 7.2, CUDA 11.0 Update 1", "host_memory_capacity": "384 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 120, "host_processor_frequency": "2.10GHz", "host_processor_interconnect": "", "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "1 TB", "host_storage_type": "NVMe SSD", "hw_notes": "ECC off", "informal_model": "3d-unet-99", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 1, "mlperf_version": 0.7, "normalize_cores": 4, "normalize_processors": 4, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/results/XE2420_T4x4_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "TensorRT 7.2, CUDA 11.0 Update 1, cuDNN 8.0.2, DALI 0.25.0; GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 16, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 24576, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/XE2420_T4x4_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 30, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "ec9e8cd324e1c99b", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 405526821390, "90.00 percentile latency (ns)": 729628876566, "95.00 percentile latency (ns)": 770197783361, "97.00 percentile latency (ns)": 786368424316, "99.00 percentile latency (ns)": 802549179977, "99.90 percentile latency (ns)": 809852927405, "Max latency (ns)": 810635193242, "Mean latency (ns)": 405459058226, "Min duration satisfied": "Yes", "Min latency (ns)": 435427301, "Min queries satisfied": "Yes", "Mode": "Performance", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 30.317, "Scenario": "offline", "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": 4, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7793, "characteristics.mean": 0.85367, "characteristics.samples_per_second": 30.317, "characteristics.samples_per_second.normalized_per_core": 7.57925, "characteristics.samples_per_second.normalized_per_processor": 7.57925, "characteristics.tumor core": 0.8692, "characteristics.whole tumor": 0.9125, "ck_system": "XE2420_T4x4_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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.9, "formal_model_link": "", "framework": "TensorRT 7.2, CUDA 11.0 Update 1", "host_memory_capacity": "384 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 120, "host_processor_frequency": "2.10GHz", "host_processor_interconnect": "", "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "1 TB", "host_storage_type": "NVMe SSD", "hw_notes": "ECC off", "informal_model": "3d-unet-99.9", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 1, "mlperf_version": 0.7, "normalize_cores": 4, "normalize_processors": 4, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/results/XE2420_T4x4_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "TensorRT 7.2, CUDA 11.0 Update 1, cuDNN 8.0.2, DALI 0.25.0; GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 16, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 24576, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/XE2420_T4x4_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 30, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "e2816aecf74d48b7", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 86416953260, "90.00 percentile latency (ns)": 155635433787, "95.00 percentile latency (ns)": 164281272948, "97.00 percentile latency (ns)": 167754106253, "99.00 percentile latency (ns)": 171233790712, "99.90 percentile latency (ns)": 172759613302, "Max latency (ns)": 172943371513, "Mean latency (ns)": 86457688852, "Min duration satisfied": "Yes", "Min latency (ns)": 217179440, "Min queries satisfied": "Yes", "Mode": "Performance", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 142.104, "Scenario": "offline", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "48 GB", "accelerator_memory_configuration": "GDDR6", "accelerator_model_name": "NVIDIA Quadro RTX 8000", "accelerator_on-chip_memories": "", "accelerators_per_node": 8, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7793, "characteristics.mean": 0.85367, "characteristics.samples_per_second": 142.104, "characteristics.samples_per_second.normalized_per_core": 17.763, "characteristics.samples_per_second.normalized_per_processor": 17.763, "characteristics.tumor core": 0.8692, "characteristics.whole tumor": 0.9125, "ck_system": "DSS8440_QuadroRTX8000x8_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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.0, "formal_model_link": "", "framework": "TensorRT 7.2, CUDA 11.0 Update 1", "host_memory_capacity": "384 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": 2, "host_storage_capacity": "1.8 TB", "host_storage_type": "NVMe SSD", "hw_notes": "ECC off", "informal_model": "3d-unet-99", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 1, "mlperf_version": 0.7, "normalize_cores": 8, "normalize_processors": 8, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/results/DSS8440_QuadroRTX8000x8_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "TensorRT 7.2, CUDA 11.0 Update 1, cuDNN 8.0.2, DALI 0.25.0; GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 16, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 24576, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/DSS8440_QuadroRTX8000x8_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 142, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "0eef4c4fab96b151", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 86416953260, "90.00 percentile latency (ns)": 155635433787, "95.00 percentile latency (ns)": 164281272948, "97.00 percentile latency (ns)": 167754106253, "99.00 percentile latency (ns)": 171233790712, "99.90 percentile latency (ns)": 172759613302, "Max latency (ns)": 172943371513, "Mean latency (ns)": 86457688852, "Min duration satisfied": "Yes", "Min latency (ns)": 217179440, "Min queries satisfied": "Yes", "Mode": "Performance", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 142.104, "Scenario": "offline", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "48 GB", "accelerator_memory_configuration": "GDDR6", "accelerator_model_name": "NVIDIA Quadro RTX 8000", "accelerator_on-chip_memories": "", "accelerators_per_node": 8, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7793, "characteristics.mean": 0.85367, "characteristics.samples_per_second": 142.104, "characteristics.samples_per_second.normalized_per_core": 17.763, "characteristics.samples_per_second.normalized_per_processor": 17.763, "characteristics.tumor core": 0.8692, "characteristics.whole tumor": 0.9125, "ck_system": "DSS8440_QuadroRTX8000x8_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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.9, "formal_model_link": "", "framework": "TensorRT 7.2, CUDA 11.0 Update 1", "host_memory_capacity": "384 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": 2, "host_storage_capacity": "1.8 TB", "host_storage_type": "NVMe SSD", "hw_notes": "ECC off", "informal_model": "3d-unet-99.9", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 1, "mlperf_version": 0.7, "normalize_cores": 8, "normalize_processors": 8, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/results/DSS8440_QuadroRTX8000x8_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "TensorRT 7.2, CUDA 11.0 Update 1, cuDNN 8.0.2, DALI 0.25.0; GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 16, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 24576, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/DSS8440_QuadroRTX8000x8_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 142, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "ae9fb2582c0d4483", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 75639049760, "90.00 percentile latency (ns)": 135990511136, "95.00 percentile latency (ns)": 143519817644, "97.00 percentile latency (ns)": 146550590630, "99.00 percentile latency (ns)": 149585824776, "99.90 percentile latency (ns)": 150911275866, "Max latency (ns)": 151029547932, "Mean latency (ns)": 75643588548, "Min duration satisfied": "Yes", "Min latency (ns)": 236332117, "Min queries satisfied": "Yes", "Mode": "Performance", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 162.723, "Scenario": "offline", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "24 GB", "accelerator_memory_configuration": "GDDR6", "accelerator_model_name": "NVIDIA Quadro RTX 6000", "accelerator_on-chip_memories": "", "accelerators_per_node": 10, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7793, "characteristics.mean": 0.85367, "characteristics.samples_per_second": 162.723, "characteristics.samples_per_second.normalized_per_core": 16.2723, "characteristics.samples_per_second.normalized_per_processor": 16.2723, "characteristics.tumor core": 0.8692, "characteristics.whole tumor": 0.9125, "ck_system": "DSS8440_QuadroRTX6000x10_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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.0, "formal_model_link": "", "framework": "TensorRT 7.2, CUDA 11.0 Update 1", "host_memory_capacity": "384 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": 2, "host_storage_capacity": "1.8 TB", "host_storage_type": "NVMe SSD", "hw_notes": "ECC off", "informal_model": "3d-unet-99", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 1, "mlperf_version": 0.7, "normalize_cores": 10, "normalize_processors": 10, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/results/DSS8440_QuadroRTX6000x10_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "TensorRT 7.2, CUDA 11.0 Update 1, cuDNN 8.0.2, DALI 0.25.0; GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 16, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 24576, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/DSS8440_QuadroRTX6000x10_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 163, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "d2e534983a32c497", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 75582330279, "90.00 percentile latency (ns)": 135919166027, "95.00 percentile latency (ns)": 143445260413, "97.00 percentile latency (ns)": 146483789275, "99.00 percentile latency (ns)": 149516452949, "99.90 percentile latency (ns)": 150842395721, "Max latency (ns)": 150961626733, "Mean latency (ns)": 75588603194, "Min duration satisfied": "Yes", "Min latency (ns)": 219645462, "Min queries satisfied": "Yes", "Mode": "Performance", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 162.796, "Scenario": "offline", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "24 GB", "accelerator_memory_configuration": "GDDR6", "accelerator_model_name": "NVIDIA Quadro RTX 6000", "accelerator_on-chip_memories": "", "accelerators_per_node": 10, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7793, "characteristics.mean": 0.85367, "characteristics.samples_per_second": 162.796, "characteristics.samples_per_second.normalized_per_core": 16.2796, "characteristics.samples_per_second.normalized_per_processor": 16.2796, "characteristics.tumor core": 0.8692, "characteristics.whole tumor": 0.9125, "ck_system": "DSS8440_QuadroRTX6000x10_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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.9, "formal_model_link": "", "framework": "TensorRT 7.2, CUDA 11.0 Update 1", "host_memory_capacity": "384 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": 2, "host_storage_capacity": "1.8 TB", "host_storage_type": "NVMe SSD", "hw_notes": "ECC off", "informal_model": "3d-unet-99.9", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 1, "mlperf_version": 0.7, "normalize_cores": 10, "normalize_processors": 10, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/results/DSS8440_QuadroRTX6000x10_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "TensorRT 7.2, CUDA 11.0 Update 1, cuDNN 8.0.2, DALI 0.25.0; GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 16, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 24576, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/DSS8440_QuadroRTX6000x10_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 163, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "34e1c90b0e072018", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 166371893266, "90.00 percentile latency (ns)": 299611520305, "95.00 percentile latency (ns)": 316270364226, "97.00 percentile latency (ns)": 322925475890, "99.00 percentile latency (ns)": 329596082093, "99.90 percentile latency (ns)": 332597851570, "Max latency (ns)": 332923599780, "Mean latency (ns)": 166387748952, "Min duration satisfied": "Yes", "Min latency (ns)": 38707316, "Min queries satisfied": "Yes", "Mode": "Performance", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 73.8187, "Scenario": "offline", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "40 GB", "accelerator_memory_configuration": "HBM2", "accelerator_model_name": "NVIDIA A100-PCIE-40GB", "accelerator_on-chip_memories": "", "accelerators_per_node": 2, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7792, "characteristics.mean": 0.8536, "characteristics.samples_per_second": 73.8187, "characteristics.samples_per_second.normalized_per_core": 36.90935, "characteristics.samples_per_second.normalized_per_processor": 36.90935, "characteristics.tumor core": 0.8691, "characteristics.whole tumor": 0.9125, "ck_system": "R7525_A100x2_TRT", "ck_used": true, "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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.0, "formal_model_link": "", "framework": "TensorRT 7.2, CUDA 11.0, cuDNN 8.0.2, cuBLAS 11.2.0, libjemalloc2, cub 1.8.0, tensorrt-laboratory mlperf branch", "host_memory_capacity": "512 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 120, "host_processor_frequency": "2.50GHz", "host_processor_interconnect": "", "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "1.84 TB", "host_storage_type": "NVMe", "hw_notes": "", "informal_model": "3d-unet-99", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 1, "mlperf_version": 0.7, "normalize_cores": 2, "normalize_processors": 2, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/results/R7525_A100x2_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "docker 19.03.12, python 3.6.8, gcc 5.5.0, onnx 1.3.0, tensorflow 1.13.1, pytorch 1.1.0, torchvision 0.3.0, pycuda 2019.1, sacrebleu 1.3.3, SimpleJSON, OpenCV 4.1.1; GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 16, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 24576, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/R7525_A100x2_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 80, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "4445b917773b9b99", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 166371893266, "90.00 percentile latency (ns)": 299611520305, "95.00 percentile latency (ns)": 316270364226, "97.00 percentile latency (ns)": 322925475890, "99.00 percentile latency (ns)": 329596082093, "99.90 percentile latency (ns)": 332597851570, "Max latency (ns)": 332923599780, "Mean latency (ns)": 166387748952, "Min duration satisfied": "Yes", "Min latency (ns)": 38707316, "Min queries satisfied": "Yes", "Mode": "Performance", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 73.8187, "Scenario": "offline", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "40 GB", "accelerator_memory_configuration": "HBM2", "accelerator_model_name": "NVIDIA A100-PCIE-40GB", "accelerator_on-chip_memories": "", "accelerators_per_node": 2, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7792, "characteristics.mean": 0.8536, "characteristics.samples_per_second": 73.8187, "characteristics.samples_per_second.normalized_per_core": 36.90935, "characteristics.samples_per_second.normalized_per_processor": 36.90935, "characteristics.tumor core": 0.8691, "characteristics.whole tumor": 0.9125, "ck_system": "R7525_A100x2_TRT", "ck_used": true, "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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.9, "formal_model_link": "", "framework": "TensorRT 7.2, CUDA 11.0, cuDNN 8.0.2, cuBLAS 11.2.0, libjemalloc2, cub 1.8.0, tensorrt-laboratory mlperf branch", "host_memory_capacity": "512 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 120, "host_processor_frequency": "2.50GHz", "host_processor_interconnect": "", "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "1.84 TB", "host_storage_type": "NVMe", "hw_notes": "", "informal_model": "3d-unet-99.9", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 1, "mlperf_version": 0.7, "normalize_cores": 2, "normalize_processors": 2, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/results/R7525_A100x2_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "docker 19.03.12, python 3.6.8, gcc 5.5.0, onnx 1.3.0, tensorflow 1.13.1, pytorch 1.1.0, torchvision 0.3.0, pycuda 2019.1, sacrebleu 1.3.3, SimpleJSON, OpenCV 4.1.1; GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 16, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 24576, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/R7525_A100x2_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 80, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "bebac64aefaeea4d", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 104993080274, "90.00 percentile latency (ns)": 187233410229, "95.00 percentile latency (ns)": 197542156445, "97.00 percentile latency (ns)": 201665758553, "99.00 percentile latency (ns)": 205777790382, "99.90 percentile latency (ns)": 207623775388, "Max latency (ns)": 207806821608, "Mean latency (ns)": 104714418588, "Min duration satisfied": "Yes", "Min latency (ns)": 114391576, "Min queries satisfied": "Yes", "Mode": "Performance", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 118.264, "Scenario": "offline", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "40 GB", "accelerator_memory_configuration": "HBM2", "accelerator_model_name": "NVIDIA A100-PCIE-40GB", "accelerator_on-chip_memories": "", "accelerators_per_node": 3, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7787, "characteristics.mean": 0.85317, "characteristics.samples_per_second": 118.264, "characteristics.samples_per_second.normalized_per_core": 39.42133333333333, "characteristics.samples_per_second.normalized_per_processor": 39.42133333333333, "characteristics.tumor core": 0.8688, "characteristics.whole tumor": 0.912, "ck_system": "R7525_A100x3_TRT", "ck_used": true, "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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.0, "formal_model_link": "", "framework": "TensorRT 7.2, CUDA 11.0, cuDNN 8.0.2, cuBLAS 11.2.0, libjemalloc2, cub 1.8.0, tensorrt-laboratory mlperf branch", "host_memory_capacity": "512 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 120, "host_processor_frequency": "2.40GHz", "host_processor_interconnect": "", "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "1.5 TB", "host_storage_type": "NVMe", "hw_notes": "", "informal_model": "3d-unet-99", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 1, "mlperf_version": 0.7, "normalize_cores": 3, "normalize_processors": 3, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/results/R7525_A100x3_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "docker 19.03.12, python 3.6.8, gcc 5.5.0, onnx 1.3.0, tensorflow 1.13.1, pytorch 1.1.0, torchvision 0.3.0, pycuda 2019.1, sacrebleu 1.3.3, SimpleJSON, OpenCV 4.1.1; GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 16, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 24576, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "preview", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/R7525_A100x3_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 142, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "e08cd4aa4a7cdfdf", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 104993080274, "90.00 percentile latency (ns)": 187233410229, "95.00 percentile latency (ns)": 197542156445, "97.00 percentile latency (ns)": 201665758553, "99.00 percentile latency (ns)": 205777790382, "99.90 percentile latency (ns)": 207623775388, "Max latency (ns)": 207806821608, "Mean latency (ns)": 104714418588, "Min duration satisfied": "Yes", "Min latency (ns)": 114391576, "Min queries satisfied": "Yes", "Mode": "Performance", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 118.264, "Scenario": "offline", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "40 GB", "accelerator_memory_configuration": "HBM2", "accelerator_model_name": "NVIDIA A100-PCIE-40GB", "accelerator_on-chip_memories": "", "accelerators_per_node": 3, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7787, "characteristics.mean": 0.85317, "characteristics.samples_per_second": 118.264, "characteristics.samples_per_second.normalized_per_core": 39.42133333333333, "characteristics.samples_per_second.normalized_per_processor": 39.42133333333333, "characteristics.tumor core": 0.8688, "characteristics.whole tumor": 0.912, "ck_system": "R7525_A100x3_TRT", "ck_used": true, "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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.9, "formal_model_link": "", "framework": "TensorRT 7.2, CUDA 11.0, cuDNN 8.0.2, cuBLAS 11.2.0, libjemalloc2, cub 1.8.0, tensorrt-laboratory mlperf branch", "host_memory_capacity": "512 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 120, "host_processor_frequency": "2.40GHz", "host_processor_interconnect": "", "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "1.5 TB", "host_storage_type": "NVMe", "hw_notes": "", "informal_model": "3d-unet-99.9", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 1, "mlperf_version": 0.7, "normalize_cores": 3, "normalize_processors": 3, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/results/R7525_A100x3_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "docker 19.03.12, python 3.6.8, gcc 5.5.0, onnx 1.3.0, tensorflow 1.13.1, pytorch 1.1.0, torchvision 0.3.0, pycuda 2019.1, sacrebleu 1.3.3, SimpleJSON, OpenCV 4.1.1; GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 16, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 24576, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "preview", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/R7525_A100x3_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 142, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "5547c4a4c4b435c8", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 75702174826, "90.00 percentile latency (ns)": 136119805938, "95.00 percentile latency (ns)": 143672484081, "97.00 percentile latency (ns)": 146680205724, "99.00 percentile latency (ns)": 149701797540, "99.90 percentile latency (ns)": 151047630696, "Max latency (ns)": 151197537249, "Mean latency (ns)": 75701247925, "Min duration satisfied": "Yes", "Min latency (ns)": 214076147, "Min queries satisfied": "Yes", "Mode": "Performance", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 162.542, "Scenario": "offline", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "48 GB", "accelerator_memory_configuration": "GDDR6", "accelerator_model_name": "NVIDIA Quadro RTX 8000", "accelerator_on-chip_memories": "", "accelerators_per_node": 10, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7793, "characteristics.mean": 0.85367, "characteristics.samples_per_second": 162.542, "characteristics.samples_per_second.normalized_per_core": 16.2542, "characteristics.samples_per_second.normalized_per_processor": 16.2542, "characteristics.tumor core": 0.8692, "characteristics.whole tumor": 0.9125, "ck_system": "DSS8440_QuadroRTX8000x10_TRT", "ck_used": true, "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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.0, "formal_model_link": "", "framework": "TensorRT 7.2, CUDA 11.0, cuDNN 8.0.2, cuBLAS 11.2.0, libjemalloc2, cub 1.8.0, tensorrt-laboratory mlperf branch", "host_memory_capacity": "768 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 120, "host_processor_frequency": "2.10GHz", "host_processor_interconnect": "", "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "1.84 TB", "host_storage_type": "NVMe", "hw_notes": "", "informal_model": "3d-unet-99", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 1, "mlperf_version": 0.7, "normalize_cores": 10, "normalize_processors": 10, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/results/DSS8440_QuadroRTX8000x10_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "docker 19.03.12, python 3.6.8, gcc 5.5.0, onnx 1.3.0, tensorflow 1.13.1, pytorch 1.1.0, torchvision 0.3.0, pycuda 2019.1, sacrebleu 1.3.3, SimpleJSON, OpenCV 4.1.1; GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 16, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 24576, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/DSS8440_QuadroRTX8000x10_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 163, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "c537f7d64617835c", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 75702174826, "90.00 percentile latency (ns)": 136119805938, "95.00 percentile latency (ns)": 143672484081, "97.00 percentile latency (ns)": 146680205724, "99.00 percentile latency (ns)": 149701797540, "99.90 percentile latency (ns)": 151047630696, "Max latency (ns)": 151197537249, "Mean latency (ns)": 75701247925, "Min duration satisfied": "Yes", "Min latency (ns)": 214076147, "Min queries satisfied": "Yes", "Mode": "Performance", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 162.542, "Scenario": "offline", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "48 GB", "accelerator_memory_configuration": "GDDR6", "accelerator_model_name": "NVIDIA Quadro RTX 8000", "accelerator_on-chip_memories": "", "accelerators_per_node": 10, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7793, "characteristics.mean": 0.85367, "characteristics.samples_per_second": 162.542, "characteristics.samples_per_second.normalized_per_core": 16.2542, "characteristics.samples_per_second.normalized_per_processor": 16.2542, "characteristics.tumor core": 0.8692, "characteristics.whole tumor": 0.9125, "ck_system": "DSS8440_QuadroRTX8000x10_TRT", "ck_used": true, "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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.9, "formal_model_link": "", "framework": "TensorRT 7.2, CUDA 11.0, cuDNN 8.0.2, cuBLAS 11.2.0, libjemalloc2, cub 1.8.0, tensorrt-laboratory mlperf branch", "host_memory_capacity": "768 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 120, "host_processor_frequency": "2.10GHz", "host_processor_interconnect": "", "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "1.84 TB", "host_storage_type": "NVMe", "hw_notes": "", "informal_model": "3d-unet-99.9", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 1, "mlperf_version": 0.7, "normalize_cores": 10, "normalize_processors": 10, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/results/DSS8440_QuadroRTX8000x10_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "docker 19.03.12, python 3.6.8, gcc 5.5.0, onnx 1.3.0, tensorflow 1.13.1, pytorch 1.1.0, torchvision 0.3.0, pycuda 2019.1, sacrebleu 1.3.3, SimpleJSON, OpenCV 4.1.1; GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 16, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 24576, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/DSS8440_QuadroRTX8000x10_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 163, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "5ca5c936ce45e5b6", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 105165571126, "90.00 percentile latency (ns)": 189683361931, "95.00 percentile latency (ns)": 200235664247, "97.00 percentile latency (ns)": 204427364925, "99.00 percentile latency (ns)": 208663451512, "99.90 percentile latency (ns)": 210561350701, "Max latency (ns)": 210802704626, "Mean latency (ns)": 105251194121, "Min duration satisfied": "Yes", "Min latency (ns)": 447590507, "Min queries satisfied": "Yes", "Mode": "Performance", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 116.583, "Scenario": "offline", "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": 16, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7793, "characteristics.mean": 0.85367, "characteristics.samples_per_second": 116.583, "characteristics.samples_per_second.normalized_per_core": 7.2864375, "characteristics.samples_per_second.normalized_per_processor": 7.2864375, "characteristics.tumor core": 0.8692, "characteristics.whole tumor": 0.9125, "ck_system": "DSS8440_T4x16_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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.0, "formal_model_link": "", "framework": "TensorRT 7.2, CUDA 11.0 Update 1", "host_memory_capacity": "384 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": 2, "host_storage_capacity": "1.84 TB", "host_storage_type": "NVMe", "hw_notes": "ECC off", "informal_model": "3d-unet-99", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 1, "mlperf_version": 0.7, "normalize_cores": 16, "normalize_processors": 16, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/results/DSS8440_T4x16_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "TensorRT 7.2, CUDA 11.0 Update 1, cuDNN 8.0.2, DALI 0.25.0; GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 16, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 24576, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/DSS8440_T4x16_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 116, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "7941e7e4529fe930", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 105165571126, "90.00 percentile latency (ns)": 189683361931, "95.00 percentile latency (ns)": 200235664247, "97.00 percentile latency (ns)": 204427364925, "99.00 percentile latency (ns)": 208663451512, "99.90 percentile latency (ns)": 210561350701, "Max latency (ns)": 210802704626, "Mean latency (ns)": 105251194121, "Min duration satisfied": "Yes", "Min latency (ns)": 447590507, "Min queries satisfied": "Yes", "Mode": "Performance", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 116.583, "Scenario": "offline", "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": 16, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7793, "characteristics.mean": 0.85367, "characteristics.samples_per_second": 116.583, "characteristics.samples_per_second.normalized_per_core": 7.2864375, "characteristics.samples_per_second.normalized_per_processor": 7.2864375, "characteristics.tumor core": 0.8692, "characteristics.whole tumor": 0.9125, "ck_system": "DSS8440_T4x16_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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.9, "formal_model_link": "", "framework": "TensorRT 7.2, CUDA 11.0 Update 1", "host_memory_capacity": "384 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": 2, "host_storage_capacity": "1.84 TB", "host_storage_type": "NVMe", "hw_notes": "ECC off", "informal_model": "3d-unet-99.9", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 1, "mlperf_version": 0.7, "normalize_cores": 16, "normalize_processors": 16, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/results/DSS8440_T4x16_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "TensorRT 7.2, CUDA 11.0 Update 1, cuDNN 8.0.2, DALI 0.25.0; GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 16, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 24576, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/DSS8440_T4x16_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 116, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "e65a6fe4593bf885", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 439029879055, "90.00 percentile latency (ns)": 795265914473, "95.00 percentile latency (ns)": 839829741177, "97.00 percentile latency (ns)": 857680650678, "99.00 percentile latency (ns)": 875506025080, "99.90 percentile latency (ns)": 883495076960, "Max latency (ns)": 884385109605, "Mean latency (ns)": 439585147498, "Min duration satisfied": "Yes", "Min latency (ns)": 396707820, "Min queries satisfied": "Yes", "Mode": "Performance", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 27.7888, "Scenario": "offline", "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": 4, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7793, "characteristics.mean": 0.85367, "characteristics.samples_per_second": 27.7888, "characteristics.samples_per_second.normalized_per_core": 6.9472, "characteristics.samples_per_second.normalized_per_processor": 6.9472, "characteristics.tumor core": 0.8692, "characteristics.whole tumor": 0.9125, "ck_system": "R7515_T4x4_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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.0, "formal_model_link": "", "framework": "TensorRT 7.2.0.14, CUDA 11.0.207", "host_memory_capacity": "256 GB", "host_memory_configuration": "DDR-4", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 120, "host_processor_frequency": "2.0GHz", "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": "NVMe SSD", "hw_notes": "ECC on", "informal_model": "3d-unet-99", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 1, "mlperf_version": 0.7, "normalize_cores": 4, "normalize_processors": 4, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/results/R7515_T4x4_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "TensorRT 7.2.0.14, CUDA 11.0.27, cuDNN 8.0.2, DALI 0.25.0; GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 16, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 24576, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/R7515_T4x4_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 60, "task": "image segmentation", "task2": "image segmentation", "total_cores": 120, "uid": "4aad93374f6a8d48", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 439029879055, "90.00 percentile latency (ns)": 795265914473, "95.00 percentile latency (ns)": 839829741177, "97.00 percentile latency (ns)": 857680650678, "99.00 percentile latency (ns)": 875506025080, "99.90 percentile latency (ns)": 883495076960, "Max latency (ns)": 884385109605, "Mean latency (ns)": 439585147498, "Min duration satisfied": "Yes", "Min latency (ns)": 396707820, "Min queries satisfied": "Yes", "Mode": "Performance", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 27.7888, "Scenario": "offline", "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": 4, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7793, "characteristics.mean": 0.85367, "characteristics.samples_per_second": 27.7888, "characteristics.samples_per_second.normalized_per_core": 6.9472, "characteristics.samples_per_second.normalized_per_processor": 6.9472, "characteristics.tumor core": 0.8692, "characteristics.whole tumor": 0.9125, "ck_system": "R7515_T4x4_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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.9, "formal_model_link": "", "framework": "TensorRT 7.2.0.14, CUDA 11.0.207", "host_memory_capacity": "256 GB", "host_memory_configuration": "DDR-4", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 120, "host_processor_frequency": "2.0GHz", "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": "NVMe SSD", "hw_notes": "ECC on", "informal_model": "3d-unet-99.9", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 1, "mlperf_version": 0.7, "normalize_cores": 4, "normalize_processors": 4, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/results/R7515_T4x4_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "TensorRT 7.2.0.14, CUDA 11.0.27, cuDNN 8.0.2, DALI 0.25.0; GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 16, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 24576, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/R7515_T4x4_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 60, "task": "image segmentation", "task2": "image segmentation", "total_cores": 120, "uid": "36daafea05d0074e", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 139365263296, "90.00 percentile latency (ns)": 251386279249, "95.00 percentile latency (ns)": 265392960018, "97.00 percentile latency (ns)": 271003651647, "99.00 percentile latency (ns)": 276610779249, "99.90 percentile latency (ns)": 279153253073, "Max latency (ns)": 279423488938, "Mean latency (ns)": 139503381937, "Min duration satisfied": "Yes", "Min latency (ns)": 414878583, "Min queries satisfied": "Yes", "Mode": "Performance", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 87.9525, "Scenario": "offline", "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": 12, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7793, "characteristics.mean": 0.85367, "characteristics.samples_per_second": 87.9525, "characteristics.samples_per_second.normalized_per_core": 7.329375, "characteristics.samples_per_second.normalized_per_processor": 7.329375, "characteristics.tumor core": 0.8692, "characteristics.whole tumor": 0.9125, "ck_system": "DSS8440_T4x12_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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.0, "formal_model_link": "", "framework": "TensorRT 7.2, CUDA 11.0 Update 1", "host_memory_capacity": "768 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": 2, "host_storage_capacity": "1 TB", "host_storage_type": "NVMe SSD", "hw_notes": "ECC off", "informal_model": "3d-unet-99", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 1, "mlperf_version": 0.7, "normalize_cores": 12, "normalize_processors": 12, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/results/DSS8440_T4x12_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "TensorRT 7.2, CUDA 11.0 Update 1, cuDNN 8.0.2, DALI 0.25.0; GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 16, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 24576, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/DSS8440_T4x12_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 88, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "814686bcebf7fb6a", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 139365263296, "90.00 percentile latency (ns)": 251386279249, "95.00 percentile latency (ns)": 265392960018, "97.00 percentile latency (ns)": 271003651647, "99.00 percentile latency (ns)": 276610779249, "99.90 percentile latency (ns)": 279153253073, "Max latency (ns)": 279423488938, "Mean latency (ns)": 139503381937, "Min duration satisfied": "Yes", "Min latency (ns)": 414878583, "Min queries satisfied": "Yes", "Mode": "Performance", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 87.9525, "Scenario": "offline", "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": 12, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7793, "characteristics.mean": 0.85367, "characteristics.samples_per_second": 87.9525, "characteristics.samples_per_second.normalized_per_core": 7.329375, "characteristics.samples_per_second.normalized_per_processor": 7.329375, "characteristics.tumor core": 0.8692, "characteristics.whole tumor": 0.9125, "ck_system": "DSS8440_T4x12_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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.9, "formal_model_link": "", "framework": "TensorRT 7.2, CUDA 11.0 Update 1", "host_memory_capacity": "768 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": 2, "host_storage_capacity": "1 TB", "host_storage_type": "NVMe SSD", "hw_notes": "ECC off", "informal_model": "3d-unet-99.9", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 1, "mlperf_version": 0.7, "normalize_cores": 12, "normalize_processors": 12, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/results/DSS8440_T4x12_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "TensorRT 7.2, CUDA 11.0 Update 1, cuDNN 8.0.2, DALI 0.25.0; GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 16, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 24576, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/DSS8440_T4x12_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 88, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "f84ac8b87c891000", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 280605627123, "90.00 percentile latency (ns)": 498979761549, "95.00 percentile latency (ns)": 525899953760, "97.00 percentile latency (ns)": 536668845178, "99.00 percentile latency (ns)": 547643537718, "99.90 percentile latency (ns)": 552549346513, "Max latency (ns)": 553078571969, "Mean latency (ns)": 279641226796, "Min duration satisfied": "Yes", "Min latency (ns)": 151318232, "Min queries satisfied": "Yes", "Mode": "Submission", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 44.4349, "Scenario": "offline", "accelerator_frequency": "1770MHz", "accelerator_host_interconnect": "PCI Express 3.0", "accelerator_interconnect": "PCI Express 3.0", "accelerator_interconnect_topology": "2 Accelerators per CPU", "accelerator_memory_capacity": "24 GB", "accelerator_memory_configuration": "GDDR6", "accelerator_model_name": "NVIDIA Quadro RTX 6000", "accelerator_on-chip_memories": "", "accelerators_per_node": 4, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7793, "characteristics.mean": 0.85367, "characteristics.samples_per_second": 44.4349, "characteristics.samples_per_second.normalized_per_core": 11.108725, "characteristics.samples_per_second.normalized_per_processor": 11.108725, "characteristics.tumor core": 0.8692, "characteristics.whole tumor": 0.9125, "ck_system": "C4140_QuadroRTX6000x4_TRT", "ck_used": true, "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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.0, "formal_model_link": "", "framework": "TensorRT 7.2, CUDA 11.0, cuDNN 8.0.2, cuBLAS 11.2.0, libjemalloc2, cub 1.8.0, tensorrt-laboratory mlperf branch", "host_memory_capacity": "384 GB", "host_memory_configuration": "6x16GB DDR4-2666 HMA82GR7AFR8N-VK RDIMM ECC", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "1.25MB+20MB+27.5MB", "host_processor_core_count": 120, "host_processor_frequency": "2.40GHz", "host_processor_interconnect": "Ultra Path Interconnect", "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "1.6 TB (1x1.6TB Dell Express Flash PM1725a 1.6TB AIC)", "host_storage_type": "3D-TLC Solid State with PCIe NVME x8 Interface", "hw_notes": "ECC off. RTX6000 is available as a special config thru Dell DSS or OEM for PowerEdge C4140", "informal_model": "3d-unet-99", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 1, "mlperf_version": 0.7, "normalize_cores": 4, "normalize_processors": 4, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/results/C4140_QuadroRTX6000x4_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "docker 19.03.12, python 3.6.8, gcc 5.5.0, onnx 1.3.0, tensorflow 1.13.1, pytorch 1.1.0, torchvision 0.3.0, pycuda 2019.1, sacrebleu 1.3.3, SimpleJSON, OpenCV 4.1.1; GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 16, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 24576, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/C4140_QuadroRTX6000x4_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 42, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "a32da743254d48ac", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 280605627123, "90.00 percentile latency (ns)": 498979761549, "95.00 percentile latency (ns)": 525899953760, "97.00 percentile latency (ns)": 536668845178, "99.00 percentile latency (ns)": 547643537718, "99.90 percentile latency (ns)": 552549346513, "Max latency (ns)": 553078571969, "Mean latency (ns)": 279641226796, "Min duration satisfied": "Yes", "Min latency (ns)": 151318232, "Min queries satisfied": "Yes", "Mode": "Submission", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 44.4349, "Scenario": "offline", "accelerator_frequency": "1770MHz", "accelerator_host_interconnect": "PCI Express 3.0", "accelerator_interconnect": "PCI Express 3.0", "accelerator_interconnect_topology": "2 Accelerators per CPU", "accelerator_memory_capacity": "24 GB", "accelerator_memory_configuration": "GDDR6", "accelerator_model_name": "NVIDIA Quadro RTX 6000", "accelerator_on-chip_memories": "", "accelerators_per_node": 4, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7793, "characteristics.mean": 0.85367, "characteristics.samples_per_second": 44.4349, "characteristics.samples_per_second.normalized_per_core": 11.108725, "characteristics.samples_per_second.normalized_per_processor": 11.108725, "characteristics.tumor core": 0.8692, "characteristics.whole tumor": 0.9125, "ck_system": "C4140_QuadroRTX6000x4_TRT", "ck_used": true, "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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.9, "formal_model_link": "", "framework": "TensorRT 7.2, CUDA 11.0, cuDNN 8.0.2, cuBLAS 11.2.0, libjemalloc2, cub 1.8.0, tensorrt-laboratory mlperf branch", "host_memory_capacity": "384 GB", "host_memory_configuration": "6x16GB DDR4-2666 HMA82GR7AFR8N-VK RDIMM ECC", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "1.25MB+20MB+27.5MB", "host_processor_core_count": 120, "host_processor_frequency": "2.40GHz", "host_processor_interconnect": "Ultra Path Interconnect", "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "1.6 TB (1x1.6TB Dell Express Flash PM1725a 1.6TB AIC)", "host_storage_type": "3D-TLC Solid State with PCIe NVME x8 Interface", "hw_notes": "ECC off. RTX6000 is available as a special config thru Dell DSS or OEM for PowerEdge C4140", "informal_model": "3d-unet-99.9", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 1, "mlperf_version": 0.7, "normalize_cores": 4, "normalize_processors": 4, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/results/C4140_QuadroRTX6000x4_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "docker 19.03.12, python 3.6.8, gcc 5.5.0, onnx 1.3.0, tensorflow 1.13.1, pytorch 1.1.0, torchvision 0.3.0, pycuda 2019.1, sacrebleu 1.3.3, SimpleJSON, OpenCV 4.1.1; GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 16, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 24576, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/C4140_QuadroRTX6000x4_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 42, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "2af497831c314a8d", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 228915073163, "90.00 percentile latency (ns)": 412412118288, "95.00 percentile latency (ns)": 435350430538, "97.00 percentile latency (ns)": 444538579781, "99.00 percentile latency (ns)": 453712207295, "99.90 percentile latency (ns)": 457842322513, "Max latency (ns)": 458287965290, "Mean latency (ns)": 228972232761, "Min duration satisfied": "Yes", "Min latency (ns)": 163046835, "Min queries satisfied": "Yes", "Mode": "Performance", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 53.6257, "Scenario": "offline", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "48 GB", "accelerator_memory_configuration": "GDDR6", "accelerator_model_name": "NVIDIA Quadro RTX 8000", "accelerator_on-chip_memories": "", "accelerators_per_node": 3, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7793, "characteristics.mean": 0.85367, "characteristics.samples_per_second": 53.6257, "characteristics.samples_per_second.normalized_per_core": 17.875233333333334, "characteristics.samples_per_second.normalized_per_processor": 17.875233333333334, "characteristics.tumor core": 0.8692, "characteristics.whole tumor": 0.9125, "ck_system": "R7525_QuadroRTX8000x3_TRT", "ck_used": false, "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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.0, "formal_model_link": "", "framework": "TensorRT 7.2, CUDA 11.0, cuDNN 8.0.2, cuBLAS 11.2.0, libjemalloc2, cub 1.8.0, tensorrt-laboratory mlperf branch", "host_memory_capacity": "512 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 32, "host_processor_frequency": "2.50GHz", "host_processor_interconnect": "", "host_processor_model_name": "AMD EPYC 7502", "host_processors_per_node": 2, "host_storage_capacity": "1.84 TB", "host_storage_type": "NVMe", "hw_notes": "", "informal_model": "3d-unet-99", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 1, "mlperf_version": 0.7, "normalize_cores": 3, "normalize_processors": 3, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/results/R7525_QuadroRTX8000x3_TRT", "number_of_nodes": 1, "operating_system": "CentOS Linux release 8.1.1911", "other_software_stack": "docker 19.03.12, python 3.6.8, gcc 5.5.0, onnx 1.3.0, tensorflow 1.13.1, pytorch 1.1.0, torchvision 0.3.0, pycuda 2019.1, sacrebleu 1.3.3, SimpleJSON, OpenCV 4.1.1", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 16, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 24576, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "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/R7525_QuadroRTX8000x3_TRT", "system_name": "Dell EMC PowerEdge R7525 (3x Quadro RTX 8000)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 54, "task": "image segmentation", "task2": "image segmentation", "total_cores": 64, "uid": "a47c5a65ab32f2c8", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 228915073163, "90.00 percentile latency (ns)": 412412118288, "95.00 percentile latency (ns)": 435350430538, "97.00 percentile latency (ns)": 444538579781, "99.00 percentile latency (ns)": 453712207295, "99.90 percentile latency (ns)": 457842322513, "Max latency (ns)": 458287965290, "Mean latency (ns)": 228972232761, "Min duration satisfied": "Yes", "Min latency (ns)": 163046835, "Min queries satisfied": "Yes", "Mode": "Performance", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 53.6257, "Scenario": "offline", "accelerator_frequency": "", "accelerator_host_interconnect": "", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "48 GB", "accelerator_memory_configuration": "GDDR6", "accelerator_model_name": "NVIDIA Quadro RTX 8000", "accelerator_on-chip_memories": "", "accelerators_per_node": 3, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7793, "characteristics.mean": 0.85367, "characteristics.samples_per_second": 53.6257, "characteristics.samples_per_second.normalized_per_core": 17.875233333333334, "characteristics.samples_per_second.normalized_per_processor": 17.875233333333334, "characteristics.tumor core": 0.8692, "characteristics.whole tumor": 0.9125, "ck_system": "R7525_QuadroRTX8000x3_TRT", "ck_used": false, "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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.9, "formal_model_link": "", "framework": "TensorRT 7.2, CUDA 11.0, cuDNN 8.0.2, cuBLAS 11.2.0, libjemalloc2, cub 1.8.0, tensorrt-laboratory mlperf branch", "host_memory_capacity": "512 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 32, "host_processor_frequency": "2.50GHz", "host_processor_interconnect": "", "host_processor_model_name": "AMD EPYC 7502", "host_processors_per_node": 2, "host_storage_capacity": "1.84 TB", "host_storage_type": "NVMe", "hw_notes": "", "informal_model": "3d-unet-99.9", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 1, "mlperf_version": 0.7, "normalize_cores": 3, "normalize_processors": 3, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/results/R7525_QuadroRTX8000x3_TRT", "number_of_nodes": 1, "operating_system": "CentOS Linux release 8.1.1911", "other_software_stack": "docker 19.03.12, python 3.6.8, gcc 5.5.0, onnx 1.3.0, tensorflow 1.13.1, pytorch 1.1.0, torchvision 0.3.0, pycuda 2019.1, sacrebleu 1.3.3, SimpleJSON, OpenCV 4.1.1", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 16, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 24576, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "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/R7525_QuadroRTX8000x3_TRT", "system_name": "Dell EMC PowerEdge R7525 (3x Quadro RTX 8000)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 54, "task": "image segmentation", "task2": "image segmentation", "total_cores": 64, "uid": "c43ac95683930ccf", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 427062520121, "90.00 percentile latency (ns)": 770867144749, "95.00 percentile latency (ns)": 813868631898, "97.00 percentile latency (ns)": 830944979277, "99.00 percentile latency (ns)": 848261896830, "99.90 percentile latency (ns)": 855877598867, "Max latency (ns)": 856707872332, "Mean latency (ns)": 427285933319, "Min duration satisfied": "Yes", "Min latency (ns)": 414440064, "Min queries satisfied": "Yes", "Mode": "Performance", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 28.6866, "Scenario": "offline", "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": 4, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7793, "characteristics.mean": 0.85367, "characteristics.samples_per_second": 28.6866, "characteristics.samples_per_second.normalized_per_core": 7.17165, "characteristics.samples_per_second.normalized_per_processor": 7.17165, "characteristics.tumor core": 0.8692, "characteristics.whole tumor": 0.9125, "ck_system": "R740_T4x4_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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.0, "formal_model_link": "", "framework": "TensorRT 7.2.0.14, CUDA 11.0.207", "host_memory_capacity": "384 GB", "host_memory_configuration": "DDR-4", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 120, "host_processor_frequency": "3.0GHz", "host_processor_interconnect": "", "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "3.84 TB", "host_storage_type": "SSD", "hw_notes": "ECC on", "informal_model": "3d-unet-99", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 1, "mlperf_version": 0.7, "normalize_cores": 4, "normalize_processors": 4, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/results/R740_T4x4_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "TensorRT 7.2.0.14, CUDA 11.0.207, cuDNN 8.0.2, DALI 0.25.0; GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 16, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 24576, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/R740_T4x4_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 60, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "dca4956289ea3772", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" }, { "50.00 percentile latency (ns)": 427062520121, "90.00 percentile latency (ns)": 770867144749, "95.00 percentile latency (ns)": 813868631898, "97.00 percentile latency (ns)": 830944979277, "99.00 percentile latency (ns)": 848261896830, "99.90 percentile latency (ns)": 855877598867, "Max latency (ns)": 856707872332, "Mean latency (ns)": 427285933319, "Min duration satisfied": "Yes", "Min latency (ns)": 414440064, "Min queries satisfied": "Yes", "Mode": "Performance", "Result is": "VALID", "SUT name": "LWIS_Server", "Samples per second": 28.6866, "Scenario": "offline", "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": 4, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.enhancing tumor": 0.7793, "characteristics.mean": 0.85367, "characteristics.samples_per_second": 28.6866, "characteristics.samples_per_second.normalized_per_core": 7.17165, "characteristics.samples_per_second.normalized_per_processor": 7.17165, "characteristics.tumor core": 0.8692, "characteristics.whole tumor": 0.9125, "ck_system": "R740_T4x4_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, "division": "closed", "formal_model": "3d-unet", "formal_model_accuracy": 99.9, "formal_model_link": "", "framework": "TensorRT 7.2.0.14, CUDA 11.0.207", "host_memory_capacity": "384 GB", "host_memory_configuration": "DDR-4", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "", "host_processor_core_count": 120, "host_processor_frequency": "3.0GHz", "host_processor_interconnect": "", "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "3.84 TB", "host_storage_type": "SSD", "hw_notes": "ECC on", "informal_model": "3d-unet-99.9", "input_data_types": "int8", "key.accuracy": "characteristics.mean", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 60000, "min_query_count": 1, "mlperf_version": 0.7, "normalize_cores": 4, "normalize_processors": 4, "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/code", "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/DellEMC/results/R740_T4x4_TRT", "number_of_nodes": 1, "operating_system": "Ubuntu 18.04.5 LTS (Linux-5.4.0-1055-azure-x86_64-with-Ubuntu-18.04-bionic)", "other_software_stack": "TensorRT 7.2.0.14, CUDA 11.0.207, cuDNN 8.0.2, DALI 0.25.0; GCC 7.5.0; Python 3.7.10", "performance_issue_same": true, "performance_issue_same_index": 0, "performance_issue_unique": true, "performance_sample_count": 16, "print_timestamps": true, "problem": false, "qsl_rng_seed": 12786827339337101903, "retraining": "N", "sample_index_rng_seed": 12640797754436136668, "samples_per_query": 24576, "schedule_rng_seed": 3135815929913719677, "starting_weights_filename": "224_224_160_dyanmic_bs.onnx", "status": "available", "submitter": "DellEMC", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC", "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/R740_T4x4_TRT", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "datacenter", "target_latency (ns)": 0, "target_qps": 60, "task": "image segmentation", "task2": "image segmentation", "total_cores": 240, "uid": "b359218482fb44a5", "use_accelerator": true, "weight_data_types": "int8,fp16", "weight_transformations": "quantization, affine fusion" } ]