[
  {
    "50.00 percentile latency (ns)": 43615635845,
    "90.00 percentile latency (ns)": 63369538657,
    "95.00 percentile latency (ns)": 64858123348,
    "97.00 percentile latency (ns)": 65470286532,
    "99.00 percentile latency (ns)": 65962604738,
    "99.90 percentile latency (ns)": 66110670110,
    "Max latency (ns)": 66110670110,
    "Mean latency (ns)": 40318863729,
    "Min duration satisfied": "Yes",
    "Min latency (ns)": 931576017,
    "Min queries satisfied": "Yes",
    "Mode": "Performance",
    "Result is": "VALID",
    "SUT name": "GNMT SERVER",
    "Samples per second": 1417.62,
    "Scenario": "offline",
    "accelerator_frequency": "-",
    "accelerator_host_interconnect": "PCIe 3.0",
    "accelerator_interconnect": "PCIe 3.0 ",
    "accelerator_interconnect_topology": "-",
    "accelerator_memory_capacity": "16 GB",
    "accelerator_memory_configuration": "GDDR6",
    "accelerator_model_name": "Tesla T4",
    "accelerator_on-chip_memories": 1,
    "accelerators_per_node": 4,
    "accuracy_log_probability": 0,
    "accuracy_log_rng_seed": 0,
    "characteristics.blue": 23.8,
    "characteristics.samples_per_second": 1417.62,
    "characteristics.samples_per_second.normalized_per_core": 354.405,
    "characteristics.samples_per_second.normalized_per_processor": 354.405,
    "ck_system": "R740_T4x4_tensorrt",
    "ck_used": true,
    "cooling": "-",
    "dataset": "WMT E-G",
    "dataset_link": "",
    "dim_x_default": "characteristics.samples_per_second",
    "dim_x_maximize": true,
    "dim_y_default": "characteristics.blue",
    "dim_y_maximize": true,
    "division": "closed",
    "formal_model": "nmt",
    "formal_model_accuracy": 99.0,
    "formal_model_link": "",
    "framework": "TensorRT NGC 19.09",
    "host_memory_capacity": "384 GB",
    "host_memory_configuration": "-",
    "host_networking": "-",
    "host_networking_topology": "-",
    "host_processor_caches": "-",
    "host_processor_core_count": 120,
    "host_processor_frequency": "3.00GHz",
    "host_processor_interconnect": "UPI",
    "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor",
    "host_processors_per_node": 2,
    "host_storage_capacity": "110 GB",
    "host_storage_type": "SSD",
    "hw_notes": "none",
    "informal_model": "gnmt",
    "input_data_types": "UTF-8",
    "key.accuracy": "characteristics.blue",
    "max_async_queries": 1,
    "max_duration (ms)": 0,
    "max_query_count": 0,
    "min_duration (ms)": 60000,
    "min_query_count": 1,
    "mlperf_version": 0.5,
    "normalize_cores": 4,
    "normalize_processors": 4,
    "note_code": "https://github.com/mlcommons/inference_results_v0.5/tree/master/closed/DellEMC/code",
    "note_details": "https://github.com/mlcommons/inference_results_v0.5/tree/master/closed/DellEMC/results/R740_T4x4_tensorrt",
    "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 19.03.2; GCC 7.5.0; Python 3.7.10",
    "performance_issue_same": true,
    "performance_issue_same_index": 0,
    "performance_issue_unique": true,
    "performance_sample_count": 3903900,
    "print_timestamps": true,
    "problem": false,
    "qsl_rng_seed": 3133965575612453542,
    "retraining": "N",
    "sample_index_rng_seed": 665484352860916858,
    "samples_per_query": 93720,
    "schedule_rng_seed": 3622009729038561421,
    "starting_weights_filename": "ende_gnmt_model_4_layer/translate.ckpt.data-00000-of-00001",
    "status": "available",
    "submitter": "DellEMC",
    "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/DellEMC",
    "sw_notes": "nonePowered 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_tensorrt",
    "system_name": "Microsoft Corporation 7.0 (Virtual Machine)",
    "target_latency (ns)": 0,
    "target_qps": 1420,
    "task": "NMT",
    "task2": "nmt",
    "total_cores": 240,
    "uid": "a7004919437fb963",
    "use_accelerator": true,
    "weight_data_types": "int8,fp16",
    "weight_transformations": "quantization, affine fusion"
  }
]