[
  {
    "50.00 percentile latency (ns)": 1096787,
    "90.00 percentile latency (ns)": 1740217,
    "95.00 percentile latency (ns)": 2817589,
    "97.00 percentile latency (ns)": 4121111,
    "99.00 percentile latency (ns)": 7459610,
    "99.90 percentile latency (ns)": 53038707,
    "Completed samples per second": 275049.7,
    "Max latency (ns)": 67461304,
    "Mean latency (ns)": 1479991,
    "Min duration satisfied": "Yes",
    "Min latency (ns)": 207274,
    "Min queries satisfied": "Yes",
    "Mode": "Performance",
    "Performance constraints satisfied": "Yes",
    "Result is": "VALID",
    "SUT name": "DLRM SERVER",
    "Scenario": "server",
    "Scheduled samples per second": 275055.77,
    "accelerator_frequency": "585 MHz",
    "accelerator_host_interconnect": "PCIe 3.0 x16",
    "accelerator_interconnect": "",
    "accelerator_interconnect_topology": "",
    "accelerator_memory_capacity": "16 GB",
    "accelerator_memory_configuration": "GDDR6",
    "accelerator_model_name": "NVIDIA Tesla T4",
    "accelerator_on-chip_memories": "L1: 40x 64KB, L2: 4MB per chip",
    "accelerators_per_node": 16,
    "accuracy_log_probability": 0,
    "accuracy_log_rng_seed": 0,
    "accuracy_log_sampling_target": 0,
    "characteristics.scheduled_queries_per_second": 275055.77,
    "characteristics.scheduled_queries_per_second.normalized_per_core": 17190.985625,
    "characteristics.scheduled_queries_per_second.normalized_per_processor": 17190.985625,
    "ck_system": "GIGABYTE_G292-Z43_16xT4",
    "ck_used": true,
    "cooling": "Air-cooled",
    "dataset": "1TB Click Logs",
    "dataset_link": "",
    "dim_x_default": "characteristics.scheduled_queries_per_second",
    "dim_y_default": "characteristics.AUC",
    "dim_y_maximize": true,
    "division": "closed",
    "formal_model": "dlrm",
    "formal_model_accuracy": 99.9,
    "formal_model_link": "",
    "framework": "TensorRT 7.2, CUDA 11.0 Update 1",
    "host_memory_capacity": "1024 GB",
    "host_memory_configuration": "",
    "host_networking": "",
    "host_networking_topology": "",
    "host_processor_caches": "L3 cache 256MB",
    "host_processor_core_count": 120,
    "host_processor_frequency": "2.25 GHz",
    "host_processor_interconnect": "1x 16-20GT/s xGMI2",
    "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor",
    "host_processors_per_node": 2,
    "host_storage_capacity": "1.92 TB",
    "host_storage_type": "SSD",
    "hw_notes": "ECC off",
    "informal_model": "dlrm-99.9",
    "input_data_types": "int8",
    "max_async_queries": 0,
    "max_duration (ms)": 0,
    "max_query_count": 0,
    "min_duration (ms)": 60000,
    "min_query_count": 270336,
    "mlperf_version": 0.7,
    "normalize_cores": 16,
    "normalize_processors": 16,
    "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/Gigabyte/code",
    "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/Gigabyte/results/GIGABYTE_G292-Z43_16xT4",
    "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": 204800,
    "print_timestamps": true,
    "problem": false,
    "qsl_rng_seed": 12786827339337101903,
    "retraining": "N",
    "sample_index_rng_seed": 12640797754436136668,
    "samples_per_query": 1,
    "schedule_rng_seed": 3135815929913719677,
    "starting_weights_filename": "tb00_40M.pt",
    "status": "preview",
    "submitter": "GIGABYTE",
    "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Gigabyte",
    "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/GIGABYTE_G292-Z43_16xT4",
    "system_name": "Microsoft Corporation 7.0 (Virtual Machine)",
    "system_type": "datacenter",
    "target_latency (ns)": 30000000,
    "target_qps": 275000,
    "task": "recommendation",
    "task2": "recommendation",
    "total_cores": 240,
    "uid": "da3a8c434717a341",
    "use_accelerator": true,
    "weight_data_types": "int8,fp16",
    "weight_transformations": "quantization, affine fusion"
  },
  {
    "50.00 percentile latency (ns)": 1096787,
    "90.00 percentile latency (ns)": 1740217,
    "95.00 percentile latency (ns)": 2817589,
    "97.00 percentile latency (ns)": 4121111,
    "99.00 percentile latency (ns)": 7459610,
    "99.90 percentile latency (ns)": 53038707,
    "Completed samples per second": 275049.7,
    "Max latency (ns)": 67461304,
    "Mean latency (ns)": 1479991,
    "Min duration satisfied": "Yes",
    "Min latency (ns)": 207274,
    "Min queries satisfied": "Yes",
    "Mode": "Performance",
    "Performance constraints satisfied": "Yes",
    "Result is": "VALID",
    "SUT name": "DLRM SERVER",
    "Scenario": "server",
    "Scheduled samples per second": 275055.77,
    "accelerator_frequency": "585 MHz",
    "accelerator_host_interconnect": "PCIe 3.0 x16",
    "accelerator_interconnect": "",
    "accelerator_interconnect_topology": "",
    "accelerator_memory_capacity": "16 GB",
    "accelerator_memory_configuration": "GDDR6",
    "accelerator_model_name": "NVIDIA Tesla T4",
    "accelerator_on-chip_memories": "L1: 40x 64KB, L2: 4MB per chip",
    "accelerators_per_node": 16,
    "accuracy_log_probability": 0,
    "accuracy_log_rng_seed": 0,
    "accuracy_log_sampling_target": 0,
    "characteristics.scheduled_queries_per_second": 275055.77,
    "characteristics.scheduled_queries_per_second.normalized_per_core": 17190.985625,
    "characteristics.scheduled_queries_per_second.normalized_per_processor": 17190.985625,
    "ck_system": "GIGABYTE_G292-Z43_16xT4",
    "ck_used": true,
    "cooling": "Air-cooled",
    "dataset": "1TB Click Logs",
    "dataset_link": "",
    "dim_x_default": "characteristics.scheduled_queries_per_second",
    "dim_y_default": "characteristics.AUC",
    "dim_y_maximize": true,
    "division": "closed",
    "formal_model": "dlrm",
    "formal_model_accuracy": 99.0,
    "formal_model_link": "",
    "framework": "TensorRT 7.2, CUDA 11.0 Update 1",
    "host_memory_capacity": "1024 GB",
    "host_memory_configuration": "",
    "host_networking": "",
    "host_networking_topology": "",
    "host_processor_caches": "L3 cache 256MB",
    "host_processor_core_count": 120,
    "host_processor_frequency": "2.25 GHz",
    "host_processor_interconnect": "1x 16-20GT/s xGMI2",
    "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor",
    "host_processors_per_node": 2,
    "host_storage_capacity": "1.92 TB",
    "host_storage_type": "SSD",
    "hw_notes": "ECC off",
    "informal_model": "dlrm-99",
    "input_data_types": "int8",
    "max_async_queries": 0,
    "max_duration (ms)": 0,
    "max_query_count": 0,
    "min_duration (ms)": 60000,
    "min_query_count": 270336,
    "mlperf_version": 0.7,
    "normalize_cores": 16,
    "normalize_processors": 16,
    "note_code": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/Gigabyte/code",
    "note_details": "https://github.com/mlcommons/inference_results_v0.7/tree/master/closed/Gigabyte/results/GIGABYTE_G292-Z43_16xT4",
    "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": 204800,
    "print_timestamps": true,
    "problem": false,
    "qsl_rng_seed": 12786827339337101903,
    "retraining": "N",
    "sample_index_rng_seed": 12640797754436136668,
    "samples_per_query": 1,
    "schedule_rng_seed": 3135815929913719677,
    "starting_weights_filename": "tb00_40M.pt",
    "status": "preview",
    "submitter": "GIGABYTE",
    "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Gigabyte",
    "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/GIGABYTE_G292-Z43_16xT4",
    "system_name": "Microsoft Corporation 7.0 (Virtual Machine)",
    "system_type": "datacenter",
    "target_latency (ns)": 30000000,
    "target_qps": 275000,
    "task": "recommendation",
    "task2": "recommendation",
    "total_cores": 240,
    "uid": "9fbb2d48bfabecf5",
    "use_accelerator": true,
    "weight_data_types": "int8,fp16",
    "weight_transformations": "quantization, affine fusion"
  }
]