[ { "50.00 percentile latency (ns)": 1780516, "90.00 percentile latency (ns)": 1899002, "90th percentile latency (ns)": 1899002, "95.00 percentile latency (ns)": 1978062, "97.00 percentile latency (ns)": 2059362, "99.00 percentile latency (ns)": 2462891, "99.90 percentile latency (ns)": 10176447, "Max latency (ns)": 33642940, "Mean latency (ns)": 1830375, "Min duration satisfied": "Yes", "Min latency (ns)": 1209962, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "QPS w/ loadgen overhead": 522.32, "QPS w/o loadgen overhead": 546.34, "Result is": "VALID", "SUT name": "QAIC_SUT", "Scenario": "singlestream", "accelerator_frequency": "", "accelerator_host_interconnect": "PCIe 4 lane Gen4", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "8 GB", "accelerator_memory_configuration": "LPDDR4x", "accelerator_model_name": "QUALCOMM Cloud AI 100 DM.2e", "accelerator_on-chip_memories": "72 MB", "accelerators_per_node": 1, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.90th_percentile_latency_ms": 1.899002, "characteristics.90th_percentile_latency_ns": 1899002.0, "characteristics.90th_percentile_latency_s": 0.001899002, "characteristics.90th_percentile_latency_us": 1899.002, "characteristics.mAP": 23.163, "characteristics.power": 0.03511948522373089, "characteristics.power.normalized_per_core": 0.03511948522373089, "characteristics.power.normalized_per_processor": 0.03511948522373089, "ck_system": "aedk_15w-qaic-v1.4.66-aic100", "ck_used": true, "cooling": "passive", "dataset": "COCO 2017 (300x300)", "dataset_link": "https://github.com/ctuning/ck/blob/master/docs/mlperf-automation/datasets/coco2017.md", "dim_x_default": "characteristics.90th_percentile_latency_ms", "dim_y_default": "characteristics.mAP", "dim_y_maximize": true, "division": "closed", "formal_model": "ssd-mobilenet", "formal_model_accuracy": 99.0, "formal_model_link": "https://github.com/mlcommons/ck-mlops/tree/main/package", "framework": "Qualcomm Cloud AI SDK v1.4.66", "host_memory_capacity": "12 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "1x (L1d: 64K; L1i: 64K; L2: 512K); 3x (L1d: 64K; L1i: 64K; L2: 256K); 4x (L1d: 64K; L1i: 64K; L2: 128K); L3: 4M", "host_processor_core_count": 8, "host_processor_frequency": "1x (Cortex-A77 @ 2840 MHz); 3x (Cortex-A77 @ 2420 MHz); 4x (Cortex-A55 @ 1800 MHz)", "host_processor_interconnect": "", "host_processor_model_name": "Qualcomm Snapdragon 865", "host_processors_per_node": 1, "host_storage_capacity": "80 GB", "host_storage_type": "Internal flash", "hw_notes": "With 15W Accelerator TDP constraints", "informal_model": "ssd-mobilenet", "input_data_types": "int8", "key.accuracy": "characteristics.mAP", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1024, "mlperf_version": 1.1, "normalize_cores": 1, "normalize_processors": 1, "note_code": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Qualcomm/code", "note_details": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Qualcomm/results/aedk_15w-qaic-v1.4.66-aic100", "number_of_nodes": 1, "operating_system": "CentOS Linux release 8.0.1905 (kernel: 4.19.81 #1 SMP PREEMPT Tue May 11 08:23:20 UTC 2021 aarch64 aarch64 aarch64 GNU/Linux)", "other_software_stack": "Collective Knowledge v1.55.5", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 256, "print_timestamps": 0, "problem": false, "qsl_rng_seed": 1624344308455410291, "retraining": "No", "sample_index_rng_seed": 517984244576520566, "samples_per_query": 1, "schedule_rng_seed": 10051496985653635065, "starting_weights_filename": "https://zenodo.org/record/3228411/files/resnet34-ssd1200.onnx", "status": "available", "submitter": "Qualcomm", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Qualcomm", "sw_notes": "Powered by Collective Knowledge v1.55.5", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/aedk_15w-qaic-v1.4.66-aic100", "system_name": "AI Development Kit", "system_type": "edge", "target_latency (ns)": 0, "target_qps": 500, "task": "object detection", "task2": "object detection", "total_cores": 8, "uid": "b40dd4cfca4fa44b", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "see calibration.md" }, { "50.00 percentile latency (ns)": 25037328, "90.00 percentile latency (ns)": 27194868, "90th percentile latency (ns)": 27194868, "95.00 percentile latency (ns)": 27804798, "97.00 percentile latency (ns)": 28140152, "99.00 percentile latency (ns)": 28781800, "99.90 percentile latency (ns)": 31925925, "Max latency (ns)": 41601002, "Mean latency (ns)": 24847537, "Min duration satisfied": "Yes", "Min latency (ns)": 17970246, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "QPS w/ loadgen overhead": 40.01, "QPS w/o loadgen overhead": 40.25, "Result is": "VALID", "SUT name": "QAIC_SUT", "Scenario": "singlestream", "accelerator_frequency": "", "accelerator_host_interconnect": "PCIe 4 lane Gen4", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "8 GB", "accelerator_memory_configuration": "LPDDR4x", "accelerator_model_name": "QUALCOMM Cloud AI 100 DM.2e", "accelerator_on-chip_memories": "72 MB", "accelerators_per_node": 1, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.90th_percentile_latency_ms": 27.194868, "characteristics.90th_percentile_latency_ns": 27194868.0, "characteristics.90th_percentile_latency_s": 0.027194868, "characteristics.90th_percentile_latency_us": 27194.868, "characteristics.mAP": 19.827, "characteristics.power": 0.47527408033136215, "characteristics.power.normalized_per_core": 0.47527408033136215, "characteristics.power.normalized_per_processor": 0.47527408033136215, "ck_system": "aedk_15w-qaic-v1.4.66-aic100", "ck_used": true, "cooling": "passive", "dataset": "COCO 2017 (300x300)", "dataset_link": "https://github.com/ctuning/ck/blob/master/docs/mlperf-automation/datasets/coco2017.md", "dim_x_default": "characteristics.90th_percentile_latency_ms", "dim_y_default": "characteristics.mAP", "dim_y_maximize": true, "division": "closed", "formal_model": "ssd-mobilenet", "formal_model_accuracy": 99.0, "formal_model_link": "https://github.com/mlcommons/ck-mlops/tree/main/package", "framework": "Qualcomm Cloud AI SDK v1.4.66", "host_memory_capacity": "12 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "1x (L1d: 64K; L1i: 64K; L2: 512K); 3x (L1d: 64K; L1i: 64K; L2: 256K); 4x (L1d: 64K; L1i: 64K; L2: 128K); L3: 4M", "host_processor_core_count": 8, "host_processor_frequency": "1x (Cortex-A77 @ 2840 MHz); 3x (Cortex-A77 @ 2420 MHz); 4x (Cortex-A55 @ 1800 MHz)", "host_processor_interconnect": "", "host_processor_model_name": "Qualcomm Snapdragon 865", "host_processors_per_node": 1, "host_storage_capacity": "80 GB", "host_storage_type": "Internal flash", "hw_notes": "With 15W Accelerator TDP constraints", "informal_model": "ssd-resnet34", "input_data_types": "int8", "key.accuracy": "characteristics.mAP", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1024, "mlperf_version": 1.1, "normalize_cores": 1, "normalize_processors": 1, "note_code": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Qualcomm/code", "note_details": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Qualcomm/results/aedk_15w-qaic-v1.4.66-aic100", "number_of_nodes": 1, "operating_system": "CentOS Linux release 8.0.1905 (kernel: 4.19.81 #1 SMP PREEMPT Tue May 11 08:23:20 UTC 2021 aarch64 aarch64 aarch64 GNU/Linux)", "other_software_stack": "Collective Knowledge v1.55.5", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 64, "print_timestamps": 0, "problem": false, "qsl_rng_seed": 1624344308455410291, "retraining": "No", "sample_index_rng_seed": 517984244576520566, "samples_per_query": 1, "schedule_rng_seed": 10051496985653635065, "starting_weights_filename": "https://zenodo.org/record/3228411/files/resnet34-ssd1200.onnx", "status": "available", "submitter": "Qualcomm", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Qualcomm", "sw_notes": "Powered by Collective Knowledge v1.55.5", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/aedk_15w-qaic-v1.4.66-aic100", "system_name": "AI Development Kit", "system_type": "edge", "target_latency (ns)": 0, "target_qps": 33.3333, "task": "object detection", "task2": "object detection", "total_cores": 8, "uid": "1df895ee4226bb4b", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "see calibration.md" }, { "50.00 percentile latency (ns)": 3826870, "90.00 percentile latency (ns)": 3902979, "90th percentile latency (ns)": 3902979, "95.00 percentile latency (ns)": 3922230, "97.00 percentile latency (ns)": 3934670, "99.00 percentile latency (ns)": 3962520, "99.90 percentile latency (ns)": 4107271, "Max latency (ns)": 6740408, "Mean latency (ns)": 3822829, "Min duration satisfied": "Yes", "Min latency (ns)": 3552029, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "QPS w/ loadgen overhead": 259.62, "QPS w/o loadgen overhead": 261.59, "Result is": "VALID", "SUT name": "QAIC_SUT", "Scenario": "singlestream", "accelerator_frequency": "", "accelerator_host_interconnect": "1x PCIe 8 lane Gen4", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "16 GB", "accelerator_memory_configuration": "LPDDR4x", "accelerator_model_name": "QUALCOMM Cloud AI 100 PCIe HHHL", "accelerator_on-chip_memories": "144 MB", "accelerators_per_node": 1, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.90th_percentile_latency_ms": 3.902979, "characteristics.90th_percentile_latency_ns": 3902979.0, "characteristics.90th_percentile_latency_s": 0.003902979, "characteristics.90th_percentile_latency_us": 3902.979, "characteristics.mAP": 23.163, "ck_system": "r282_z93_q1-qaic-v1.5.6-aic100", "ck_used": true, "cooling": "air", "dataset": "COCO 2017 (300x300)", "dataset_link": "https://github.com/ctuning/ck/blob/master/docs/mlperf-automation/datasets/coco2017.md", "dim_x_default": "characteristics.90th_percentile_latency_ms", "dim_y_default": "characteristics.mAP", "dim_y_maximize": true, "division": "closed", "formal_model": "ssd-mobilenet", "formal_model_accuracy": 99.0, "formal_model_link": "https://github.com/mlcommons/ck-mlops/tree/main/package", "framework": "Qualcomm Cloud AI SDK v1.5.6", "host_memory_capacity": "8x 64 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "L1d: 32K; L1i: 32K; L2: 512K; L3: 16384K", "host_processor_core_count": 16, "host_processor_frequency": "2800 MHz", "host_processor_interconnect": "1x 16-20GT/s xGMI2", "host_processor_model_name": "AMD EPYC 7282 16-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "1 TB", "host_storage_type": "1x SSD", "hw_notes": "With 75W Accelerator TDP constraints", "informal_model": "ssd-mobilenet", "input_data_types": "int8", "key.accuracy": "characteristics.mAP", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1024, "mlperf_version": 1.1, "normalize_cores": 1, "normalize_processors": 1, "note_code": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Qualcomm/code", "note_details": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Qualcomm/results/r282_z93_q1-qaic-v1.5.6-aic100", "number_of_nodes": 1, "operating_system": "CentOS 7 (Linux kernel: 5.4.1-1.el7.elrepo.x86_64 #1 SMP Fri Nov 29 10:21:13 EST 2019 x86_64 x86_64 x86_64 GNU/Linux)", "other_software_stack": "Docker 20.10.7; Collective Knowledge v1.55.5", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 256, "print_timestamps": 0, "problem": false, "qsl_rng_seed": 1624344308455410291, "retraining": "No", "sample_index_rng_seed": 517984244576520566, "samples_per_query": 1, "schedule_rng_seed": 10051496985653635065, "starting_weights_filename": "https://zenodo.org/record/3228411/files/resnet34-ssd1200.onnx", "status": "available", "submitter": "Qualcomm", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Qualcomm", "sw_notes": "Powered by Collective Knowledge v1.55.5", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/r282_z93_q1-qaic-v1.5.6-aic100", "system_name": "Gigabyte R282-Z93 1x QAIC100", "system_type": "edge", "target_latency (ns)": 0, "target_qps": 250, "task": "object detection", "task2": "object detection", "total_cores": 32, "uid": "79464ea1ae6e1ed0", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "see calibration.md" }, { "50.00 percentile latency (ns)": 38404536, "90.00 percentile latency (ns)": 39592504, "90th percentile latency (ns)": 39592504, "95.00 percentile latency (ns)": 40013135, "97.00 percentile latency (ns)": 40331389, "99.00 percentile latency (ns)": 41221456, "99.90 percentile latency (ns)": 42592366, "Max latency (ns)": 44318259, "Mean latency (ns)": 38530427, "Min duration satisfied": "Yes", "Min latency (ns)": 36610711, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "QPS w/ loadgen overhead": 25.92, "QPS w/o loadgen overhead": 25.95, "Result is": "VALID", "SUT name": "QAIC_SUT", "Scenario": "singlestream", "accelerator_frequency": "", "accelerator_host_interconnect": "1x PCIe 8 lane Gen4", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "16 GB", "accelerator_memory_configuration": "LPDDR4x", "accelerator_model_name": "QUALCOMM Cloud AI 100 PCIe HHHL", "accelerator_on-chip_memories": "144 MB", "accelerators_per_node": 1, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.90th_percentile_latency_ms": 39.592504, "characteristics.90th_percentile_latency_ns": 39592504.0, "characteristics.90th_percentile_latency_s": 0.039592504, "characteristics.90th_percentile_latency_us": 39592.504, "characteristics.mAP": 19.827, "ck_system": "r282_z93_q1-qaic-v1.5.6-aic100", "ck_used": true, "cooling": "air", "dataset": "COCO 2017 (300x300)", "dataset_link": "https://github.com/ctuning/ck/blob/master/docs/mlperf-automation/datasets/coco2017.md", "dim_x_default": "characteristics.90th_percentile_latency_ms", "dim_y_default": "characteristics.mAP", "dim_y_maximize": true, "division": "closed", "formal_model": "ssd-mobilenet", "formal_model_accuracy": 99.0, "formal_model_link": "https://github.com/mlcommons/ck-mlops/tree/main/package", "framework": "Qualcomm Cloud AI SDK v1.5.6", "host_memory_capacity": "8x 64 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "L1d: 32K; L1i: 32K; L2: 512K; L3: 16384K", "host_processor_core_count": 16, "host_processor_frequency": "2800 MHz", "host_processor_interconnect": "1x 16-20GT/s xGMI2", "host_processor_model_name": "AMD EPYC 7282 16-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "1 TB", "host_storage_type": "1x SSD", "hw_notes": "With 75W Accelerator TDP constraints", "informal_model": "ssd-resnet34", "input_data_types": "int8", "key.accuracy": "characteristics.mAP", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1024, "mlperf_version": 1.1, "normalize_cores": 1, "normalize_processors": 1, "note_code": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Qualcomm/code", "note_details": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Qualcomm/results/r282_z93_q1-qaic-v1.5.6-aic100", "number_of_nodes": 1, "operating_system": "CentOS 7 (Linux kernel: 5.4.1-1.el7.elrepo.x86_64 #1 SMP Fri Nov 29 10:21:13 EST 2019 x86_64 x86_64 x86_64 GNU/Linux)", "other_software_stack": "Docker 20.10.7; Collective Knowledge v1.55.5", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 1024, "print_timestamps": 0, "problem": false, "qsl_rng_seed": 1624344308455410291, "retraining": "No", "sample_index_rng_seed": 517984244576520566, "samples_per_query": 1, "schedule_rng_seed": 10051496985653635065, "starting_weights_filename": "https://zenodo.org/record/3228411/files/resnet34-ssd1200.onnx", "status": "available", "submitter": "Qualcomm", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Qualcomm", "sw_notes": "Powered by Collective Knowledge v1.55.5", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/r282_z93_q1-qaic-v1.5.6-aic100", "system_name": "Gigabyte R282-Z93 1x QAIC100", "system_type": "edge", "target_latency (ns)": 0, "target_qps": 24.3902, "task": "object detection", "task2": "object detection", "total_cores": 32, "uid": "1dabce9f34d48168", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "see calibration.md" }, { "50.00 percentile latency (ns)": 3829326, "90.00 percentile latency (ns)": 3900176, "90th percentile latency (ns)": 3900176, "95.00 percentile latency (ns)": 3917987, "97.00 percentile latency (ns)": 3930616, "99.00 percentile latency (ns)": 3951807, "99.90 percentile latency (ns)": 4014027, "Max latency (ns)": 5252932, "Mean latency (ns)": 3828591, "Min duration satisfied": "Yes", "Min latency (ns)": 3626755, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "QPS w/ loadgen overhead": 259.68, "QPS w/o loadgen overhead": 261.19, "Result is": "VALID", "SUT name": "QAIC_SUT", "Scenario": "singlestream", "accelerator_frequency": "", "accelerator_host_interconnect": "5x PCIe 8 lane Gen4", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "16 GB", "accelerator_memory_configuration": "LPDDR4x", "accelerator_model_name": "QUALCOMM Cloud AI 100 PCIe HHHL", "accelerator_on-chip_memories": "144 MB", "accelerators_per_node": 5, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.90th_percentile_latency_ms": 3.900176, "characteristics.90th_percentile_latency_ns": 3900176.0, "characteristics.90th_percentile_latency_s": 0.003900176, "characteristics.90th_percentile_latency_us": 3900.176, "characteristics.mAP": 23.163, "characteristics.power": 0.8705427237206242, "characteristics.power.normalized_per_core": 0.17410854474412485, "characteristics.power.normalized_per_processor": 0.17410854474412485, "ck_system": "r282_z93_q5-qaic-v1.5.6-aic100", "ck_used": true, "cooling": "air", "dataset": "COCO 2017 (300x300)", "dataset_link": "https://github.com/ctuning/ck/blob/master/docs/mlperf-automation/datasets/coco2017.md", "dim_x_default": "characteristics.90th_percentile_latency_ms", "dim_y_default": "characteristics.mAP", "dim_y_maximize": true, "division": "closed", "formal_model": "ssd-mobilenet", "formal_model_accuracy": 99.0, "formal_model_link": "https://github.com/mlcommons/ck-mlops/tree/main/package", "framework": "Qualcomm Cloud AI SDK v1.5.6", "host_memory_capacity": "8x 64 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "L1d: 32K; L1i: 32K; L2: 512K; L3: 16384K", "host_processor_core_count": 16, "host_processor_frequency": "2800 MHz", "host_processor_interconnect": "1x 16-20GT/s xGMI2", "host_processor_model_name": "AMD EPYC 7282 16-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "1 TB", "host_storage_type": "1x SSD", "hw_notes": "With 75W Accelerator TDP constraints", "informal_model": "ssd-mobilenet", "input_data_types": "int8", "key.accuracy": "characteristics.mAP", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1024, "mlperf_version": 1.1, "normalize_cores": 5, "normalize_processors": 5, "note_code": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Qualcomm/code", "note_details": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Qualcomm/results/r282_z93_q5-qaic-v1.5.6-aic100", "number_of_nodes": 1, "operating_system": "CentOS 7 (Linux kernel: 5.4.1-1.el7.elrepo.x86_64 #1 SMP Fri Nov 29 10:21:13 EST 2019 x86_64 x86_64 x86_64 GNU/Linux)", "other_software_stack": "Docker 20.10.7; Collective Knowledge v1.55.5", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 256, "print_timestamps": 0, "problem": false, "qsl_rng_seed": 1624344308455410291, "retraining": "No", "sample_index_rng_seed": 517984244576520566, "samples_per_query": 1, "schedule_rng_seed": 10051496985653635065, "starting_weights_filename": "https://zenodo.org/record/3228411/files/resnet34-ssd1200.onnx", "status": "available", "submitter": "Qualcomm", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Qualcomm", "sw_notes": "Powered by Collective Knowledge v1.55.5", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/r282_z93_q5-qaic-v1.5.6-aic100", "system_name": "Gigabyte R282-Z93 5x QAIC100", "system_type": "edge", "target_latency (ns)": 0, "target_qps": 250, "task": "object detection", "task2": "object detection", "total_cores": 32, "uid": "5fa8be5444b44c84", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "see calibration.md" }, { "50.00 percentile latency (ns)": 38466485, "90.00 percentile latency (ns)": 39660174, "90th percentile latency (ns)": 39660174, "95.00 percentile latency (ns)": 40090991, "97.00 percentile latency (ns)": 40397814, "99.00 percentile latency (ns)": 41234604, "99.90 percentile latency (ns)": 42665236, "Max latency (ns)": 44152082, "Mean latency (ns)": 38586546, "Min duration satisfied": "Yes", "Min latency (ns)": 36616155, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "QPS w/ loadgen overhead": 25.87, "QPS w/o loadgen overhead": 25.92, "Result is": "VALID", "SUT name": "QAIC_SUT", "Scenario": "singlestream", "accelerator_frequency": "", "accelerator_host_interconnect": "5x PCIe 8 lane Gen4", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "16 GB", "accelerator_memory_configuration": "LPDDR4x", "accelerator_model_name": "QUALCOMM Cloud AI 100 PCIe HHHL", "accelerator_on-chip_memories": "144 MB", "accelerators_per_node": 5, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.90th_percentile_latency_ms": 39.660174, "characteristics.90th_percentile_latency_ns": 39660174.0, "characteristics.90th_percentile_latency_s": 0.039660174, "characteristics.90th_percentile_latency_us": 39660.174, "characteristics.mAP": 19.827, "characteristics.power": 9.560774210340496, "characteristics.power.normalized_per_core": 1.9121548420680992, "characteristics.power.normalized_per_processor": 1.9121548420680992, "ck_system": "r282_z93_q5-qaic-v1.5.6-aic100", "ck_used": true, "cooling": "air", "dataset": "COCO 2017 (300x300)", "dataset_link": "https://github.com/ctuning/ck/blob/master/docs/mlperf-automation/datasets/coco2017.md", "dim_x_default": "characteristics.90th_percentile_latency_ms", "dim_y_default": "characteristics.mAP", "dim_y_maximize": true, "division": "closed", "formal_model": "ssd-mobilenet", "formal_model_accuracy": 99.0, "formal_model_link": "https://github.com/mlcommons/ck-mlops/tree/main/package", "framework": "Qualcomm Cloud AI SDK v1.5.6", "host_memory_capacity": "8x 64 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "L1d: 32K; L1i: 32K; L2: 512K; L3: 16384K", "host_processor_core_count": 16, "host_processor_frequency": "2800 MHz", "host_processor_interconnect": "1x 16-20GT/s xGMI2", "host_processor_model_name": "AMD EPYC 7282 16-Core Processor", "host_processors_per_node": 2, "host_storage_capacity": "1 TB", "host_storage_type": "1x SSD", "hw_notes": "With 75W Accelerator TDP constraints", "informal_model": "ssd-resnet34", "input_data_types": "int8", "key.accuracy": "characteristics.mAP", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1024, "mlperf_version": 1.1, "normalize_cores": 5, "normalize_processors": 5, "note_code": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Qualcomm/code", "note_details": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Qualcomm/results/r282_z93_q5-qaic-v1.5.6-aic100", "number_of_nodes": 1, "operating_system": "CentOS 7 (Linux kernel: 5.4.1-1.el7.elrepo.x86_64 #1 SMP Fri Nov 29 10:21:13 EST 2019 x86_64 x86_64 x86_64 GNU/Linux)", "other_software_stack": "Docker 20.10.7; Collective Knowledge v1.55.5", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 1024, "print_timestamps": 0, "problem": false, "qsl_rng_seed": 1624344308455410291, "retraining": "No", "sample_index_rng_seed": 517984244576520566, "samples_per_query": 1, "schedule_rng_seed": 10051496985653635065, "starting_weights_filename": "https://zenodo.org/record/3228411/files/resnet34-ssd1200.onnx", "status": "available", "submitter": "Qualcomm", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Qualcomm", "sw_notes": "Powered by Collective Knowledge v1.55.5", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/r282_z93_q5-qaic-v1.5.6-aic100", "system_name": "Gigabyte R282-Z93 5x QAIC100", "system_type": "edge", "target_latency (ns)": 0, "target_qps": 24.3902, "task": "object detection", "task2": "object detection", "total_cores": 32, "uid": "f45a9a74e0da4205", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "see calibration.md" }, { "50.00 percentile latency (ns)": 1911197, "90.00 percentile latency (ns)": 2004945, "90th percentile latency (ns)": 2004945, "95.00 percentile latency (ns)": 2056454, "97.00 percentile latency (ns)": 2095151, "99.00 percentile latency (ns)": 2219002, "99.90 percentile latency (ns)": 3931090, "Max latency (ns)": 23094990, "Mean latency (ns)": 1929740, "Min duration satisfied": "Yes", "Min latency (ns)": 1446992, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "QPS w/ loadgen overhead": 497.42, "QPS w/o loadgen overhead": 518.2, "Result is": "VALID", "SUT name": "QAIC_SUT", "Scenario": "singlestream", "accelerator_frequency": "", "accelerator_host_interconnect": "PCIe 4 lane Gen4", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "16 GB", "accelerator_memory_configuration": "LPDDR4x", "accelerator_model_name": "QUALCOMM Cloud AI 100 DM.2", "accelerator_on-chip_memories": "144 MB", "accelerators_per_node": 1, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.90th_percentile_latency_ms": 2.004945, "characteristics.90th_percentile_latency_ns": 2004945.0, "characteristics.90th_percentile_latency_s": 0.002004945, "characteristics.90th_percentile_latency_us": 2004.945, "characteristics.mAP": 23.163, "characteristics.power": 0.03787876076898744, "characteristics.power.normalized_per_core": 0.03787876076898744, "characteristics.power.normalized_per_processor": 0.03787876076898744, "ck_system": "aedk_20w-qaic-v1.4.66-aic100", "ck_used": true, "cooling": "passive", "dataset": "COCO 2017 (300x300)", "dataset_link": "https://github.com/ctuning/ck/blob/master/docs/mlperf-automation/datasets/coco2017.md", "dim_x_default": "characteristics.90th_percentile_latency_ms", "dim_y_default": "characteristics.mAP", "dim_y_maximize": true, "division": "closed", "formal_model": "ssd-mobilenet", "formal_model_accuracy": 99.0, "formal_model_link": "https://github.com/mlcommons/ck-mlops/tree/main/package", "framework": "Qualcomm Cloud AI SDK v1.4.66", "host_memory_capacity": "12 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "1x (L1d: 64K; L1i: 64K; L2: 512K); 3x (L1d: 64K; L1i: 64K; L2: 256K); 4x (L1d: 64K; L1i: 64K; L2: 128K); L3: 4M", "host_processor_core_count": 8, "host_processor_frequency": "1x (Cortex-A77 @ 2840 MHz); 3x (Cortex-A77 @ 2420 MHz); 4x (Cortex-A55 @ 1800 MHz)", "host_processor_interconnect": "", "host_processor_model_name": "Qualcomm Snapdragon 865", "host_processors_per_node": 1, "host_storage_capacity": "80 GB", "host_storage_type": "Internal flash", "hw_notes": "With 20W Accelerator TDP constraints", "informal_model": "ssd-mobilenet", "input_data_types": "int8", "key.accuracy": "characteristics.mAP", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1024, "mlperf_version": 1.1, "normalize_cores": 1, "normalize_processors": 1, "note_code": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Qualcomm/code", "note_details": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Qualcomm/results/aedk_20w-qaic-v1.4.66-aic100", "number_of_nodes": 1, "operating_system": "CentOS Linux release 8.0.1905 (kernel: 4.19.81 #1 SMP PREEMPT Tue May 11 08:23:20 UTC 2021 aarch64 aarch64 aarch64 GNU/Linux)", "other_software_stack": "Collective Knowledge v1.55.5", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 256, "print_timestamps": 0, "problem": false, "qsl_rng_seed": 1624344308455410291, "retraining": "No", "sample_index_rng_seed": 517984244576520566, "samples_per_query": 1, "schedule_rng_seed": 10051496985653635065, "starting_weights_filename": "https://zenodo.org/record/3228411/files/resnet34-ssd1200.onnx", "status": "available", "submitter": "Qualcomm", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Qualcomm", "sw_notes": "Powered by Collective Knowledge v1.55.5", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/aedk_20w-qaic-v1.4.66-aic100", "system_name": "AI Development Kit", "system_type": "edge", "target_latency (ns)": 0, "target_qps": 500, "task": "object detection", "task2": "object detection", "total_cores": 8, "uid": "228126d1789439e9", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "see calibration.md" }, { "50.00 percentile latency (ns)": 25394201, "90.00 percentile latency (ns)": 27519143, "90th percentile latency (ns)": 27519143, "95.00 percentile latency (ns)": 28137772, "97.00 percentile latency (ns)": 28454066, "99.00 percentile latency (ns)": 28946864, "99.90 percentile latency (ns)": 31489815, "Max latency (ns)": 59221974, "Mean latency (ns)": 25155968, "Min duration satisfied": "Yes", "Min latency (ns)": 18111327, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "QPS w/ loadgen overhead": 39.5, "QPS w/o loadgen overhead": 39.75, "Result is": "VALID", "SUT name": "QAIC_SUT", "Scenario": "singlestream", "accelerator_frequency": "", "accelerator_host_interconnect": "PCIe 4 lane Gen4", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "16 GB", "accelerator_memory_configuration": "LPDDR4x", "accelerator_model_name": "QUALCOMM Cloud AI 100 DM.2", "accelerator_on-chip_memories": "144 MB", "accelerators_per_node": 1, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.90th_percentile_latency_ms": 27.519143, "characteristics.90th_percentile_latency_ns": 27519143.0, "characteristics.90th_percentile_latency_s": 0.027519143, "characteristics.90th_percentile_latency_us": 27519.143, "characteristics.mAP": 19.827, "characteristics.power": 0.49474180982904864, "characteristics.power.normalized_per_core": 0.49474180982904864, "characteristics.power.normalized_per_processor": 0.49474180982904864, "ck_system": "aedk_20w-qaic-v1.4.66-aic100", "ck_used": true, "cooling": "passive", "dataset": "COCO 2017 (300x300)", "dataset_link": "https://github.com/ctuning/ck/blob/master/docs/mlperf-automation/datasets/coco2017.md", "dim_x_default": "characteristics.90th_percentile_latency_ms", "dim_y_default": "characteristics.mAP", "dim_y_maximize": true, "division": "closed", "formal_model": "ssd-mobilenet", "formal_model_accuracy": 99.0, "formal_model_link": "https://github.com/mlcommons/ck-mlops/tree/main/package", "framework": "Qualcomm Cloud AI SDK v1.4.66", "host_memory_capacity": "12 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "1x (L1d: 64K; L1i: 64K; L2: 512K); 3x (L1d: 64K; L1i: 64K; L2: 256K); 4x (L1d: 64K; L1i: 64K; L2: 128K); L3: 4M", "host_processor_core_count": 8, "host_processor_frequency": "1x (Cortex-A77 @ 2840 MHz); 3x (Cortex-A77 @ 2420 MHz); 4x (Cortex-A55 @ 1800 MHz)", "host_processor_interconnect": "", "host_processor_model_name": "Qualcomm Snapdragon 865", "host_processors_per_node": 1, "host_storage_capacity": "80 GB", "host_storage_type": "Internal flash", "hw_notes": "With 20W Accelerator TDP constraints", "informal_model": "ssd-resnet34", "input_data_types": "int8", "key.accuracy": "characteristics.mAP", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1024, "mlperf_version": 1.1, "normalize_cores": 1, "normalize_processors": 1, "note_code": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Qualcomm/code", "note_details": "https://github.com/mlcommons/inference_results_v1.1/tree/master/closed/Qualcomm/results/aedk_20w-qaic-v1.4.66-aic100", "number_of_nodes": 1, "operating_system": "CentOS Linux release 8.0.1905 (kernel: 4.19.81 #1 SMP PREEMPT Tue May 11 08:23:20 UTC 2021 aarch64 aarch64 aarch64 GNU/Linux)", "other_software_stack": "Collective Knowledge v1.55.5", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 64, "print_timestamps": 0, "problem": false, "qsl_rng_seed": 1624344308455410291, "retraining": "No", "sample_index_rng_seed": 517984244576520566, "samples_per_query": 1, "schedule_rng_seed": 10051496985653635065, "starting_weights_filename": "https://zenodo.org/record/3228411/files/resnet34-ssd1200.onnx", "status": "available", "submitter": "Qualcomm", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Qualcomm", "sw_notes": "Powered by Collective Knowledge v1.55.5", "system_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.system/aedk_20w-qaic-v1.4.66-aic100", "system_name": "AI Development Kit", "system_type": "edge", "target_latency (ns)": 0, "target_qps": 33.3333, "task": "object detection", "task2": "object detection", "total_cores": 8, "uid": "d3fd2330166c4681", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "see calibration.md" }, { "50.00 percentile latency (ns)": 28192123, "90.00 percentile latency (ns)": 30435498, "90th percentile latency (ns)": 30435498, "95.00 percentile latency (ns)": 31079646, "97.00 percentile latency (ns)": 31449376, "99.00 percentile latency (ns)": 32030505, "99.90 percentile latency (ns)": 34758502, "Max latency (ns)": 46670145, "Mean latency (ns)": 28254589, "Min duration satisfied": "Yes", "Min latency (ns)": 20417290, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "QPS w/ loadgen overhead": 35.15, "QPS w/o loadgen overhead": 35.39, "Result is": "VALID", "SUT name": "QAIC_SUT", "Scenario": "singlestream", "accelerator_frequency": "", "accelerator_host_interconnect": "PCIe 4 lane Gen4", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "16 GB", "accelerator_memory_configuration": "LPDDR4x", "accelerator_model_name": "QUALCOMM Cloud AI 100 DM.2", "accelerator_on-chip_memories": "144 MB", "accelerators_per_node": 1, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.90th_percentile_latency_ms": 30.435498, "characteristics.90th_percentile_latency_ns": 30435498.0, "characteristics.90th_percentile_latency_s": 0.030435498, "characteristics.90th_percentile_latency_us": 30435.498, "characteristics.mAP": 19.87, "characteristics.power": 0.5953544536768118, "characteristics.power.normalized_per_core": 0.5953544536768118, "characteristics.power.normalized_per_processor": 0.5953544536768118, "ck_system": "aedk_20w-qaic-v1.3.14-aic100", "ck_used": true, "cooling": "passive", "dataset": "COCO 2017 (300x300)", "dataset_link": "https://github.com/ctuning/ck/blob/master/docs/mlperf-automation/datasets/coco2017.md", "dim_x_default": "characteristics.90th_percentile_latency_ms", "dim_y_default": "characteristics.mAP", "dim_y_maximize": true, "division": "closed", "formal_model": "ssd-mobilenet", "formal_model_accuracy": 99.0, "formal_model_link": "https://github.com/mlcommons/ck-mlops/tree/main/package", "framework": "Qualcomm Cloud AI SDK v1.3.14", "host_memory_capacity": "12 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "1x (L1d: 64K; L1i: 64K; L2: 512K); 3x (L1d: 64K; L1i: 64K; L2: 256K); 4x (L1d: 64K; L1i: 64K; L2: 128K); L3: 4M", "host_processor_core_count": 120, "host_processor_frequency": "1x (Cortex-A77 @ 2840 MHz); 3x (Cortex-A77 @ 2420 MHz); 4x (Cortex-A55 @ 1800 MHz)", "host_processor_interconnect": "", "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 1, "host_storage_capacity": "894 GB", "host_storage_type": "NVMe SSD", "hw_notes": "With 20W Accelerator TDP constraints", "informal_model": "ssd-resnet34", "input_data_types": "int8", "key.accuracy": "characteristics.mAP", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1024, "mlperf_version": 1.0, "normalize_cores": 1, "normalize_processors": 1, "note_code": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/Qualcomm/code", "note_details": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/Qualcomm/results/aedk_20w-qaic-v1.3.14-aic100", "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": "GCC 7.5.0; Python 3.7.10", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 64, "print_timestamps": 0, "problem": false, "qsl_rng_seed": 7322528924094909334, "retraining": "No", "sample_index_rng_seed": 1570999273408051088, "samples_per_query": 1, "schedule_rng_seed": 3507442325620259414, "starting_weights_filename": "https://zenodo.org/record/3228411/files/resnet34-ssd1200.onnx", "status": "available", "submitter": "Qualcomm", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Qualcomm", "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/aedk_20w-qaic-v1.3.14-aic100", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "edge", "target_latency (ns)": 0, "target_qps": 33.3333, "task": "object detection", "task2": "object detection", "total_cores": 120, "uid": "dbe5b2542aa9b4b6", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "see calibration/" }, { "50.00 percentile latency (ns)": 29482357, "90.00 percentile latency (ns)": 31703549, "90th percentile latency (ns)": 31703549, "95.00 percentile latency (ns)": 32236973, "97.00 percentile latency (ns)": 32603109, "99.00 percentile latency (ns)": 33188875, "99.90 percentile latency (ns)": 34913621, "Max latency (ns)": 59885836, "Mean latency (ns)": 29503958, "Min duration satisfied": "Yes", "Min latency (ns)": 21273352, "Min queries satisfied": "Yes", "Mode": "PerformanceOnly", "QPS w/ loadgen overhead": 33.62, "QPS w/o loadgen overhead": 33.89, "Result is": "VALID", "SUT name": "QAIC_SUT", "Scenario": "singlestream", "accelerator_frequency": "", "accelerator_host_interconnect": "PCIe 4 lane Gen4", "accelerator_interconnect": "", "accelerator_interconnect_topology": "", "accelerator_memory_capacity": "8 GB", "accelerator_memory_configuration": "LPDDR4x", "accelerator_model_name": "QUALCOMM Cloud AI 100 DM.2e", "accelerator_on-chip_memories": "72 MB", "accelerators_per_node": 1, "accuracy_log_probability": 0, "accuracy_log_rng_seed": 0, "accuracy_log_sampling_target": 0, "characteristics.90th_percentile_latency_ms": 31.703549, "characteristics.90th_percentile_latency_ns": 31703549.0, "characteristics.90th_percentile_latency_s": 0.031703549, "characteristics.90th_percentile_latency_us": 31703.549, "characteristics.mAP": 19.87, "characteristics.power": 0.6272178085003224, "characteristics.power.normalized_per_core": 0.6272178085003224, "characteristics.power.normalized_per_processor": 0.6272178085003224, "ck_system": "aedk_15w-qaic-v1.3.14-aic100", "ck_used": true, "cooling": "passive", "dataset": "COCO 2017 (300x300)", "dataset_link": "https://github.com/ctuning/ck/blob/master/docs/mlperf-automation/datasets/coco2017.md", "dim_x_default": "characteristics.90th_percentile_latency_ms", "dim_y_default": "characteristics.mAP", "dim_y_maximize": true, "division": "closed", "formal_model": "ssd-mobilenet", "formal_model_accuracy": 99.0, "formal_model_link": "https://github.com/mlcommons/ck-mlops/tree/main/package", "framework": "Qualcomm Cloud AI SDK v1.3.14", "host_memory_capacity": "12 GB", "host_memory_configuration": "", "host_networking": "", "host_networking_topology": "", "host_processor_caches": "1x (L1d: 64K; L1i: 64K; L2: 512K); 3x (L1d: 64K; L1i: 64K; L2: 256K); 4x (L1d: 64K; L1i: 64K; L2: 128K); L3: 4M", "host_processor_core_count": 120, "host_processor_frequency": "1x (Cortex-A77 @ 2840 MHz); 3x (Cortex-A77 @ 2420 MHz); 4x (Cortex-A55 @ 1800 MHz)", "host_processor_interconnect": "", "host_processor_model_name": "AMD EPYC 7V13 64-Core Processor", "host_processors_per_node": 1, "host_storage_capacity": "894 GB", "host_storage_type": "NVMe SSD", "hw_notes": "With 15W Accelerator TDP constraints", "informal_model": "ssd-resnet34", "input_data_types": "int8", "key.accuracy": "characteristics.mAP", "max_async_queries": 1, "max_duration (ms)": 0, "max_query_count": 0, "min_duration (ms)": 600000, "min_query_count": 1024, "mlperf_version": 1.0, "normalize_cores": 1, "normalize_processors": 1, "note_code": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/Qualcomm/code", "note_details": "https://github.com/mlcommons/inference_results_v1.0/tree/master/closed/Qualcomm/results/aedk_15w-qaic-v1.3.14-aic100", "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": "GCC 7.5.0; Python 3.7.10", "performance_issue_same": 0, "performance_issue_same_index": 0, "performance_issue_unique": 0, "performance_sample_count": 64, "print_timestamps": 0, "problem": false, "qsl_rng_seed": 7322528924094909334, "retraining": "No", "sample_index_rng_seed": 1570999273408051088, "samples_per_query": 1, "schedule_rng_seed": 3507442325620259414, "starting_weights_filename": "https://zenodo.org/record/3228411/files/resnet34-ssd1200.onnx", "status": "available", "submitter": "Qualcomm", "submitter_link": "https://github.com/ctuning/ck-mlperf-inference/tree/main/bench.mlperf.submitter/Qualcomm", "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/aedk_15w-qaic-v1.3.14-aic100", "system_name": "Microsoft Corporation 7.0 (Virtual Machine)", "system_type": "edge", "target_latency (ns)": 0, "target_qps": 33.3333, "task": "object detection", "task2": "object detection", "total_cores": 120, "uid": "542bb9007d03b1ca", "use_accelerator": true, "weight_data_types": "int8", "weight_transformations": "see calibration/" } ]