Edge Software Device Qualification

Test Report

The Intel® Edge Software Device Qualification is a Command Line Interface (CLI) tool that enables customers to test and qualify their platform for a specific Intel® Edge Software package.

Tool Suite Information

Package Name esdq
Package Version 11.1.0
Package CLI Version 11.1.0

1. metro-device-qualification

Test Module Information

Module Name Metro Device Qualification Test
Module Version 2.0

Test Case Execution Summary

Total Test Cases 8
Total Test Cases Passed 8
Total Test Cases Failed 0
Total Test Cases with Unknown Result 0
Pass Percentage (%) 100.00
Verdict PASSED

Test Module Components

Category Name Version Status
generic Docker CE 28.4.0 Installed
Docker Images metro2.0/media_performance_benchmark_runner 2.0 Installed
Docker Images metro2.0/ai_frequency_measure_runner 2.0 Installed
Docker Images metro2.0/proxy_pipeline_benchmark_runner 2.0 Installed
Docker Images metro2.0/openvino_benchmark_runner 2.0 Installed
Docker Images metro2.0/stream_memory_benchmark 2.0 Installed

Additional Data

System Info

Section Field Value
HARDWARE INFO Device Manufacturer O.E.M.
HARDWARE INFO Hardware Architecture x86_64
HARDWARE INFO Processor Intel(R) Core(TM) Ultra 7 155U
HARDWARE INFO GPU PCI ID 7d45 (Intel® Graphics)
HARDWARE INFO GPU Memory Size 16M (Intel® Graphics)
HARDWARE INFO GPU Memory Size 256M (Intel® Graphics)
HARDWARE INFO Memory Size 30Gi
HARDWARE INFO Monitor Monitors: 1
HARDWARE INFO Monitor 0: +*DP-1 1920/480x1080/270+0+0 DP-1
HDD Configurations Model TS1TSSD4 72K-I (scsi)
HDD Configurations Disk /dev/sda 1024GB
HDD Configurations Sector size (logical/physical) 512B/512B
HDD Configurations Partition Table gpt
HDD Configurations Disk Flags Number Start End Size File system Name Flags
HDD Configurations Disk Flags 1 1049kB 1128MB 1127MB fat32 boot, esp
HDD Configurations Disk Flags 2 1128MB 1024GB 1023GB ext4
BIOS Configuration

bios_config.txt

CPU Power Configuration

cpu_power_config.txt

Software Info

Section Field Value
SOFTWARE INFO OS Version Ubuntu 24.04.2 LTS
SOFTWARE INFO Kernel Version 6.14.0-29-generic
SOFTWARE INFO OpenVINO Version (Base) 2025.2.0
SOFTWARE INFO OpenVINO Version (for DLStreamer) 2025.2.0
SOFTWARE INFO OpenVINO Version (for OpenCV/FFMpeg) 2025.2.0
SOFTWARE INFO Docker CE Version 28.4.0
SOFTWARE INFO Docker Compose Version 2.39.2
SOFTWARE INFO Intel Level Zero for GPU NA
SOFTWARE INFO Intel Graphics Compiler for OpenCL 25.05.32567.19-1099
SOFTWARE INFO Libva / VAAPI Driver NA
SOFTWARE INFO Media-driver Version 25.2.4-1146~24.04
SOFTWARE INFO Intel OneVPL 1:2.15.0.0-1140
SOFTWARE INFO Mesa NA
SOFTWARE INFO OpenCV Version 4.11.0.86
SOFTWARE INFO DLStreamer Version 1.26.4
SOFTWARE INFO FFmpeg Version 6.1.1-3ubuntu5
SOFTWARE INFO NPU Version 1.19.0.20250707-16111289554
SOFTWARE INFO Current System Time 2025-10-29 01:45:41

Test Case Results

performance

Category Name Value Type Logic Type Expected Value Value Parameters Status
generic OpenVINO Benchmark Test text eq No Error No Error na PASSED
generic Memory Benchmark Test text eq No Error No Error na PASSED
generic Media Performance Test text eq No Error No Error na PASSED
generic GPU Freq Performance Test text eq No Error No Error na PASSED
generic Smart NVR Proxy Pipeline Benchmark Test text eq No Error No Error na PASSED
generic Headed Visual AI Proxy Pipeline Benchmark Test text eq No Error No Error na PASSED
generic VSaaS Gateway with Storage Proxy Pipeline Benchmark Test text eq No Error No Error na PASSED
generic License Plate Recognition(LPR) AI Proxy Pipeline Benchmark text eq No Error No Error na PASSED

Additional Data

resnet-50-tf OV Benchmark Result

Model Precision Device Throughput
(fps)		 Latency
(ms)		 Reference Platform Reference Throughput (fps) Reference GPU Freq (GHz) CPU Avg Freq
(GHz)		 CPU Util (norm)
(%)		 Memory Util
(%)		 GPU Freq
(GHz)		 GPU EU Util
(%)		 GPU VDBox Util
(%)		 GPU Power Util
(W)		 Package Power
(W)		 Result
resnet-50-tf INT8 GPU.0 427.85 597.22 MTL 165H (32G Mem) 675.8 0.9 1.38 9.02 2.52 0.69 86.01 0.0 6.85 14.94 No Error
resnet-50-tf INT8 NPU 652.50 6.06 MTL 165H (32G Mem) 717.62 1.02 1.77 1.48 1.04 14.71 No Error
resnet-50-tf OV Benchmark Result Chart

resnet-50-tf OV Benchmark Result Chart
efficientnet-b0 OV Benchmark Result

Model Precision Device Throughput
(fps)		 Latency
(ms)		 Reference Platform Reference Throughput (fps) Reference GPU Freq (GHz) CPU Avg Freq
(GHz)		 CPU Util (norm)
(%)		 Memory Util
(%)		 GPU Freq
(GHz)		 GPU EU Util
(%)		 GPU VDBox Util
(%)		 GPU Power Util
(W)		 Package Power
(W)		 Result
efficientnet-b0 INT8 GPU.0 727.40 175.77 MTL 165H (32G Mem) 1253.92 0.17 1.56 11.64 3.05 0.71 85.26 0.0 4.72 14.97 No Error
efficientnet-b0 INT8 NPU 476.03 8.33 MTL 165H (32G Mem) 485.84 1.23 1.62 1.09 0.84 11.48 No Error
efficientnet-b0 OV Benchmark Result Chart

efficientnet-b0 OV Benchmark Result Chart
ssdlite_mobilenet_v2 OV Benchmark Result

Model Precision Device Throughput
(fps)		 Latency
(ms)		 Reference Platform Reference Throughput (fps) Reference GPU Freq (GHz) CPU Avg Freq
(GHz)		 CPU Util (norm)
(%)		 Memory Util
(%)		 GPU Freq
(GHz)		 GPU EU Util
(%)		 GPU VDBox Util
(%)		 GPU Power Util
(W)		 Package Power
(W)		 Result
ssdlite_mobilenet_v2 INT8 GPU.0 489.70 8.09 MTL 165H (32G Mem) 480.35 0.9 1.58 17.27 1.75 0.62 94.39 0.0 3.78 15.68 No Error
ssdlite_mobilenet_v2 INT8 NPU 796.70 4.96 MTL 165H (32G Mem) 817.27 0.96 1.69 1.43 0.82 13.0 No Error
ssdlite_mobilenet_v2 OV Benchmark Result Chart

ssdlite_mobilenet_v2 OV Benchmark Result Chart
mobilenet-v2-pytorch OV Benchmark Result

Model Precision Device Throughput
(fps)		 Latency
(ms)		 Reference Platform Reference Throughput (fps) Reference GPU Freq (GHz) CPU Avg Freq
(GHz)		 CPU Util (norm)
(%)		 Memory Util
(%)		 GPU Freq
(GHz)		 GPU EU Util
(%)		 GPU VDBox Util
(%)		 GPU Power Util
(W)		 Package Power
(W)		 Result
mobilenet-v2-pytorch INT8 GPU.0 1710.73 74.06 MTL 165H (32G Mem) 3102.01 0.9 1.52 13.54 2.27 0.59 86.32 0.0 4.3 15.19 No Error
mobilenet-v2-pytorch INT8 NPU 1965.83 1.98 MTL 165H (32G Mem) 2007.79 0.96 1.95 1.96 0.79 15.16 No Error
mobilenet-v2-pytorch OV Benchmark Result Chart

mobilenet-v2-pytorch OV Benchmark Result Chart
yolo-v5s OV Benchmark Result

Model Precision Device Throughput
(fps)		 Latency
(ms)		 Reference Platform Reference Throughput (fps) Reference GPU Freq (GHz) CPU Avg Freq
(GHz)		 CPU Util (norm)
(%)		 Memory Util
(%)		 GPU Freq
(GHz)		 GPU EU Util
(%)		 GPU VDBox Util
(%)		 GPU Power Util
(W)		 Package Power
(W)		 Result
yolo-v5s INT8 GPU.0 169.23 377.67 MTL 165H (32G Mem) 285.99 0.88 1.14 9.42 2.3 0.65 83.92 0.0 6.2 14.87 No Error
yolo-v5s INT8 NPU 114.19 34.94 MTL 165H (32G Mem) 111.83 0.79 1.34 0.41 0.87 10.77 No Error
yolo-v5s OV Benchmark Result Chart

yolo-v5s OV Benchmark Result Chart
yolo-v8s OV Benchmark Result

Model Precision Device Throughput
(fps)		 Latency
(ms)		 Reference Platform Reference Throughput (fps) Reference GPU Freq (GHz) CPU Avg Freq
(GHz)		 CPU Util (norm)
(%)		 Memory Util
(%)		 GPU Freq
(GHz)		 GPU EU Util
(%)		 GPU VDBox Util
(%)		 GPU Power Util
(W)		 Package Power
(W)		 Result
yolo-v8s INT8 GPU.0 71.34 55.99 MTL 165H (32G Mem) 116.1 0.66 1.61 15.45 1.72 0.63 95.04 0.0 4.87 15.79 No Error
yolo-v8s OV Benchmark Result Chart

yolo-v8s OV Benchmark Result Chart
clip-vit-base-patch16 OV Benchmark Result

Model Precision Device Throughput
(fps)		 Latency
(ms)		 Reference Platform Reference Throughput (fps) Reference GPU Freq (GHz) CPU Avg Freq
(GHz)		 CPU Util (norm)
(%)		 Memory Util
(%)		 GPU Freq
(GHz)		 GPU EU Util
(%)		 GPU VDBox Util
(%)		 GPU Power Util
(W)		 Package Power
(W)		 Result
clip-vit-base-patch16 INT8 GPU.0 9.64 414.46 MTL 165H (32G Mem) 0 0 1.35 11.9 2.44 0.77 83.8 0.0 5.93 15.11 No Error
clip-vit-base-patch16 OV Benchmark Result Chart

clip-vit-base-patch16 OV Benchmark Result Chart
Memory Benchmark Test Result

This test runs a memory benchmark based on STREAM to ensure sufficient memory bandwidth for media processing.

memory_benchmark_runner_test Function Best Rate MB/s Avg time Min time Max time Result
Memory_Benchmark_Test Copy 34313.9 0.056199 0.055954 0.056677 No Error
Memory_Benchmark_Test Scale 26711.1 0.074053 0.071880 0.075055 No Error
Memory_Benchmark_Test Add 29723.2 0.099898 0.096894 0.101036 No Error
Memory_Benchmark_Test Triad 29001.2 0.100625 0.099306 0.101643 No Error
Media Performance Test Result

Media Encode and Decode Performance tests.

Media Performance Benchmark Device Used Codec Bitrate Resolution Number of Monitors Max Channels Reference Platform Reference Value AVG CPU Freq(MHz) AVG CPU Util(%) AVG Memory Util(%) AVG GPU Freq(MHz) AVG GPU EU(RCS) Util(%) AVG GPU VDBox(VCS) Util(%) Avg SoC Power (W) AVG GPU power(W) Duration(s) Result
Media Encode Benchmark iGPU h264 4Mbps 1080p@30 N/A 22 MTL 165H (32G Mem) 25 2512.83 132.48 0.90 409.07 0.00 97.03 14.93 0.13 560.65 No Error
Media Encode Benchmark iGPU h264 16Mbps 4K@30 N/A 5 MTL 165H (32G Mem) 5 2286.70 23.19 0.30 582.22 0.00 99.65 14.99 0.14 372.84 No Error
Media Encode Benchmark iGPU h265 2Mbps 1080p@30 N/A 15 MTL 165H (32G Mem) 18 1967.28 66.91 2.20 415.10 0.00 0.00 15.07 0.14 424.28 No Error
Media Encode Benchmark iGPU h265 8Mbps 4K@30 N/A 3 MTL 165H (32G Mem) 3 2159.35 14.60 1.50 434.31 0.00 0.00 14.88 0.17 313.62 No Error
Media Performance Test Result

Media Encode and Decode Performance tests.

Media Performance Test Result
Media Performance Test Result

Media Encode and Decode Performance tests.

Media Performance Benchmark Device Used Input Codec Input Bitrate Input Resolution Number of Monitors Max Channels Reference Platform Reference Value AVG CPU Freq(MHz) AVG CPU Util(%) AVG Memory Util(%) AVG GPU Freq(MHz) AVG GPU EU(RCS) Util(%) AVG GPU VDBox(VCS) Util(%) Avg SoC Power (W) AVG GPU power(W) Duration(s) Result
Media Decode Benchmark iGPU h264 4Mbps 1080p@30 N/A 56 MTL 165H (32G Mem) 77 2521.49 137.45 0.69 474.15 0.00 0.00 14.77 0.15 803.47 No Error
Media Decode Benchmark iGPU h264 16Mbps 4K@30 N/A 16 MTL 165H (32G Mem) 23 2401.92 18.83 0.30 590.84 0.00 0.00 13.45 0.16 568.25 No Error
Media Decode Benchmark iGPU h265 2Mbps 1080p@30 N/A 70 MTL 165H (32G Mem) 107 2022.62 238.51 0.78 429.67 0.00 0.00 14.96 0.16 853.43 No Error
Media Decode Benchmark iGPU h265 8Mbps 4K@30 N/A 25 MTL 165H (32G Mem) 28 2640.05 27.82 0.40 590.00 0.00 0.00 15.08 0.17 555.36 No Error
Media Decode + 6*6 Compose Benchmark to 1080p@30FPS iGPU h264 4Mbps 1080p@30 1 27 MTL 165H (32G Mem) 17 1383.83 361.84 0.70 645.22 11.13 0.00 15.00 0.44 1522.61 No Error
Media Decode + 6*6 Compose Benchmark to 4K@30FPS iGPU h264 16Mbps 4K@30 1 8 MTL 165H (32G Mem) 4 2852.17 30.48 0.39 677.56 5.02 0.00 14.63 0.33 631.27 No Error
Media Decode + 6*6 Compose Benchmark to 1080p@30FPS iGPU h265 2Mbps 1080p@30 1 24 MTL 165H (32G Mem) 17 2594.19 162.72 0.69 559.60 8.91 0.00 15.02 0.38 1211.51 No Error
Media Decode + 6*6 Compose Benchmark to 4K@30FPS iGPU h265 8Mbps 4K@30 1 8 MTL 165H (32G Mem) 4 2930.83 30.85 0.30 682.62 5.24 0.00 14.97 0.33 637.24 No Error
Media Performance Test Result

Media Encode and Decode Performance tests.

Media Performance Test Result
Details of Media Performance Test

media_performance_benchmark.tar.gz

GPU Freq Performance Test Result

This test uses the OpenVINO benchmark app to exercise the GPU for an extended period and while tracking its frequency.

GPU Freq Performance Test Result
Smart NVR Proxy Pipeline Benchmark Test Introduction

Benchmark Specification: Input: 1080p@30 H.264 4Mbps (video with 10 objects); Storage: 1080p@30 H.264 4Mbps; Output: VideoWall, 4K@30 H264 16Mbps; Benchmark metric: Number of video channels with inference.

Smart NVR Proxy Pipeline Benchmark Test Introduction
Smart NVR Proxy Pipeline Benchmark Test Result

Benchmark Specification: Input: 1080p@30 H.264 4Mbps (video with 10 objects); Storage: 1080p@30 H.264 4Mbps; Output: VideoWall, 4K@30 H264 16Mbps; Benchmark metric: Number of video channels with inference.

Pipeline_Test Device Used Input Codec Input Resolution Input Channels Model Compose Number of Monitors AI Channels Reference Platform Reference Value AVG CPU Freq(MHz) AVG CPU Util(%) AVG Memory Util(%) AVG GPU Freq(MHz) AVG GPU EU(RCS) Util(%) AVG GPU VDBox(VCS) Util(%) AVG package power(W) AVG GPU power(W) Duration(s) Result
Smart NVR Pipeline iGPU H264 (4Mbps) 1080p@20 25 yolov5s-416 5x5 1 9 MTL 165H (32G Mem) 23 1929.46 333.28 2.67 645.81 6.91 45.19 14.67 1.83 912.19 No Error
Smart NVR Pipeline iGPU H264 (4Mbps) 1080p@20 25 yolov5m-416 5x5 1 4 MTL 165H (32G Mem) 12 1932.10 316.09 2.53 750.41 8.72 55.89 15.05 2.30 793.51 No Error
Smart NVR Pipeline iGPU H264 (4Mbps) 1080p@20 25 yolov5m-416+efficientnet-b0 5x5 1 1 MTL 165H (32G Mem) 3 1949.45 227.63 2.55 577.62 10.69 21.06 15.02 0.95 409.38 No Error
Smart NVR Proxy Pipeline Benchmark Test Result

Benchmark Specification: Input: 1080p@30 H.264 4Mbps (video with 10 objects); Storage: 1080p@30 H.264 4Mbps; Output: VideoWall, 4K@30 H264 16Mbps; Benchmark metric: Number of video channels with inference.

Smart NVR Proxy Pipeline Benchmark Test Result
Headed Visual AI Proxy Pipeline Benchmark Test Introduction

Benchmark Specification: Input: 1080p@30 H.264 4Mbps (video with 10 objects); Output: VideoWall, 4K@30 H264 16Mbps; Benchmark metric: Number of video channels.

Headed Visual AI Proxy Pipeline Benchmark Test Introduction
Headed Visual AI Proxy Pipeline Benchmark Test Result

Benchmark Specification: Input: 1080p@30 H.264 4Mbps (video with 10 objects); Output: VideoWall, 4K@30 H264 16Mbps; Benchmark metric: Number of video channels.

Pipeline_Test Device Used Input Codec Input Resolution Input Channels Model Compose Number of Monitors AI Channels Reference Platform Reference Value AVG CPU Freq(MHz) AVG CPU Util(%) AVG Memory Util(%) AVG GPU Freq(MHz) AVG GPU EU(RCS) Util(%) AVG GPU VDBox(VCS) Util(%) AVG package power(W) AVG GPU power(W) Duration(s) Result
Headed Visual AI Pipeline iGPU H264 (4Mbps) 1080p@30 9 yolov5s-416 4x4 1 9 MTL 165H (32G Mem) 16 1810.80 264.12 2.05 614.20 7.58 0.00 14.03 2.03 978.95 No Error
Headed Visual AI Pipeline iGPU H264 (4Mbps) 1080p@30 5 yolov5m-416 4x4 1 5 MTL 165H (32G Mem) 14 1906.50 249.09 2.02 734.63 6.36 0.00 14.93 3.31 705.89 No Error
Headed Visual AI Pipeline iGPU H264 (4Mbps) 1080p@30 1 yolov5m-416+efficientnet-b0 4x4 1 1 MTL 165H (32G Mem) 4 2605.62 62.64 1.57 520.39 4.81 0.00 15.05 1.53 302.54 No Error
Headed Visual AI Proxy Pipeline Benchmark Test Result

Benchmark Specification: Input: 1080p@30 H.264 4Mbps (video with 10 objects); Output: VideoWall, 4K@30 H264 16Mbps; Benchmark metric: Number of video channels.

Headed Visual AI Proxy Pipeline Benchmark Test Result
Details of Headed Visual AI Proxy Pipeline Benchmark Test

headed_visual_ai_proxy_pipeline_runner.tar.gz

VSaaS Gateway with Storage Proxy Pipeline Benchmark Test Introduction

Benchmark Specification: Input: 1080p@30 H.264 4Mbps (video with 10 objects); Storage: 1080p@30 H.265 2Mbps; Output: 1080p@30 H.265 2Mbps; Benchmark metric: Number of video channels.

VSaaS Gateway with Storage Proxy Pipeline Benchmark Test Introduction
VSaaS Gateway with Storage Proxy Pipeline Benchmark Test Result

Benchmark Specification: Input: 1080p@30 H.264 4Mbps (video with 10 objects); Storage: 1080p@30 H.265 2Mbps; Output: 1080p@30 H.265 2Mbps; Benchmark metric: Number of video channels.

Pipeline_Test Device Used Input Codec Input Resolution Input Channels Model Compose Number of Monitors AI Channels Reference Platform Reference Value AVG CPU Freq(MHz) AVG CPU Util(%) AVG Memory Util(%) AVG GPU Freq(MHz) AVG GPU EU(RCS) Util(%) AVG GPU VDBox(VCS) Util(%) AVG package power(W) AVG GPU power(W) Duration(s) Result
AI VSaaS Gateway Pipeline iGPU H264 (4Mbps) 1080p@30 6 yolov5s-416 N/A N/A 6 MTL 165H (32G Mem) 27 1598.85 219.90 1.48 676.88 0.00 0.00 14.96 2.39 591.63 No Error
AI VSaaS Gateway Pipeline iGPU H264 (4Mbps) 1080p@30 6 yolov5m-416 N/A N/A 6 MTL 165H (32G Mem) 16 1731.08 215.12 1.60 665.98 0.00 0.00 15.07 3.33 692.00 No Error
AI VSaaS Gateway Pipeline iGPU H264 (4Mbps) 1080p@30 1 yolov5m-416+efficientnet-b0 N/A N/A 1 MTL 165H (32G Mem) 4 2325.53 50.57 1.39 336.65 0.00 0.00 15.06 1.22 297.40 No Error
VSaaS Gateway with Storage Proxy Pipeline Benchmark Test Result

Benchmark Specification: Input: 1080p@30 H.264 4Mbps (video with 10 objects); Storage: 1080p@30 H.265 2Mbps; Output: 1080p@30 H.265 2Mbps; Benchmark metric: Number of video channels.

VSaaS Gateway with Storage Proxy Pipeline Benchmark Test Result
Details of VSaaS Gateway with Storage Proxy Pipeline Benchmark Test

ai_vsaas_proxy_pipeline_runner.tar.gz

License Plate Recognition(LPR) AI Proxy Pipeline Benchmark Introduction

License Plate Recognition(LPR) Pipeline Benchmark

License Plate Recognition(LPR) AI Proxy Pipeline Benchmark Introduction
License Plate Recognition(LPR) AI Proxy Pipeline Benchmark Result

License Plate Recognition(LPR) Pipeline Benchmark

Pipeline Test Model Mode Devices Used AI Channels Avg FPS Reference Platform Reference Value AVG CPU Freq(MHz) AVG CPU Util(%) AVG Memory Util(%) AVG GPU Freq(MHz) AVG GPU EU(RCS) Util(%) AVG GPU VDBox(VCS) Util(%) AVG package power(W) AVG GPU power(W) Duration(s) Result
LPR Pipeline yolov8_license_plate_detector+ch_PP-OCRv4_rec_infer Mode 2 iGPU(Dec)/iGPU(Det)/iGPU(Cls) 7 32.2 MTL 165H (32G Mem) 14 1848.39 234.75 2.71 477.26 0.00 0.00 14.09 2.66 397.20 No Error
LPR Pipeline yolov8_license_plate_detector+ch_PP-OCRv4_rec_infer Mode 3 iGPU(Dec)/iGPU(Det)/NPU(Cls) 8 30.32 MTL 165H (32G Mem) 14 1864.20 115.16 1.50 714.42 0.00 0.00 15.28 3.86 342.71 No Error
LPR Pipeline yolov8_license_plate_detector+ch_PP-OCRv4_rec_infer Mode 4 iGPU(Dec)/NPU(Det)/NPU(Cls) 5 32.88 MTL 165H (32G Mem) 14 2409.74 58.68 1.20 108.43 0.00 0.00 13.80 0.17 319.65 No Error
License Plate Recognition(LPR) AI Proxy Pipeline Benchmark Result

License Plate Recognition(LPR) Pipeline Benchmark

License Plate Recognition(LPR) AI Proxy Pipeline Benchmark Result
Details of License Plate Recognition(LPR) AI Proxy Pipeline Benchmark

lpr_proxy_pipeline_runner.tar.gz

Back to Top

Edge Software Device Qualification