Semi-automated OpenVINO benchmark_app with variable parameters

Related tags

Deep Learning benchmark automation deep-neural-networks deep-learning intel inference deeplearning performance-test performance-testing automation-test openvino performance-ben
Overview

Semi-automated OpenVINO benchmark_app with variable parameters

Description

This program allows the users to specify variable parameters in the OpenVINO benchmark_app and run the benchmark with all combinations of the given parameters automatically.
The program will generate the report file in the CSV format with coded date and time file name ('result_DDmm-HHMMSS.csv'). You can analyze or visualize the benchmark result with MS Excel or a spreadsheet application.

The program is just a front-end for the OpenVINO official benchmark_app.
This program utilizes the benchmark_app as the benchmark core logic. So the performance result measured by this program must be consistent with the one measured by the benchmark_app.
Also, the command line parameters and their meaning are compatible with the benchmark_app.

Requirements

  • OpenVINO 2022.1 or higher
    This program is not compatible with OpenVINO 2021.

How to run

  1. Install required Python modules.
python -m pip install --upgrade pip setuptools
python -m pip install -r requirements.txt
  1. Run the auto benchmark (command line example)
python auto_benchmark_app.py -m resnet.xml -niter 100 -nthreads %1,2,4,8 -nstreams %1,2 -d %CPU,GPU -cdir cache

With this command line, -nthreads has 4 options (1,2,4,8), -nstreams has 2 options (1,2), and -d option has 2 options (CPU,GPU). As the result, 16 (4x2x2) benchmarks will be performed in total.

Parameter options

You can specify variable parameters by adding following prefix to the parameters.

Prefix Type Description/Example
$ range $1,8,2 == range(1,8,2) => [1,3,5,7]
All range() compatible expressions are possible. e.g. $1,5 or $5,1,-1
% list %CPU,GPU => ['CPU', 'GPU'], %1,2,4,8 => [1,2,4,8]
@ ir-models @models == IR models in the './models' dir => ['resnet.xml', 'googlenet.xml', ...]
This option will recursively search the '.xml' files in the specified directory.

Examples of command line

python auto_benchmark_app.py -cdir cache -m resnet.xml -nthreads $1,5 -nstreams %1,2,4,8 -d %CPU,GPU

  • Run benchmark with -nthreads=range(1,5)=[1,2,3,4], -nstreams=[1,2,4,8], -d=['CPU','GPU']. Total 32 combinations.

python auto_benchmark_app.py -m @models -niter 100 -nthreads %1,2,4,8 -nstreams %1,2 -d CPU -cdir cache

  • Run benchmark with -m=[all .xml files in models directory], -nthreads = [1,2,4,8], -nstreams=[1,2].

Example of a result file

The last 4 items in each line are the performance data in the order of 'count', 'duration (ms)', 'latency AVG (ms)', and 'throughput (fps)'.

#CPU: Intel(R) Core(TM) i7-10700K CPU @ 3.80GHz
#MEM: 33947893760
#OS: Windows-10-10.0.22000-SP0
#OpenVINO: 2022.1.0-7019-cdb9bec7210-releases/2022/1
#Last 4 items in the lines : test count, duration (ms), latency AVG (ms), and throughput (fps)
benchmark_app.py,-m,models\FP16\googlenet-v1.xml,-niter,100,-nthreads,1,-nstreams,1,-d,CPU,-cdir,cache,100,772.55,30.20,129.44
benchmark_app.py,-m,models\FP16\resnet-50-tf.xml,-niter,100,-nthreads,1,-nstreams,1,-d,CPU,-cdir,cache,100,1917.62,75.06,52.15
benchmark_app.py,-m,models\FP16\squeezenet1.1.xml,-niter,100,-nthreads,1,-nstreams,1,-d,CPU,-cdir,cache,100,195.28,7.80,512.10
benchmark_app.py,-m,models\FP16-INT8\googlenet-v1.xml,-niter,100,-nthreads,1,-nstreams,1,-d,CPU,-cdir,cache,104,337.09,24.75,308.53
benchmark_app.py,-m,models\FP16-INT8\resnet-50-tf.xml,-niter,100,-nthreads,1,-nstreams,1,-d,CPU,-cdir,cache,100,1000.39,38.85,99.96
benchmark_app.py,-m,models\FP16-INT8\squeezenet1.1.xml,-niter,100,-nthreads,1,-nstreams,1,-d,CPU,-cdir,cache,104,64.22,4.69,1619.38
benchmark_app.py,-m,models\FP32\googlenet-v1.xml,-niter,100,-nthreads,1,-nstreams,1,-d,CPU,-cdir,cache,100,778.90,30.64,128.39
benchmark_app.py,-m,models\FP32\resnet-50-tf.xml,-niter,100,-nthreads,1,-nstreams,1,-d,CPU,-cdir,cache,100,1949.73,76.91,51.29
benchmark_app.py,-m,models\FP32\squeezenet1.1.xml,-niter,100,-nthreads,1,-nstreams,1,-d,CPU,-cdir,cache,100,182.59,7.58,547.69
benchmark_app.py,-m,models\FP32-INT8\googlenet-v1.xml,-niter,100,-nthreads,1,-nstreams,1,-d,CPU,-cdir,cache,104,331.73,24.90,313.51
benchmark_app.py,-m,models\FP32-INT8\resnet-50-tf.xml,-niter,100,-nthreads,1,-nstreams,1,-d,CPU,-cdir,cache,100,968.38,38.45,103.27
benchmark_app.py,-m,models\FP32-INT8\squeezenet1.1.xml,-niter,100,-nthreads,1,-nstreams,1,-d,CPU,-cdir,cache,104,67.70,5.04,1536.23
benchmark_app.py,-m,models\FP16\googlenet-v1.xml,-niter,100,-nthreads,2,-nstreams,1,-d,CPU,-cdir,cache,100,1536.14,15.30,65.10
benchmark_app.py,-m,models\FP16\resnet-50-tf.xml,-niter,100,-nthreads,2,-nstreams,1,-d,CPU,-cdir,cache,100,3655.59,36.50,27.36
benchmark_app.py,-m,models\FP16\squeezenet1.1.xml,-niter,100,-nthreads,2,-nstreams,1,-d,CPU,-cdir,cache,100,366.73,3.68,272.68
benchmark_app.py,-m,models\FP16-INT8\googlenet-v1.xml,-niter,100,-nthreads,2,-nstreams,1,-d,CPU,-cdir,cache,100,872.87,8.66,114.56
benchmark_app.py,-m,models\FP16-INT8\resnet-50-tf.xml,-niter,100,-nthreads,2,-nstreams,1,-d,CPU,-cdir,cache,100,1963.67,19.54,50.93
benchmark_app.py,-m,models\FP16-INT8\squeezenet1.1.xml,-niter,100,-nthreads,2,-nstreams,1,-d,CPU,-cdir,cache,100,242.28,2.34,412.74
benchmark_app.py,-m,models\FP32\googlenet-v1.xml,-niter,100,-nthreads,2,-nstreams,1,-d,CPU,-cdir,cache,100,1506.14,14.96,66.39
benchmark_app.py,-m,models\FP32\resnet-50-tf.xml,-niter,100,-nthreads,2,-nstreams,1,-d,CPU,-cdir,cache,100,3593.88,35.88,27.83
benchmark_app.py,-m,models\FP32\squeezenet1.1.xml,-niter,100,-nthreads,2,-nstreams,1,-d,CPU,-cdir,cache,100,366.28,3.56,273.01
benchmark_app.py,-m,models\FP32-INT8\googlenet-v1.xml,-niter,100,-nthreads,2,-nstreams,1,-d,CPU,-cdir,cache,100,876.52,8.69,114.09
benchmark_app.py,-m,models\FP32-INT8\resnet-50-tf.xml,-niter,100,-nthreads,2,-nstreams,1,-d,CPU,-cdir,cache,100,1934.72,19.25,51.69

END

You might also like...
Python wrapper class for OpenVINO Model Server. User can submit inference request to OVMS with just a few lines of code

Python wrapper class for OpenVINO Model Server. User can submit inference request to OVMS with just a few lines of code.

Yasunori Shimura 7 Jul 27, 2022
Experimental Python implementation of OpenVINO Inference Engine (very slow, limited functionality). All codes are written in Python. Easy to read and modify.
Experimental Python implementation of OpenVINO Inference Engine (very slow, limited functionality). All codes are written in Python. Easy to read and modify.

PyOpenVINO - An Experimental Python Implementation of OpenVINO Inference Engine (minimum-set) Description The PyOpenVINO is a spin-off product from my

Yasunori Shimura 7 Oct 31, 2022
CoSMA: Convolutional Semi-Regular Mesh Autoencoder. From Paper
CoSMA: Convolutional Semi-Regular Mesh Autoencoder. From Paper "Mesh Convolutional Autoencoder for Semi-Regular Meshes of Different Sizes"

Mesh Convolutional Autoencoder for Semi-Regular Meshes of Different Sizes Implementation of CoSMA: Convolutional Semi-Regular Mesh Autoencoder arXiv p

Fraunhofer SCAI 10 Oct 11, 2022
CausalNLP is a practical toolkit for causal inference with text as treatment, outcome, or "controlled-for" variable.

CausalNLP CausalNLP is a practical toolkit for causal inference with text as treatment, outcome, or "controlled-for" variable. Install pip install -U

Arun S. Maiya 95 Jan 3, 2023
Python Implementation of Chess Playing AI with variable difficulty

Chess AI with variable difficulty level implemented using the MiniMax AB-Pruning Algorithm

Ali Imran 7 Feb 20, 2022
The official implementation of Variable-Length Piano Infilling (VLI).
The official implementation of Variable-Length Piano Infilling (VLI).

Variable-Length-Piano-Infilling The official implementation of Variable-Length Piano Infilling (VLI). (paper: Variable-Length Music Score Infilling vi

null 29 Sep 1, 2022
Benchmark for Answering Existential First Order Queries with Single Free Variable
Benchmark for Answering Existential First Order Queries with Single Free Variable

EFO-1-QA Benchmark for First Order Query Estimation on Knowledge Graphs This repository contains an entire pipeline for the EFO-1-QA benchmark. EFO-1

HKUST-KnowComp 14 Oct 24, 2022
Llvlir - Low Level Variable Length Intermediate Representation

Low Level Variable Length Intermediate Representation Low Level Variable Length

Michael Clark 2 Jan 24, 2022
Automates Machine Learning Pipeline with Feature Engineering and Hyper-Parameters Tuning :rocket:
Automates Machine Learning Pipeline with Feature Engineering and Hyper-Parameters Tuning :rocket:

MLJAR Automated Machine Learning Documentation: https://supervised.mljar.com/ Source Code: https://github.com/mljar/mljar-supervised Table of Contents

MLJAR 2.4k Dec 31, 2022
Owner
Yasunori Shimura
Yasunori Shimura
GitHub
This is a repository for a semantic segmentation inference API using the OpenVINO toolkit

BMW-IntelOpenVINO-Segmentation-Inference-API This is a repository for a semantic segmentation inference API using the OpenVINO toolkit. It's supported

BMW TechOffice MUNICH 34 Nov 24, 2022
📚 A collection of Jupyter notebooks for learning and experimenting with OpenVINO 👓

A collection of ready-to-run Python* notebooks for learning and experimenting with OpenVINO developer tools. The notebooks are meant to provide an introduction to OpenVINO basics and teach developers how to leverage our APIs for optimized deep learning inference in their applications.

OpenVINO Toolkit 840 Jan 3, 2023
MoveNet Single Pose on OpenVINO

MoveNet Single Pose tracking on OpenVINO Running Google MoveNet Single Pose models on OpenVINO. A convolutional neural network model that runs on RGB

null 35 Nov 11, 2022
A high-performance anchor-free YOLO. Exceeding yolov3~v5 with ONNX, TensorRT, NCNN, and Openvino supported.

YOLOX is an anchor-free version of YOLO, with a simpler design but better performance! It aims to bridge the gap between research and industrial communities. For more details, please refer to our report on Arxiv.

null 7.7k Jan 6, 2023
YOLOX is a high-performance anchor-free YOLO, exceeding yolov3~v5 with ONNX, TensorRT, ncnn, and OpenVINO supported.

Introduction YOLOX is an anchor-free version of YOLO, with a simpler design but better performance! It aims to bridge the gap between research and ind

null 7.7k Jan 3, 2023
Running Google MoveNet Multipose Tracking models on OpenVINO.

MoveNet MultiPose Tracking on OpenVINO

null 60 Nov 17, 2022
Demonstrates how to divide a DL model into multiple IR model files (division) and introduce a simplest way to implement a custom layer works with OpenVINO IR models.

Demonstration of OpenVINO techniques - Model-division and a simplest-way to support custom layers Description: Model Optimizer in Intel(r) OpenVINO(tm

Yasunori Shimura 12 Nov 9, 2022
A repository that shares tuning results of trained models generated by TensorFlow / Keras. Post-training quantization (Weight Quantization, Integer Quantization, Full Integer Quantization, Float16 Quantization), Quantization-aware training. TensorFlow Lite. OpenVINO. CoreML. TensorFlow.js. TF-TRT. MediaPipe. ONNX. [.tflite,.h5,.pb,saved_model,tfjs,tftrt,mlmodel,.xml/.bin, .onnx]

PINTO_model_zoo Please read the contents of the LICENSE file located directly under each folder before using the model. My model conversion scripts ar

Katsuya Hyodo 2.4k Jan 5, 2023
Monocular 3D pose estimation. OpenVINO. CPU inference or iGPU (OpenCL) inference.

human-pose-estimation-3d-python-cpp RealSenseD435 (RGB) 480x640 + CPU Corei9 45 FPS (Depth is not used) 1. Run 1-1. RealSenseD435 (RGB) 480x640 + CPU

Katsuya Hyodo 8 Oct 3, 2022
WHENet - ONNX, OpenVINO, TFLite, TensorRT, EdgeTPU, CoreML, TFJS, YOLOv4/YOLOv4-tiny-3L

HeadPoseEstimation-WHENet-yolov4-onnx-openvino ONNX, OpenVINO, TFLite, TensorRT, EdgeTPU, CoreML, TFJS, YOLOv4/YOLOv4-tiny-3L 1. Usage $ git clone htt

Katsuya Hyodo 49 Sep 21, 2022