Skip to content

Latest commit

 

History

History
 
 

109-performance-tricks

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
 
 
 
 
 
 

Introduction to Performance Tricks in OpenVINO™

These notebooks show tips on how to optimize inference in OpenVINO™.

There are different performance targets for various use case scenarios. For example, video conferencing usually requires the latency to be as short as possible. While for high-resolution video/image analysis, high throughput is typically the performance target. As a result, different optimization tricks should be applied to achieve other performance targets. In this notebook, we’ll show a set of performance tricks for optimizing inferencing latency.

Notebook Contents

  1. Performance tricks in OpenVINO for latency mode

Colab

This notebook demonstrates how to optimize the inference latency in OpenVINO™. A set of optimization tricks, including model conversion with different data precision, “AUTO” device with latency mode, shared memory, inference with a further configuration, inference on GPU, etc., are introduced.

  1. Performance tricks in OpenVINO for throughput mode

Colab

This notebook demonstrates how to optimize the inference throughput in OpenVINO™. A set of optimization tricks, including bigger batch size, “AUTO” device with throughput and cumulative throughput mode, asynchronous inference mode, etc., are introduced.

Installation Instructions

If you have not installed all required dependencies, follow the Installation Guide.