If you’re running deep learning inference on CPUs that support low-precision math, balancing model accuracy with energy-efficient performance is important for accelerating inference while reducing memory bandwidth and improving operations per cycle..
In this webinar, Intel Senior Software Engineer Andrey Malyshev will show you how to achieve these benefits using the latest release of the Intel® Distribution of OpenVINO™ toolkit. Topics covered include:
- Introduction to a post-training quantization process with support for INT8 model inference on Intel® processors
- Best practices for leveraging model precision to improve inference throughput
- Parallelization techniques to boost CPU performance in multicore systems
Get the software
Be sure to download the latest release of Intel Distribution of OpenVINO. It’s free.
Additional resources
- Read the blog Introducing INT8 Quantization for Fast CPU Inference
- Get the developer guide Intel Distribution of OpenVINO toolkit Model Optimizer
- Low-Precision 8-bit Integer Inference
OpenVINO is a trademark of Intel Corporation or its subsidiaries in the U.S. and/or other countries.