Discover how to leverage Intel® Deep Learning Boost technology, also known as Vector Neural Network Instructions (VNNI), to accelerate deep learning inference in this advanced talk from EuroPython 2019. Explore the new set of AVX-512 instructions designed for more efficient deep learning acceleration, and learn to perform faster low-precision (INT8) inference on hardware supporting the VNNI instruction set, such as 2nd generation Intel Xeon Scalable processors. Witness the benefits of this hardware technology through a live Jupyter notebook demonstration. Gain insights into quantization, OpenVino, and calibration techniques. Prior knowledge of deep learning, inferencing, and basic awareness of hardware instruction sets is recommended for this 29-minute presentation.
Accelerate Your Deep Learning Inferencing with the Intel DL Boost Technology
EuroPython Conference via YouTube
Overview
Syllabus
Intro
What is Intel DL Boost
Quantization
OpenVino
Demonstration
Questions
Calibration
Taught by
EuroPython Conference