Profile TensorFlow* Workloads to Uncover Optimization Opportunities

TensorFlow* is among the most popular frameworks used to create machine learning and deep learning applications. Applications can include image and voice recognition, real-time translation, drug discovery, and recommender systems. (Google* ads are powered by TensorFlow, by the way.)

This session is for TensorFlow developers who recurrently (because it is necessary) give a thorough performance analysis—uncovering bottlenecks, threading inefficiencies, scalability issues, and more—to deliver highly performant machine learning and deep learning applications.

Topics include:

• An introduction to Intel® VTune™ Profiler, Intel's flagship performance analysis tool for applications, systems, and system configurations.
• How to use key capabilities of the tool to analyze TensorFlow workloads and identify bottlenecks and optimization opportunities.
• A demo.

Skill level: Intermediate
 

Get a stand-alone version of Intel VTune Profiler or download as part of the Intel® oneAPI Base Toolkit.