Skip to main content

Accelerating Machine Learning with OpenCL
Accelerating Machine Learning with OpenCL Banner
May 11, 2022

Event is now over

Presentations and other assets from this event are presented here. For information on upcoming events, click here.

Accelerating Machine Learning with OpenCL

Discover how the latest developments in OpenCL can improve ML performance

  • Wednesday, May 11, 2022
  • Duration: 60 mins
  • Start Time: 10:00 PDT, 13:00 EDT, 19:00 CEST

In this webinar members of the OpenCL working group at Khronos will share the latest updates to the OpenCL language and ecosystem that can directly benefit Machine Learning workflow performance.



  • Welcome and Introduction
    • Speaker: Neil Trevett, Khronos President, OpenCL Working Group Chair, and NVIDIA
    • An introduction to the Khronos Machine Learning Forum and update on OpenCL support for Machine Learning
  • Qualcomm Extensions for Advancing Machine Learning Acceleration
    • Speaker: Balaji Calidas, Director of Engineering. Qualcomm
    • View Abstract

      An introduction to the Qualcomm OpenCL extensions that accelerate Machine Learning. These extensions accelerate ML operations, enable edge training, reduce the CPU overhead of dispatching ML workloads and add new math builtins. This talk presents an overview of these features and Qualcomm’s ongoing investment in accelerating Machine Learning..

      Speaker Bio: Balaji is the engineering lead for OpenCL at Qualcomm and is a member of the Khronos OpenCL working group. He is actively engaged in identifying and implementing features for GPGPU use cases such as Machine Learning and Image Processing.

  • A Case Study on OpenCL vs GPU Assembly for Machine Learning Performance
    • Speaker: Roy Oursler, Intel
    • View Abstract

      The project oneDNN is an open-source cross-platform performance library for accelerating deep learning applications. In this talk, we give an analysis on implementing optimized convolutions with OpenCL C vs generating optimized assembly within oneDNN.

      Speaker Bio: Roy Oursler is an AI Algorithm Engineer working on oneDNN to improve deep learning performance on Intel GPUs.

  • Ask the Experts Q&A and Panel Discussion

Here’s your opportunity to put all questions to our panel of speakers and OpenCL ML experts.   During this session we will also encourage feedback from application and framework developers on any requirements which are not currently addressed by the existing OpenCL solutions, or development pain points which could be alleviated.

  • Panel Chair: To be confirmed
  • Panel Members: All the speakers plus additional members of the Vulkan ML subgroup.

Get All Your OpenCL ML Questions Answered

  • Attendees will be able to post questions to our speakers throughout the live event.
  • Our panel of experts will be on-hand to answer all your questions during the final Q&A session.


Khronos videos, presentations, and upcoming events. Skip to the Footer