PyTorch ATX: Accelerating GPU Performance with Triton | Sponsored by Red Hat

The Triton framework provides a hardware agnostic way of programming and targeting GPUs. As Triton becomes more widely adopted, it will be essential in understanding how to write, optimize and troubleshoot the Triton kernel in order to optimize GPU efficiency for algorithms.

Join the PyTorch ATX community to learn how Red Hat, Intel, AMD, IBM Research, and the University of Texas are working on developing Triton kernels.

Presentations

• Introduction - Steve Watt, Red Hat
• Why Texas? Powering the AI Revolution - Jason Meaux, PyTorch ATX
• Triton: Developing for Vendor Neutral Hardware Acceleration - Steven Royer, Red Hat
• Triton for vLLM - Burkhard Ringlein and Jamie Yang, IBM Research; and Rishi Astra, University of Texas at Austin
• Triton Use at AMD - Jason Furmanek, AMD
• Triton Framework and Release Process - Areg Melik-Adamyan, Intel

When: Wednesday April 30th, 5pm-8pm
Where: Lavaca Classroom (room 4.204) on the 4th floor of Robert B. Rowling Hall on the UT Austin campus. Set your Google Maps directions for the Rowling Hall Pay Garage address: 1907 Guadalupe Street, Austin, TX 78705
Parking: Park in the Rowling Hall Pay Garage (entrance on 20th Street). From parking levels B5 or B6, take the Zlotnik Family Ballroom Elevators to the 4th floor.

Food and beverages will be provided.

Sign‑ups are capped at 100 due to room capacity. If you can no longer attend, please change your RSVP status to "Not Going" on Meetup so that folks on the waitlist can attend.