Programming Novel AI Accelerators for Scientific Computing

Tutorial at SciFM Summer-school-2024 at University of Michigan Scientific applications are increasingly adopting Artificial Intelligence (AI) techniques to advance science. There are specialized hardware accelerators designed and built to run AI applications efficiently. With a wide diversity in the hardware architectures and software stacks of these systems, it is challenging to understand the differences between these accelerators, their capabilities, programming approaches, and how they perform, particularly for scientific applications.

In this tutorial, we will cover an overview of the AI accelerators landscape with a focus on SambaNova, Cerebras, Graphcore, Groq, and Nvidia systems along with architectural features and details of their software stacks. We will have hands-on exercises that will help attendees understand how to program these systems by learning how to refactor codes written in standard AI framework implementations and compile and run the models on these systems. The tutorial will enable the attendees with an understanding of the key capabilities of emerging AI accelerators and their performance implications for scientific applications.

Agenda

Date: Wednesday, 17 July 2024 Time: 10:30 AM - 4:45 PM EST

Time (EST) Topic
10.30 - 10:45 Murali Emani(ANL) [Slides]
10.45 - 11.15 Sylvia Howland (Cerebras Systems) [Slides]
11.15 - 11.45 Vijay Tatkar (SambaNova Systems) [Slides]
11.45 - 12.00 Break
12.00 - 12.30 Chad Martin (Graphcore)[Slides]
12.30 - 01.00 Sanjif Shanmugavelu, Hatice Ozen (Groq) [Slides]
01.00 - 02.00 Lunch
02.00 - 02:30 Sam Foreman (ANL) (LLMs on Nvidia) [Slides]
02.30 - 04:00 Hands session on the AI Testbed: Sid Raskar, Varuni Sastry (ANL)
04.00 - 04.15 Break
04.15 - 04:45 Open Discussion, Q/A

Hands-On Session

Director’s Discretionary Allocation Program

To gain access to AI Testbeds at ALCF, you may apply for Director’s Discretionary Allocation Program

The ALCF Director’s Discretionary program provides “start up” awards to researchers working to achieve computational readiness for for a major allocation award. Once this request is approved, please use the following steps to set up accounts.

Useful Links

Acknowledgements

Contributors: Siddhisanket (Sid) Raskar, Varuni Sastry, Bill Arnold, Murali Emani.

This research used resources of the Argonne Leadership Computing Facility, which is a DOE Office of Science User Facility supported under Contract DE-AC02-06CH11357.