Skip to yearly menu bar Skip to main content


Poster
in
Workshop: ML with New Compute Paradigms

Accelerating AI Performance using Anderson Extrapolation on GPUs

Saleem Al Dajani · David Keyes

[ ] [ Project Page ]
Sun 15 Dec noon PST — 1:40 p.m. PST

Abstract:

We present a novel approach for accelerating AI performance by leveraging Anderson extrapolation, a vector-to-vector mapping technique based on a window of historical iterations. By identifying the crossover point where a mixing penalty is incurred, the method focuses on reducing iterations to convergence, with fewer more compute-intensive but generally cacheable iterations, balancing speed and memory usage with accuracy and algorithmic stability, respectively. We demonstrate significant improvements, in both training and inference, motivated by scalability and efficiency extensions to the realm of high-performance computing (HPC).

Chat is not available.