Accelerating AI Performance using Anderson Extrapolation on GPUs

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Dajani, Saleem Abdul Fattah Ahmed Al, Keyes, David E.
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866913617809506304
author Dajani, Saleem Abdul Fattah Ahmed Al
Keyes, David E.
author_facet Dajani, Saleem Abdul Fattah Ahmed Al
Keyes, David E.
contents 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 (Fig. 1) 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).
format Preprint
id arxiv_https___arxiv_org_abs_2410_19460
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Accelerating AI Performance using Anderson Extrapolation on GPUs
Dajani, Saleem Abdul Fattah Ahmed Al
Keyes, David E.
Machine Learning
Artificial Intelligence
Numerical Analysis
Performance
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 (Fig. 1) 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).
title Accelerating AI Performance using Anderson Extrapolation on GPUs
topic Machine Learning
Artificial Intelligence
Numerical Analysis
Performance
url https://arxiv.org/abs/2410.19460