Accelerating AI Performance using Anderson Extrapolation on GPUs
This work addresses efficiency and scalability issues in high-performance computing for AI applications, but appears incremental as it applies an existing technique to a new context.
The paper tackles the problem of accelerating AI performance by using Anderson extrapolation on GPUs to reduce iterations to convergence, demonstrating significant improvements in both training and inference.
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).