Author: Morgridge Institute for Research

Breaking the GPU Bottleneck: How Distributed Computing is Expanding AI Training

For researchers at smaller institutions or those attempting to train “ensembles” — multiple versions of a machine learning model to ensure accuracy and robustness — the requirement for high-end GPUs often creates an insurmountable barrier. A Morgridge Institute and Center for High Throughput Computing team supported by the NSF presents a new path: breaking AI training into “small bites” and distributing them across a nationwide network of computing providers.