92c/MFlops/s, Ultra-Large-Scale Neural-Network Training on a PIII Cluster

TL;DR

This paper demonstrates the distributed training of ultra-large neural networks on a cluster of Pentium III processors, achieving high performance and cost efficiency in recognizing Japanese characters from a large dataset.

Contribution

It introduces a technique for distributed training of ultra-large neural networks on commodity hardware, with a practical experiment showing high performance and cost-effectiveness.

Findings

Achieved 163.3 GFlops/s performance in training

Trained a 1.73 million parameter network on 9 million patterns

Cost-effective training with 92.4c/MFlops/s price/performance ratio

Abstract

Artificial neural networks with millions of adjustable parameters and a similar number of training examples are a potential solution for difficult, large-scale pattern recognition problems in areas such as speech and face recognition, classification of large volumes of web data, and finance. The bottleneck is that neural network training involves iterative gradient descent and is extremely computationally intensive. In this paper we present a technique for distributed training of Ultra Large Scale Neural Networks (ULSNN) on Bunyip, a Linux-based cluster of 196 Pentium III processors. To illustrate ULSNN training we describe an experiment in which a neural network with 1.73 million adjustable parameters was trained to recognize machine-printed Japanese characters from a database containing 9 million training patterns. The training runs with a average performance of 163.3 GFlops/s (single precision). With a machine cost of \$150,913, this yields a price/performance ratio of 92.4c/MFlops/s (single precision). For comparison purposes, training using double precision and the ATLAS DGEMM produces a sustained performance of 70 MFlops/s or \$2.16 / MFlop/s (double precision).