StreamBrain: An HPC DSL for Brain-Like Neural Networks on Heterogeneous Systems

Authors: Artur Podobas, Martin Svedin, and Steven W. D. Chien (KTH Royal Institute of Technology); Ivy B. Peng (Lawrence Livermore National Laboratory); Stefano Markidis and Pawel Herman (KTH Royal Institute of Technology); Anders Lansner (KTH Royal Institute of Technology, Stockholm University); and Naresh Balaji Ravichandran (KTH Royal Institute of Technology)

Abstract: We introduce StreamBrain: a high-performance DSL for brain-like neural networks. StreamBrain supports multiple backends such as FPGAs, GPUs and CPUs on heterogeneous HPC systems while providing a convenient Keras-like interface to users. We show that training an MNIST dataset on the BCPNN model only takes 15 seconds. We empirically show that batching is critical for the BCPNN model as it allows the computational intensity to be controlled. Finally, we explored the resilience of the BCPNN model to reduced width of the numerical representation and showed that the mantissa of the double-precision (DP) computation could be reduced to a 9-bit representation, yielding nearly twice the performance of the original DP implementation.

Best Poster Finalist (BP): no

Poster: PDF
Poster summary: PDF


Back to Poster Archive Listing