NetSmith: An Optimization Framework for Machine-Discovered Network Topologies

Conor Green,Mithuna Thottethodi
2024-04-03
Abstract:Over the past few decades, network topology design for general purpose, shared memory multicores has been primarily driven by human experts who use their insights to arrive at network designs that balance the competing goals of performance requirements (e.g., latency, bandwidth) and cost constraints (e.g., router radix, router counts). On the other hand, there have been automatic NoC synthesis methods for SoCs to optimize for application-specific communication and objectives such as resource usage or power. Unfortunately, these techniques do not lend themselves to the general-purpose context, where directly applying these previous NoC synthesis techniques in the general-purpose context yields poor results, even worse than expert-designed networks. We design and develop an automatic network design methodology - NetSmith - to design networks for general-purpose, shared memory multicores that comprehensively outperform expert-designed networks.
Hardware Architecture,Distributed, Parallel, and Cluster Computing
What problem does this paper attempt to address?