Performance Analysis of Deep Learning Workloads on a Composable System

TL;DR

This paper evaluates the performance of deep learning workloads on a flexible, composable system architecture, demonstrating how resource reconfiguration impacts performance and enabling early bottleneck detection.

Contribution

It introduces a composable infrastructure for deep learning workloads and provides experimental insights into resource configuration effects.

Findings

Resource aggregation impacts deep learning performance

Reconfigurable system helps identify bottlenecks early

Flexible infrastructure supports diverse workload experimentation

Abstract

A composable infrastructure is defined as resources, such as compute, storage, accelerators and networking, that are shared in a pool and that can be grouped in various configurations to meet application requirements. This freedom to 'mix and match' resources dynamically allows for experimentation early in the design cycle, prior to the final architectural design or hardware implementation of a system. This design provides flexibility to serve a variety of workloads and provides a dynamic co-design platform that allows experiments and measurements in a controlled manner. For instance, key performance bottlenecks can be revealed early on in the experimentation phase thus avoiding costly and time consuming mistakes. Additionally, various system-level topologies can be evaluated when experimenting with new System on Chip (SoCs) and new accelerator types. This paper details the design of an enterprise composable infrastructure that we have implemented and made available to our partners in the IBM Research AI Hardware Center (AIHC). Our experimental evaluations on the composable system give insights into how the system works and evaluates the impact of various resource aggregations and reconfigurations on representative deep learning benchmarks.