Fire-Flyer AI-HPC: A Cost-Effective Software-Hardware Co-Design for Deep Learning

TL;DR

The paper presents Fire-Flyer AI-HPC, a hardware-software co-design framework that significantly reduces costs and energy consumption in deep learning HPC systems while maintaining high performance.

Contribution

It introduces a novel co-design architecture and software stack that enhances scalability and efficiency in AI-HPC, with practical deployment on large GPU clusters.

Findings

Cost reduced by 50% compared to DGX-A100

Energy consumption decreased by 40%

Achieved high scalability through overlapping computation and communication

Abstract

The rapid progress in Deep Learning (DL) and Large Language Models (LLMs) has exponentially increased demands of computational power and bandwidth. This, combined with the high costs of faster computing chips and interconnects, has significantly inflated High Performance Computing (HPC) construction costs. To address these challenges, we introduce the Fire-Flyer AI-HPC architecture, a synergistic hardware-software co-design framework and its best practices. For DL training, we deployed the Fire-Flyer 2 with 10,000 PCIe A100 GPUs, achieved performance approximating the DGX-A100 while reducing costs by half and energy consumption by 40%. We specifically engineered HFReduce to accelerate allreduce communication and implemented numerous measures to keep our Computation-Storage Integrated Network congestion-free. Through our software stack, including HaiScale, 3FS, and HAI-Platform, we achieved substantial scalability by overlapping computation and communication. Our system-oriented experience from DL training provides valuable insights to drive future advancements in AI-HPC.