Towards Easy and Realistic Network Infrastructure Testing for Large-scale Machine Learning

2025/04/29 by Yoo, Jinsun, ChonLam Lao, Lianjie Cao +10 · 1 citation
Computer Science · #Advanced Neural Network Applications #Artificial Intelligence (cs.AI) #Distributed #FOS: Computer and information sciences #FOS: Electrical engineering #Networking and Internet Architecture (cs.NI) #Parallel #Parallel Computing and Optimization Techniques #Software-Defined Networks and 5G #Systems and Control (eess.SY) #and Cluster Computing (cs.DC) #electronic engineering #information engineering

paper · pdf · doi:10.48550/arxiv.2504.20854

openalex publication_date 2025/04/29 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

This paper lays the foundation for Genie, a testing framework that captures the impact of real hardware network behavior on ML workload performance, without requiring expensive GPUs. Genie uses CPU-initiated traffic over a hardware testbed to emulate GPU to GPU communication, and adapts the ASTRA-sim simulator to model interaction between the network and the ML workload.

Cited by

• STAGE: A Symbolic Tensor grAph GEnerator for distributed AI system co-design

Related

• An Extensible Software Transport Layer for GPU Networking
• Homunculus: Auto-Generating Efficient Data-Plane ML Pipelines for Datacenter Networks
• MAD Max Beyond Single-Node: Enabling Large Machine Learning Model Acceleration on Distributed Systems
• Towards Universal Performance Modeling for Machine Learning Training on Multi-GPU Platforms
• TonY: An Orchestrator for Distributed Machine Learning Jobs