5 papers
The Multipath Reliable Connection (MRC) Transport
Rip Sohan, Eric Spada, Eric Davis +36
MRC is an open, production-grade transport designed for large-scale AI/ML training over best-effort Ethernet. It extends RoCEv2 with explicit, composable primitives for per-packet…
Resilient AI Supercomputer Networking using MRC and SRv6
Joao Araujo, Alex Chow, Mark Handley +47
Tail latency dominates the performance of synchronous pretraining jobs when running at very large scales. We describe a three-pronged approach: (1) a new RDMA-based transport proto…
Demystifying NCCL: An In-depth Analysis of GPU Communication Protocols and Algorithms
Zhiyi Hu, Siyuan Shen, Tommaso Bonato +6
The NVIDIA Collective Communication Library (NCCL) is a critical software layer enabling high-performance collectives on large-scale GPU clusters. Despite being open source with a…
In-Network Collective Operations: Game Changer or Challenge for AI Workloads?
Torsten Hoefler, Mikhail Khalilov, Josiah Clark +9
This paper summarizes the opportunities of in-network collective operations (INC) for accelerated collective operations in AI workloads. We provide sufficient detail to make this i…
Ultra Ethernet's Design Principles and Architectural Innovations
Torsten Hoefler, Karen Schramm, Eric Spada +13
The recently released Ultra Ethernet (UE) 1.0 specification defines a transformative High-Performance Ethernet standard for future Artificial Intelligence (AI) and High-Performance…