MSCCL++: Rethinking GPU Communication Abstractions for AI Inference
Authors:
- Changho HwangCHChangho Hwang
- Peng ChengPCPeng Cheng
- Roshan DathathriRDRoshan Dathathri
- Abhinav JangdaAJAbhinav Jangda
- Saeed MalekiSMSaeed Maleki
- Madan MusuvathiMMMadan Musuvathi
- Olli SaarikiviOSOlli Saarikivi
- Aashaka ShahASAashaka Shah
- Ziyue YangZYZiyue Yang
- Binyang LiBLBinyang Li
- Caio RochaCRCaio Rocha
- Qinghua ZhouQZQinghua Zhou
Open Access
- 22 March 2026
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 1201-1215
Abstract
No abstract availableAll Related Versions
This publication has 3 references indexed in Scilit:
- Splitwise: Efficient Generative LLM Inference Using Phase Splitting2024
- Breaking the computation and communication abstraction barrier in distributed machine learning workloads | Proceedings of the 27th ACM International Conference on Architectural Support for Programming Languages and Operating Systems2022
- Synthesizing optimal collective algorithms2021