Salvato in:
| Autori principali: | , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2407.15266 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866909266442452992 |
|---|---|
| author | Le, Yanfang Pan, Rong Newman, Peter Blendin, Jeremias Kabbani, Abdul Jain, Vipin Sivaramu, Raghava Matus, Francis |
| author_facet | Le, Yanfang Pan, Rong Newman, Peter Blendin, Jeremias Kabbani, Abdul Jain, Vipin Sivaramu, Raghava Matus, Francis |
| contents | Emerging artificial intelligence (AI) and machine learning (ML) workloads present new challenges of managing the collective communication used in distributed training across hundreds or even thousands of GPUs. This paper presents STrack, a novel hardware-offloaded reliable transport protocol aimed at improving the performance of AI /ML workloads by rethinking key aspects of the transport layer. STrack optimizes congestion control and load balancing in tandem: it incorporates an adaptive load balancing algorithm leveraging ECN, while adopts RTT as multi-bit congestion indicators for precise congestion window adjustment. Additionally, STrack facilitates out-of-order delivery, selective retransmission, and swift loss recovery in hardware for multipath environment. The extensive simulation comparing STrack and RoCEv2 demonstrates that STrack outperforms RoCEv2 by up to 6X with synthetic workloads and by 27.4% with collective workloads, even with the optimized RoCEv2 system setup. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2407_15266 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | STrack: A Reliable Multipath Transport for AI/ML Clusters Le, Yanfang Pan, Rong Newman, Peter Blendin, Jeremias Kabbani, Abdul Jain, Vipin Sivaramu, Raghava Matus, Francis Networking and Internet Architecture Emerging artificial intelligence (AI) and machine learning (ML) workloads present new challenges of managing the collective communication used in distributed training across hundreds or even thousands of GPUs. This paper presents STrack, a novel hardware-offloaded reliable transport protocol aimed at improving the performance of AI /ML workloads by rethinking key aspects of the transport layer. STrack optimizes congestion control and load balancing in tandem: it incorporates an adaptive load balancing algorithm leveraging ECN, while adopts RTT as multi-bit congestion indicators for precise congestion window adjustment. Additionally, STrack facilitates out-of-order delivery, selective retransmission, and swift loss recovery in hardware for multipath environment. The extensive simulation comparing STrack and RoCEv2 demonstrates that STrack outperforms RoCEv2 by up to 6X with synthetic workloads and by 27.4% with collective workloads, even with the optimized RoCEv2 system setup. |
| title | STrack: A Reliable Multipath Transport for AI/ML Clusters |
| topic | Networking and Internet Architecture |
| url | https://arxiv.org/abs/2407.15266 |