Speaking as WG member: Hey Xiaohu,
I don't have a strong opinion on the algorithm or the effectiveness in the stated use case. However, if you're going to modify the existing SPF, you should use the flex algo framework which will provide backward compatibility. Thanks, Acee > On Sep 2, 2024, at 5:49 AM, Tiger Xu <[email protected]> wrote: > > Hi all, > > In the latest version, we added: > > - definition of a Path Bandwidth sub-TLV which is prefix-specific > - Improvement on the weighted ECMP load-balancing scheme in 5-stage CLOS > networks > > Any further comments or suggestions are welcome. > > Best regards, > Xiaohu > > > 发件人: [email protected] <[email protected]> > 日期: 星期一, 2024年9月2日 10:16 > 收件人: Hang Wu <[email protected]>, Hongyi Huang <[email protected]>, > Junjie Wang <[email protected]>, Peilong Wang <[email protected]>, > Qingliang Zhang <[email protected]>, Shraddha Hegde > <[email protected]>, Xiaohu Xu <[email protected]>, Yadong Liu > <[email protected]>, Yinben Xia <[email protected]>, Zongying He > <[email protected]> > 主题: New Version Notification for draft-xu-lsr-fare-03.txt > > A new version of Internet-Draft draft-xu-lsr-fare-03.txt has been successfully > submitted by Xiaohu Xu and posted to the > IETF repository. > > Name: draft-xu-lsr-fare > Revision: 03 > Title: Fully Adaptive Routing Ethernet using LSR > Date: 2024-09-01 > Group: Individual Submission > Pages: 10 > URL: https://www.ietf.org/archive/id/draft-xu-lsr-fare-03.txt > Status: https://datatracker.ietf.org/doc/draft-xu-lsr-fare/ > HTMLized: https://datatracker.ietf.org/doc/html/draft-xu-lsr-fare > Diff: https://author-tools.ietf.org/iddiff?url2=draft-xu-lsr-fare-03 > > Abstract: > > Large language models (LLMs) like ChatGPT have become increasingly > popular in recent years due to their impressive performance in > various natural language processing tasks. These models are built by > training deep neural networks on massive amounts of text data, often > consisting of billions or even trillions of parameters. However, the > training process for these models can be extremely resource- > intensive, requiring the deployment of thousands or even tens of > thousands of GPUs in a single AI training cluster. Therefore, three- > stage or even five-stage CLOS networks are commonly adopted for AI > networks. The non-blocking nature of the network become increasingly > critical for large-scale AI models. Therefore, adaptive routing is > necessary to dynamically distribute traffic to the same destination > over multiple equal-cost paths, based on network capacity and even > congestion information along those paths. > > > > The IETF Secretariat > > > _______________________________________________ > Lsr mailing list -- [email protected] > To unsubscribe send an email to [email protected] _______________________________________________ Lsr mailing list -- [email protected] To unsubscribe send an email to [email protected]
