The number of members of the Ultra Ethernet Alliance increases to 55, and some details of the upcoming HPC backbone technology are revealed.

The Ultra Ethernet Consortium (UEC) announced this week that the Next Generation Interconnect Alliance has grown to 55 members. As the group works to develop an initial version of the SuperSpeed ​​Ethernet standard, they have released some of the first technical details about the upcoming standard.

UEC was established in summer 2023 to develop new interconnect standards that meet the needs of AI and HPC data centers as actually (if not in Law) Alternative to InfiniBand, which is currently mainly controlled by NVIDIA. The UEC started accepting new members back in November and has added 45 new members in just five months, highlighting the huge interest in this new technology. The alliance currently has 55 members and 715 industry experts working in eight technical groups.

There’s still a lot of work to do at the UEC, as the organization lays out in its latest development blog post as the consortium works to build an Ethernet-based unified communications stack for high-performance capabilities that support artificial intelligence and high-performance networks. network. Performance computing cluster. The consortium’s technical goals include developing specifications, APIs, and source code for HyperEthernet communications, updating existing protocols, and introducing new mechanisms for telemetry, signaling, security, and congestion management. In particular, Ultra Ethernet introduces UEC Transport (UET) to improve network utilization and reduce tail latency, thereby accelerating RDMA (Remote Direct Memory Access) operations over Ethernet. Key features include multipath packet injection, flexible sequencing, and advanced congestion control to ensure efficient and reliable data transfer.

These enhancements are designed to cater to the needs of large AI and HPC clusters – there are separate profiles for each type of deployment – although everything is done in a surgical manner to enhance the technology, reusing as much of the existing Ethernet as possible network to maintain cost efficiency and interoperability.

Founding members of the alliance include AMD, Arista, Broadcom, Cisco, Eviden (an Atos company), HPE, Intel, Meta and Microsoft. After the Ultra Ethernet Alliance (UEC) began accepting new members in October 2023, many industry giants have joined the alliance, including Baidu, Dell, Huawei, IBM, Nokia, Lenovo, Super Micro and Tencent.

The consortium currently plans to publicly release the initial 1.0 version of the UEC specification sometime in the third quarter of 2024.

“There has always been a perception that the UEC is meeting the needs of the industry,” said J Metz, chair of the UEC Steering Committee. “There is a strong desire to have an open, accessible, Ethernet-based network designed specifically to meet the needs of artificial intelligence and high-performance computing workloads. This level of participation is encouraging; it helps us achieve broad interoperability and stability goals.”

While it’s clear that the Ultra Ethernet Alliance is gaining support across the industry, it’s still unclear where other industry giants like AWS and Google stand. While hardware companies can design Hyper Ethernet support into their hardware and systems, the technology ultimately exists to serve large data center and HPC system operators. So it will be interesting to see their interest in (and how quickly they adopt) emerging Ethernet backbone technologies once the hardware to incorporate them is ready.

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *