"The networking for this system will be a scale-up Ethernet implementation that will use UALink over Ethernet, featuring a purpose-built Juniper Networks switch based on Broadcom's Tomahawk 6 network silicon, which boasts 102.4 Tbps of aggregate bandwidth. UALink, or Ultra Accelerator Link, is an open standard alternative to Nvidia's NVLink technology for interconnecting clusters of GPUs, the specifications for which were published earlier this year."
"HPE is throwing its weight behind AMD's Helios rack-scale architecture and will offer this as part of its AI portfolio next year, including a purpose-built Juniper Networks scale-up switch. Announced ahead of HPE's Discover event in Barcelona this week, the company claims it will be one of the first companies to offer a turnkey rack system for large-scale AI training and inference, based on AMD's reference design."
"However, it appears that HPE and Broadcom, which is also working to develop the scale-up switch, believe that you don't need to build a network using actual UALink hardware if you can simply run the protocol over Ethernet, so that's what is happening here. "This is an industry first scale-up solution using Ethernet, standard Ethernet. So that means it's 100 percent open standard and avoids proprietary vendor lock-in, leverages proven HPE Juniper networking technology to deliver scale and optimal performance for AI workloads,""
HPE will provide a turnkey rack system based on AMD's Helios rack-scale architecture for large-scale AI training and inference, with global availability planned for 2026. Helios is designed to treat a rack of accelerator-equipped nodes as a single large GPU, positioned against systems like Nvidia's DGX GB200 NVL72. The platform will leverage AMD Instinct MI455X GPUs and 6th-gen Epyc Venice CPUs expected next year. Networking uses a scale-up Ethernet approach running the UALink protocol over Ethernet and a purpose-built Juniper switch using Broadcom Tomahawk 6 silicon delivering 102.4 Tbps. HPE and Broadcom favor protocol-over-Ethernet to avoid proprietary hardware lock-in.
Read at Theregister
Unable to calculate read time
Collection
[
|
...
]