Technology industry giants AMD, Broadcom, Cisco, Google, Hewlett Packard Enterprise (HPE), Intel, Meta, and Microsoft have joined forces to establish a new industry standard known as UALink (Ultra Accelerator Link) aimed at enhancing high-speed and low-latency communication for AI systems in data centers.
The primary objective of UALink is to create an open industry standard that enables more efficient communication between AI accelerators by providing OEMs, IT professionals, and system integrators with a pathway to easier integration, greater flexibility, and scalability of AI-connected data centers. The companies involved bring extensive experience in developing large-scale AI and high-performance computing (HPC) solutions that are rooted in open standards, efficiency, and robust ecosystem support.
The UALink 1.0 specification is expected to be available in Q3 of 2024 and made available to companies that join the UALink Consortium. It will enable the connection of up to 1,024 accelerators within an AI computing pod as well as allow for direct loads and stores between the memory attached to GPUs and other accelerators within the pod.
The most noticeable absentee from the UALink Consortium is market leader Nvidia, which has established its own high-speed, low latency interconnect standards.
Long time technology industry fan here in Taiwan.
