Cerebras and AMD Forge AI Hardware Alliance, Integrating Helios Servers and WSE Technology
Translated from Korean, summarized and contextualized by DistantNews.
At a glance
- AMD and Cerebras have formed an alliance in AI hardware technology, integrating Cerebras's AI chips with AMD's Helios server system.
- Cerebras will adopt AMD's Helios all-in-one server and jointly release solutions related to its Wafer Scale Engine (WSE).
- Cerebras's WSE-3 boasts significantly higher transistor counts and computational speed compared to Nvidia's B200.
In a significant move for the artificial intelligence hardware landscape, AMD and Cerebras have announced a strategic alliance focused on AI infrastructure. The partnership will see Cerebras integrate its advanced AI chips with AMD's newly unveiled Helios all-in-one server system, aiming to deliver enhanced performance and efficiency for AI workloads.
The collaboration, revealed at AMD's annual 'Advancing AI' conference in San Francisco, signifies a deepening commitment to heterogeneous computing and cloud cooperation. Cerebras will not only adopt AMD's Helios servers for its own data centers but also co-develop solutions tied to its proprietary Wafer Scale Engine (WSE). This joint effort seeks to leverage the strengths of both companies' technologies to push the boundaries of AI processing capabilities.
Cerebras, a startup founded in 2015 and known for its groundbreaking WSE technology, positions its chips as a formidable competitor in the AI hardware market. The WSE-3, its latest offering, features an impressive 900,000 cores and delivers 125 petaflops of computational power. This performance metric is substantially higher than that of Nvidia's B200, with Cerebras claiming 19 times the transistors and 28 times the processing speed. The WSE-3 also incorporates 44GB of on-chip SRAM, enabling an internal bandwidth of 27 petabytes per second and 21 petabytes of memory.
Originally published by Dong-A Ilbo in Korean. Translated, summarized, and contextualized by our editorial team with added local perspective. Read our editorial standards.