Marvell Unveils New Custom HBM Compute Architecture to Power Cloud AI Accelerators

Marvell has just revealed its new custom HBM compute architecture, enabling XPUs to have even higher levels of compute performance and memory density.

SEE THE GALLERY – 2 IMAGES

The new HBM compute architecture is available across all of its custom silicon partners, with Marvell working with its cloud customers and leading HBM manufacturers SK hynix, Samsung and Micron to develop custom HBM solutions for next-generation XPUs.

Will Chu, Senior Vice President and General Manager of the Custom, Compute and Storage Group at Marvell, said: “Leading cloud data center operators have evolved with custom infrastructure. Improving XPUs by tailoring HBM to specific performance, power, and total cost ownership is the latest step in a new paradigm in how AI accelerators are designed and delivered. We are very grateful to be working with leading memory designers to accelerate this revolution and help cloud data center operators continue to evolve their XPUs and infrastructure at this time. The era of AI.

Raj Narasimhan, senior vice president and general manager of Micron’s Compute and Networking business unit, added: “Increasing memory capacity and bandwidth will help cloud operators efficiently scale their infrastructure to the era of AI. Strategic collaborations focused on energy efficiency, like the one we have with Marvell, will leverage Micron’s industry-leading HBM power specifications and provide hyperscalers with a robust platform to deliver the capabilities and performance optimal conditions required to evolve AI”.

Harry Yoon, Executive Vice President of Samsung Electronics and Head of Americas Product and Solution Planning, said: “HBM’s optimization for specific XPUs and software environments will significantly improve the performance of device infrastructure. cloud operators and will ensure efficient energy use. depends on these targeted efforts. We look forward to collaborating with Marvell, a leader in custom computing silicon innovation.”

Sunny Kang, vice president of DRAM technology at SK hynix America, said: “By collaborating with Marvell, we can help our customers produce a more optimized solution for their workloads and infrastructure. As one of HBM’s leading pioneers, we look forward to shaping this next stage of technology evolution.”

Patrick Moorhead, CEO and Founder of Moor Insights & Strategy, added: “Customized XPUs deliver higher performance and performance per watt than general-purpose commercial solutions for specific, cloud-unique workloads. Marvell, already a player in custom compute silicon, already provides tailored solutions to leading cloud computing companies. Their latest HBM Custom Compute Architecture platform provides additional leverage to improve the total cost of ownership of custom silicon. Through strategic collaboration with leading memory manufacturers, Marvell is poised to enable cloud operators to scale their production. “XPU and accelerated infrastructure, paving the way for them to enable the future of AI.”