AMD is shopping for Samsung’s HBM4 high-bandwidth reminiscence for its upcoming “AI accelerator”, the Intuition MI455X GPU, in addition to sixth gen AMD Epyc CPUs codenamed Venice. AI information middle operators can mix Intuition GPUs with Epyc CPUs and rack-scale architectures such because the AMD Helios platform to help next-gen AI techniques.
The memorandum of understanding between Samsung and AMD on these issues was introduced at present at Samsung’s most superior chip manufacturing advanced in Pyeongtaek, Korea. The signing ceremony was attended by AMD CEO Dr. Lisa Su and Samsung Electronics CEO and Vice Chairman Younger Hyun Jun.
Younger Hyun Jun mentioned:
Samsung and AMD share a dedication to advancing AI computing, and this settlement displays the rising scope of our collaboration. From industry-leading HBM4 and next-generation reminiscence architectures to cutting-edge foundry and superior packaging, Samsung is uniquely positioned to ship unmatched turnkey capabilities that help AMD’s evolving AI roadmap.
Dr. Lisa Su mentioned:
Powering the subsequent era of AI infrastructure requires deep collaboration throughout the {industry}. We’re thrilled to broaden our work with Samsung, bringing collectively their management in superior reminiscence with our Intuition GPUs, Epyc CPUs and rack-scale platforms. Integration throughout the complete computing stack, from silicon to system to rack, is important to accelerating AI innovation that interprets into real-world impression at scale.
Samsung’s HBM4 is made on a 10nm course of and a 4nm logic base die, and options processing speeds of as much as 13Gbps and a most bandwidth of three.3TBps. Samsung and AMD will even work collectively on high-performance DDR5 reminiscence optimized for the sixth gen Epyc CPUs. The 2 firms will even “focus on alternatives for foundry partnership”, the official press launch says, “by which Samsung would supply foundry companies for next-generation AMD merchandise”.
Supply
