SK hynix has just announced its new HBM3E memory standard, built for the future of AI applications and more especially -- NVIDIA's just-announced GH200 Grace Hopper GPU . SK hynix says its new HBM3E memory not only meets the industry's highest standards of speeds, which is the most important part of AI memory products, but it also boats huge capacity, heat dissipation, and user-friendliness.
We are looking at SK hynix's new HBM3E memory pumping away at 1.15TB/sec of memory bandwidth, with a 10% improvement in heat dissipation, something the company achieved by using cutting-edge technology called Advanced Mass Reflow Molded Underfill, or MR-MUF. This also provides backward compatibility that helps the adoption of HBM3E into systems that have been prepared for HBM3 (no E) without a design or structure modification.
Ian Buck, Vice President of Hyperscale and HPC Computing at NVIDIA explains: "We have a long history of working with SK hynix on High Bandwidth Memory for leading edge accelerated computing solutions. We look forward to continuing our collaboration with HBM3E to deliver the next generation of AI computing".
Meanwhile, Sungsoo Ryu, Head of DRAM Product Planning at SK hynix said that the company strengthened its market leadership while developing its new HBM3E memory by enhancing the completeness of its HBM product lineup, which is front and centre when it comes to the development of AI technology. Ryu said: "By increasing the supply share of the high-value HBM products, SK hynix will also seek a fast business turnaround".