NVIDIA has opened the doors to a future of possibilities with the announcement of the NVIDIA GH200 Grace Hopper Superchip Platform. Created to manage the world’s most challenging generative AI workloads, this groundbreaking platform sets the stage for a new era of innovation and performance.
Nvidia GH200: The World’s First HBM3e Processor
The NVIDIA GH200 Grace Hopper Superchip Platform heralds the world’s first HBM3e processor. Offering a groundbreaking 3.5x more memory capacity and 3x more bandwidth compared to the current generation, it delivers unparalleled performance. “HBM3e memory, which is 50% faster than current HBM3, allows the new platform to run models 3.5x larger,” the company shared in its announcement.
Multi-GPU Connectivity and Exceptional Performance
The innovative design of the platform allows for multiple Superchips to connect via NVIDIA NVLink™, enabling deployment of giant models used for generative AI. “This high-speed, coherent technology gives the GPU full access to the CPU memory, providing a combined 1.2TB of fast memory,” NVIDIA states.
Scalable Server Design and Compatibility
The GH200 Grace Hopper Superchip Platform is designed to scale across an entire data center without compromise. NVIDIA CEO, Jensen Huang, emphasized its importance, saying,
“To meet surging demand for generative AI, data centers require accelerated computing platforms with specialized needs. The new platform delivers this with exceptional memory technology.”
Growing Demand and Availability
Leading manufacturers have already embraced the technology, and systems based on the NVIDIA GH200 Grace Hopper Superchip Platform are expected to be available in Q2 of 2024. This broad adoption was facilitated by the full compatibility with the NVIDIA MGX™ server specification.
Watch Huang’s SIGGRAPH keynote address on demand to learn more about Grace Hopper:
The NVIDIA GH200 Grace Hopper Superchip Platform stands as a testament to innovation, marking a significant milestone in the world of accelerated computing and generative AI. Its introduction promises to reshape our understanding of what’s possible, with its cutting-edge design and exceptional performance. We invite our readers to share their thoughts and insights in the comments section below.
Check out our other article on NVIDIA MGX: Next-Gen Architecture for Accelerated Computing, right here.
