NVIDIA's new Hopper GPU architecture has been unveiled and with it comes some new DGX H100 systems powered by the monster new NVIDIA H100 GPU.
The new NVIDIA DGX H100 system has 8 x H100 GPUs per system, all connected as one gigantic insane GPU through 4th-Generation NVIDIA NVLink connectivity. This enables up to 32 petaflops at new FP8 precision, a gigantic 6x performance improvement over the previous-gen Ampere-based GPUs.
NVIDIA will be using the new Hopper-based DGX H100 systems as the "building blocks" of the next-gen NVIDIA DGX POD and NVIDIA DGX SuperPOD AI infrastructure platforms. NVIDIA's new DGX SuperPOD architecture has the new NVIDIA NVLink Switch System -- capable of up to 32 nodes and a total of 256 x H100 GPUs. At this level, we're talking about 1 exaflops of FP8 AI performance, also 6x more than its predecessor.
- Read more: NVIDIA reveals next-gen Hopper GPU architecture, H100 GPU announced
- Read more: NVIDIA can sustain the world's internet traffic with 20 x H100 GPUs
- Read more: NVIDIA Hopper GPU is up to 40x faster with new DPX instructions
- Read more: NVIDIA is turning data centers into 'AI factories' with Hopper GPU
- Read more: NVIDIA Eos: the world's fastest AI supercomputer, 4608 x DGX H100 GPUs
NVIDIA founder and CEO Jensen Huang explains: "AI has fundamentally changed what software can do and how it is produced. Companies revolutionizing their industries with AI realize the importance of their AI infrastructure. Our new DGX H100 systems will power enterprise AI factories to refine data into our most valuable resource - intelligence".