Nvidia announces DGX GH200 AI supercomputer

[{"selector":"#anim-ce26612d-7e90-4399-846c-21554a64e7d7","keyframes":{"opacity":[0,1]},"delay":250,"duration":2000,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-ecd9ecf9-6bc1-4da2-beb5-83a8b1052e35","keyframes":{"transform":["translate3d(-177.04918%, 0px, 0)","translate3d(0px, 0px, 0)"]},"delay":250,"duration":2000,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] NVIDIA announces new large-memory AI supercomputer DGX GH200, offering 1 exaflop of performance and 144 terabytes of shared memory

[{"selector":"#anim-e1158909-b7ed-46a4-9ce3-6171feb2604a [data-leaf-element=\"true\"]","keyframes":{"transform":["translate3d(-37.96630763314801%, 0, 0)","translate3d(0%, 0, 0)"]},"delay":0,"duration":3000,"easing":"cubic-bezier(.3,0,.55,1)","fill":"both"}] DGX GH200 uses NVLink interconnect technology with the NVLink Switch System to combine 256 GH200 superchips as a single GPU

[{"selector":"#anim-3c17f03d-03b9-4008-ab5f-fe6eb5dd00b0","keyframes":{"opacity":[0,1]},"delay":250,"duration":2000,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-c0900744-fd94-4d91-9ce8-50be0a25bb35","keyframes":{"transform":["translate3d(-177.04918%, 0px, 0)","translate3d(0px, 0px, 0)"]},"delay":250,"duration":2000,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] GH200 superchips eliminate need for traditional CPU-to-GPU PCIe connection and increases bandwidth between GPU and CPU by 7x

Save time. Stay Informed with FYI: A short news app for working professionals.

[{"selector":"#anim-5f7f7f0a-fc74-43f6-bf62-1696a63a400f [data-leaf-element=\"true\"]","keyframes":{"transform":["translate3d(-37.96630763314801%, 0, 0)","translate3d(0%, 0, 0)"]},"delay":0,"duration":3000,"easing":"cubic-bezier(.3,0,.55,1)","fill":"both"}] DGX GH200 is the first supercomputer to pair Grace Hopper Superchips with the NVIDIA NVLink Switch System, providing 48x more NVLink bandwidth

[{"selector":"#anim-f5367356-bc07-43f7-a611-14ef6fe3cca9","keyframes":{"opacity":[0,1]},"delay":250,"duration":2000,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-7ddfc14f-6257-4870-9772-6667859cd4a8","keyframes":{"transform":["translate3d(-177.04918%, 0px, 0)","translate3d(0px, 0px, 0)"]},"delay":250,"duration":2000,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] Google Cloud, Meta, and Microsoft are expected to gain access to DGX GH200 for generative AI workloads and large-scale AI initiatives

[{"selector":"#anim-c07dc8e9-b0f2-4a86-b0b9-b508a9f7378b [data-leaf-element=\"true\"]","keyframes":{"transform":["translate3d(-37.96630763314801%, 0, 0)","translate3d(0%, 0, 0)"]},"delay":0,"duration":3000,"easing":"cubic-bezier(.3,0,.55,1)","fill":"both"}] NVIDIA DGX GH200 supercomputers are expected to be available by the end of the year Via