According to the latest rumors, Nvidia’ is planning to soon tape out its Hopper GPU’s which Nvidia is for its data center market. This new GPU will be the first for the company to feature MCM technology. The rumor comes from a cryptic tweet posted by Greymon55 which was quickly decrypted by another Twitter user which 3DCenter caught sight of.
NVIDIA’s Hopper GPU architecture, designed for the Data Center market will be exclusive for the HPC segment. As mentioned above the GPU will also be the first from NVIDIA to feature an MCM or Multi-Chip-Module design. The technology is already being utilized by its main competitors, Intel (Xe-HPC) and AMD (CDNA 2) to power their next-generation Ponte Vecchio & Instinct MI200 accelerators.
We have seen Nvidia’s roadmap recently, and according to it, the Hopper GPU was placed for launch in 2022, so officially we still have at least two years after the launch of Ampere to see the Hopper GPU. It was expected for a long time that the graphics maker plans to announce its new GPU during the next GTC event in Q2 2022.
However, the recent rumors that claim that the company is soon going to tape out its first Hopper GPUs throw quite a lot of information around. The chip is rumored to be based on TSMC’s 5nm process node and we are still unclear about the architecture of the card.
According to previous leaks and rumors, we have heard that the Hopper GPU could be comprised of two GPM (Graphics Processing Modules), each of which could feature 8 GPC (4 x 2) for a total of 144 SM units. And based on other information we could probably see the Hopper GPUs having a total of 288 SM units.
For more information related to Nvidia’s Hopper GPU for data centers stay tuned.