[Hardware] Nvidia’s next-gen Hopper GPU revealed: 3 things PC gamers need to know

[PrO_____]

 

Welcome to the next generation of GPUs. During his GTC 2022 keynote on Tuesday morning, Nvidia CEO Jensen Huang formally revealed “Hopper,” the graphics architecture powering the company’s next generation of products, and welp, it looks like an absolute monster—at least in data-center form.  Hopper’s debut takes shape in the data center-focused Nvidia H100, a beastly chip wielding a whopping 80 million transistors, the first use of next-gen HBM3 memory, and advances in the company’s blazing-fast NVLink interconnect technology. Hopper is “our largest generational leap ever,” Huang proudly declared.  Introducing new architectures at GTC is par for the course for Nvidia; Ampere, Volta, and Pascal also stepped into the spotlight there. And just like with those past reveals, while the H100 is heavily specialized for data-center uses, with a tensor core-laden hardware configuration designed to optimize deep-learning tasks, that doesn’t mean we humble PC gamers can’t infer some information from Hopper’s AI-centric reveal, even if the RTX 4080 and its ilk wind up running on a different “Lovelace” architecture like some current rumors suggest.