The newly rebranded ai company, (as all tech giants do lately) Nvidia announced its latest innovation, the B200 GPU.
Specifically intended for training Large Models (LLM), this impressive powerhose of a GPU comes at a $40.000 price.
But can it run Crysis?
I surely think it can! Without going into too many technical details, we mention: 20 petaflops of FP4 processing power, thanks to its 208 billion transistors. And for the GB200, which integrates two B200 GPU, Nvidia claims to boost performance for Large Language Model inference tasks by 30-fold.
And as always, the most tangible improvment is the energy consumption—by up to 25 times compared to the older H100 model, according to Nvidia.
More to read on Nvidia.com.