NVIDIA Tesla V100 Tensor Core

Decrease your computing time from weeks and months to mere hours with NVIDIA Tesla 100V. This device’s 640 tensor cores allowed it to overcome the long-standing 100 teraFLOPS barrier and give the world a new meaning of deep learning performance. Several Tesla V100 GPUs can be connected together with speeds of up to 300 GB per second, giving you the opportunity to put together the most powerful servers to date. 

High performance computing is crucial to every branch of science in today’s world: whether it’s used by a pharmaceutical company to design new drugs, by a meteorology station to predict weather or by geneticists to analyze the genomes of millions of people. But everything HPC does can be made even more effective by combining it with artificial intelligence. NVIDIA Tesla V100 is made specifically for the task of bringing HPC and AI together due to its tensor cores and NVIDIA CUDA cores that are combined in one architecture.  

With an increasing demand in the world hyperscale companies have been becoming increasingly more involved with AI, but so far demand is greatly outweighing supply. This can be readily mitigated with Tesla V100 that delivers top-notch performance in standard hyperscale server racks. A single V100 GPU provides 47 times better inference than a regular CPU server.