NVIDIA Server

NVIDIA’s products present a range of professional graphics cards designed for both rackmount servers and powerful workstations. These are very high-quality graphic products with large memory and a special set of drivers that are adapted to work in CAD and other professional applications as well as remote collaboration, virtual desktop, 3D visualization and rendering, but also for machine learning. Compared to desktop graphics cards, they are often also optimized for lower consumption.

Theoretical performance
535.28 TFLOPS
Graphics Engine
Hopper

Graphics memory
VRAM, these days primarily of GDDR type, is a synchronous memory, similar to standard RAM. However, in the case of graphic memory, memory chips with faster throughput and multiple data transfer rates are concerned. The result is a much faster buffering of data that the graphics card or coprocessor calculates and passes to the processor.

Memory size
640 GB
Memory type
HBM2

CUDA Technology
Users of professional applications can benefit from CUDA graphics stream processors thanks to CUDA architecture. This allows the raw power of the graphics card to be used for specific calculations, accelerating tasks significantly compared to a classic processor, which is limited by fewer cores.

Stream processors
135,168

Number of tensor cores
4,224

PCI Express
PCI Express is an interface that typically takes the form of an expansion slot, ensuring the modularity of the entire system, whether for GPUs, network cards, controllers, M.2 drives, or other expansion cards. Newer generations and wider interfaces offer higher performance and throughput. Most modern graphics cards use 16 lanes to connect with the processor. Currently, the most up-to-date generation is PCI Express 4.0 with a speed of 2 Gb/s per lane.

BUS
PCIe 5.0 x16