Description

Theoretical Performance:
267.64 TFLOPS

Graphics Engine:
Hopper

Graphics Memory:
VRAM, primarily of GDDR type, is a synchronous memory similar to standard RAM. However, graphics memory uses chips with faster throughput and multiple data transfer rates, allowing for much faster buffering of data calculated by the graphics card or coprocessor and passed to the processor.

Memory Size:
320 GB

Memory Type:
HBM3

CUDA Technology:
Users of professional applications benefit from CUDA graphics stream processors, leveraging CUDA architecture. This allows the raw power of the graphics card to be used for specific calculations, accelerating tasks significantly compared to traditional processors, which are limited by fewer cores.

Stream Processors:
67,584

Number of Tensor Cores:
2,112

Cooling Type:
Passive
Passively cooled graphics cards have no powered moving components. They rely on optimized airflow, especially in a server environment, to dissipate heat outside the case. This cooling type is commonly used for high-performance graphics coprocessors.

PCI Express:
PCI Express is an interface that ensures modularity for various system components such as GPUs, network cards, controllers, M.2 drives, and more. Newer generations and wider interfaces offer higher performance and throughput. Most modern graphics cards use 16 lanes to connect to the processor. The latest generation, PCI Express 4.0, provides a speed of 2 Gb/s per lane.

BUS:
PCIe 5.0 x16