Published in Graphics

Nvida shows off new GPU powered-AI supercomputer

by on28 March 2018


Updated DGX-2

The boss of the outfit named after a Roman vengeance demon, Nvidia CEO Jensen Huang took to the stage at GTC to unveil some GPU-powered innovations for machine learning, including a new AI supercomputer and an updated version of the company's  Tesla V100 GPU that now has  32GB of onboard HBM2 memory.

Last year, Nvidia was showing off the DGX-1 AI supercomputer, and now new DGX-2 has double the number of Tesla V100 processing modules for double the number of hamsters running around Nvidia's GPU wheels. The DGX-2 has four times the available memory space, thanks to the updated Tesla V100's larger 32GB of memory.

Nvidia's new NVSwitch technology is a full crossbar GPU interconnect fabric that allows NVIDIA's platform to scale to up to 16 GPUs and use their memory space contiguously. The DGX-1  platform was limited to eight total GPU complexes and associated memory.

Nvidia claims NVSwitch is five times faster than the fastest PCI Express switch and offers an aggregate 2.4TB per second of bandwidth.

 

 

Last modified on 28 March 2018
Rate this item
(0 votes)

Read more about: