Nvidia has introduced its deep-learning supercomputer – the DGX-1 – into the Australian market.
The supercomputer allows data scientists to quickly deploy a platform to speed up research in deep learning – a field of study in artificial intelligence that allows computers to process vast amounts of data.
DGX-1 is Nvidia’s first device to use its new Pascal GPU architecture – a purpose built engine for supercomputers. It uses the Tesla P100 advanced data centre accelerator to provide the compute power of 250 CPU-based servers in one system.
According to the vendor, it also includes the NVLink interconnect for application scalability, 16nm FinFET fabrication for energy efficiency, and can deliver more than 21 teraflops of peak performance.
It comes with Nvidia’s software suite for designing deep neural networks, access to cloud management service for container creation, system updates and application repository.
Australian Centre for Robotic Vision chief investigator Tom Drummond said the new system to “accelerate the pace of discovery, while industry can develop smarter products and services.”
“We have been leveraging Nvidia technologies to research ways to make robots see, learn and respond just like humans. The Nvidia DGX-1 is a powerful solution that will help us accelerate our research and uncover findings.”
The DGX-1 will be available in the third quarter of 2016 directly from Nvidia or through select partners.
-1.jpg&w=100&c=1&s=0)
.png&w=100&c=1&s=0)
