The Personal Supercomputer for Leading-Edge AI Development
Your data science team depends on computing performance to gain insights,and innovate faster through the power of deep learning and data analytics. Until now, AI supercomputing was confined to the data centre, limiting the experimentation needed to develop and test deep neural networks prior to training at scale. Now there’s a solution, offering the power to experiment with deep learning while bringing AI supercomputing performance within arm’s reach.
Groundbreaking AI, at Your Desk
Now you can get the computing capacity of 400 CPUs, in a workstation that conveniently fits under your desk, drawing less than 1/20th the power. NVIDIA® DGX Station™ delivers incredible deep learning and analytics performance, designed for the office and whisper quiet with only 1/10th the noise of other workstations. Data scientists and AI researchers can instantly boost their productivity with a workstation that includes access to optimised deep learning software and runs popular analytics software.
Get Started in Deep Learning, Faster
DGX Station breaks through the limitations of building your own deep learning platform. You could spend a month or longer, procuring, integrating, and testing hardware and software. Then additional expertise and effort are needed to optimise frameworks, libraries, and drivers. That’s valuable time and money spent on systems integration and software engineering that could be spent training and experimenting. NVIDIA DGX Station is designed to kickstart your AI initiative, with a streamlined plug-in and power-up experience that can have you training deep neural networks in just one day.
Productivity That Takes You from Desk to Data Centre
Deep learning platforms require software engineering expertise to keep today’s frameworks optimised for maximum performance, with time spent waiting on stable versions of open source software. This means hundreds of thousands of dollars in lost productivity, dwarfing the initial hardware cost. NVIDIA DGX Station includes the same software stack found in all DGX solutions. This innovative, integrated system includes access to popular deep learning frameworks, updated monthly, each optimised by NVIDIA engineers for maximised performance. It also includes access to NVIDIA DIGITS™ deep learning training application, third-party accelerated solutions, the NVIDIA Deep Learning SDK (e.g. cuDNN, cuBLAS, NCCL), CUDA® Toolkit, and NVIDIA drivers. Built on container technology powered by NVIDIA Docker, this unified deep learning software stack simplifies workflow, saving you days in re-compilation time when you need to scale your work and deploy your models in the data centre or cloud. The same workload running on DGX Station can be effortlessly migrated to a DGX-1 or the cloud, without modification.
Supercomputing Performance, at Your Desk
DGX Station brings the incredible performance of an AI supercomputer in a workstation form factor that takes advantage of innovative engineering and a water-cooled system that runs whisper-quiet. The NVIDIA DGX Station packs 500 teraFLOPS of performance, with the first and only workstation built on four NVIDIA Tesla®V100 accelerators, including innovations like next generation NVLink™ and new Tensor Core architecture. This groundbreaking solution offers:
- 72X the performance for deep learning training, compared with CPU-based servers
- 100X in speed-up on large data set analysis, compared with a 20 node Spark server cluster
- 5X increase in bandwidth compared to PCIe with NVIDIA NVLink technology
- Maximised versatility with deep learning training and over 30,000 images/second inferencing