A lot can happen in a year …
- a scientist discovers how to predict growth problems in children, with the accuracy of a radiologist
- an engineer upgrades the brain of their autonomous vehicle program
- a professor finds a way to improve baseball player performance and predict likelihood of injury
- a supercomputer breaks world records and powers the search for a cure to cancer
These are just some of the highlights from the 12 months since we announced the NVIDIA DGX-1 AI supercomputer. If you haven’t been along for the ride, check out how we brought this system to market as well as this retrospective on our first year.
Since last April, organizations across the globe have embraced the solution, making it a cornerstone of the effort to “AI-Accelerate” their business or field of scientific research. Here’s a bird’s-eye-view of what a year in the life of DGX-1 looks like:
If you’re wondering what compels a data science practice to go with a DGX-1, instead of building from the ground up, check out this blog post on a whitepaper that explores the innovations, architecture and performance advantages of DGX-1. (Read the whitepaper itself by clicking on the image below.)
Some of its key realizations:
- More than just its hardware, DGX-1 delivers innovative software, inspired by the demands of deep learning and data science. This integrated system redefines how practitioners can develop, tune and scale neural nets with the only stack that includes deep learning frameworks, libraries and drivers, all engineered by NVIDIA for maximized, GPU-optimized performance found nowhere else.
- DGX-1 delivers many industry “firsts,” including being the first solution built on eight NVIDIA Tesla P100 GPUs, connected via NVIDIA NVLink, configured in a hybrid cube-mesh topology, with multisystem strong-scaling built on Infiniband EDR IB.
- DGX-1 enables teams to rethink and simplify how they experiment, optimize and collaborate across organizations. With NVIDIA Docker-based container technology, practitioners can experiment on multiple frameworks, iterate on configurations in parallel non-disruptively, share their work with peers and push production-ready models to any node or cluster of nodes they have access to.
DGX-1 a Full Year Old
You may be wondering why there’s no picture of a DGX-shaped cake with ice cream. Parents give their offspring daily accolades, but we’ll let others do the talking. Here’s a sampling of what the world has had to say about our exceptional one-year-old:
- The DGX-1 will enable new levels of deep learning for AI applications. – Forbes
- This is AI in a box. – ITBusinessEdge
- Move over Watson – if you haven’t already – TechRadar
- NVIDIA’s DGX-1 supercomputer packs the power of 250 servers. – Computerworld
- NVIDIA’s insane DGX-1 is a computer tailor-made for deep learning. – Engadget
- The DGX-1 is designed to help researchers and scientists comprehend and analyze huge reams of data, as well as unlock the potential of artificial intelligence technology. – InformationWeek
- The Pint-Sized Supercomputer That Companies are Scrambling to Get – MIT Technology Review
- It is hands-down the most powerful system NVIDIA has ever brought to market – Gizmodo
And we’re just getting started. We’re only five weeks away from GTC 2017. This year’s event is the place to be for all things AI and deep learning, and is the perfect backdrop for exciting DGX news. If you don’t want to miss any of it, sign up (if you haven’t already) and check out these DGX-focused, must-attend sessions.
For more info on DGX-1, visit www.nvidia.com/dgx-1.