Enabling AI-Powered Innovation from Data Center to Deskside with DGX Station

by Tony Paikeday

Business processes that follow workers, instead of the other way around. Customer service operations that predict churn before a client is lost to a competitor. Brand marketing investments that yield measureable ROI. This is what the digital enterprise, or more correctly the AI enterprise, looks like – designed by you, and fueled by an insatiable appetite for your data.

We’re helping today’s enterprise GPU-accelerate every aspect of their R&D, supply chain and marketing operations to make these capabilities a reality. But to get there, an organization needs to think about AI and machine learning in a way that goes beyond the confines of the data center, and puts supercomputing power, literally the computational capacity of hundreds of CPUs, in the hands of every researcher, product developer, business process engineer and data scientist – namely, every innovator.

The challenge quickly becomes one of cost and scale. Our original DGX-1 based on Pascal GPUs packed 170 TFLOPs in a single chassis, delivering the equivalent of 250 x86 servers. Our latest DGX-1 based on Volta GPUs raises the bar to almost a petaflop (960 TFLOPs to be precise). As a device installed in the data center, it enables organizations to reap the benefits of their own AI-cloud. But before you can make the leap to machine learning at such scale, the developers in your enterprise need fast, unimpeded access to GPU computing when and where inspiration hits them. Extending the benefits of a DGX-1 to each and every innovator across the organization may not always be practical.

This means “re-thinking” a few things, including:

  • extending that kind of power to within arm’s reach
  • bringing the tools to where the data resides
  • deploying the platform where it seamlessly integrates within their existing workstyle and workflow
  • making it cost-effective enough to be offered to every innovator in the enterprise

So, what if we designed a DGX-1 that could:

  • sit neatly underneath your employee’s desk
  • was whisper-quiet and simply plugged into a standard wall outlet
  • effortlessly ingest mountains of data that until now was insurmountable
  • deliver the price-to-performance ratio to make it tenable for your valued innovators

The Personal AI Supercomputer: The Newest Addition to Your Innovator “Welcome Kit”

The “what-if’s” above essentially describe the NVIDIA DGX Station. A desk-side supercomputer that can be

DGX Station delivers the power of 400 CPU’s

cost-effectively used by every person in your organization responsible for innovating your processes and products. It’s changing the thinking around what the standard “loadout” (in gaming parlance) should be for the people tasked with solving your company’s biggest problems.

We can’t imagine today’s knowledge worker without the aid of a laptop, smartphone, and broadband internet, and similarly, it’s going to be very hard to imagine your prized innovators working without the advantage of a personal AI supercomputer. With this essential tool, they’re going to help you digitize your business, transform customer relationships, and unlock new revenue frontiers. Check out our new DGX Station Infographic to learn more.

Enabling the Intelligent Enterprise with SAP and NVIDIA

NVIDIA and SAP are helping businesses realize this transformative effect of AI-powered enterprise applications. With 76 percent of the world’s transaction revenue running through SAP systems, SAP is in a unique position to help organizations arrive at the crossroads of AI-powered innovation and business transformation. Their Innovation Center Network is helping businesses everywhere re-imagine their operations from Marketing to Design to Logistics and more, now powered by NVIDIA DGX Systems.

In our recent webinar with Frank Wu – SAP Director of Machine Learning at the SAP Innovation Center Network, we heard first-hand how DGX Station yielded a 170x performance improvement in their deep learning training environment. As Frank shared, the streamlined deep learning workflow and training performance boost is helping SAP create new AI-powered business applications such as:

  • service ticketing that uses deep learning to intelligently route customer help requests to the right agent for faster resolution.
  • SAP Brand Impact enabling companies to measure ROI impact of brand marketing investments

Additionally, Frank shared key benefits of DGX Station that appealed to his team at the SAP Innovation Center Network, including:

  • the dramatic acceleration of neural net training, moving from days between training to near real-time iteration
  • the simplified experience of leveraging NVIDIA Docker, along with the ease of deploying and using NVIDIA-optimized frameworks like TensorFlow, Theano, and PyTorch
  • the ability to support multiple teams accessing the DGX remotely
  • the quiet, office-optimized user experience of DGX Station

To learn more about how SAP is helping organizations transform their business using NVIDIA DGX Station check out the webinar replay.