Cisco GPU-Accelerated Data Center

NVIDIA GPU Technology Conference, aka GTC, is around the corner and it is a good time to reflect on the rise of GPUs in the data center and what they make possible.

When I first attended GTC four years ago, machine learning and deep learning were barely on anybody’s radar outside academic circles and research lab; today these technologies are effectively deployed in many industries from healthcare, to banking, to autonomous vehicles and many more.

Four years ago, at our first GTC, Cisco had just a tabletop with black drapes and a PowerPoint presentation on a laptop. This year, we are a platinum sponsor showcasing our GPU-accelerated data center solution.

One pillar of the Cisco GPU-accelerated data center is support for virtual workstations running on Cisco UCS and Cisco HyperFlex systems. With support for the NVIDIA M6 Tesla card on our UCS blade servers, and M60 & M10 cards in our UCS rack servers and HyperFlex hyperconverged nodes, Cisco can deliver the infrastructure needed to run the most demanding graphics applications on virtualized workstations, effectively untethering engineers from their desks.

CannonDesign, an architectural and engineering firm, has successfully deployed NVIDIA M6 GPU in Cisco UCS blade servers to deliver virtual desktops to run their graphics-intensive applications. Catch their GTC session for more details. Mayfran International, a manufacturing company, is using Cisco hyperconverged system, HyperFlex, and NVIDIA GRID to speed up the design of its conveyor belts. By having their Autodesk Inventor files located in the data center next to their virtual desktops, they are able to load these files in about 4 minutes instead of 20.

The combination of Cisco UCS Integrated Infrastructure for big data and analytics with NVIDIA K80 or P100 cards provides the robust infrastructure needed to run real-time analytics. See Cisco and NVIDIA solution with Kinetica here.

Cisco is also working with deep learning and machine learning in our IoT practice.

Stop by our booth #311 and see for yourself. We will be demonstrating how the right infrastructure can support your virtual workstations as well as your Windows 10 virtual desktops. We will also show an application of deep learning/machine learning for IoT, with a spatial predictive analytics demo where we will analyze and count crowds in a live setting.

We will also be introducing our Cisco Spark Board for team collaboration, powered by NVIDIA Jetson TX1. Read the NVIDIA blog on Cisco’s use of the Jetson platform in the Spark Board.

Finally, you won’t want to miss Hugo Latapie’s session, The Future of IoT and intelligent Applications, on Thursday at 2:00 PM, when he will discuss Cisco’s foray into deep learning and machine learning and talk about the new capabilities and use cases Cisco is targeting with this breakthrough technology.