GPU Startup Story: Fuzzy Logix Brings Clarity to Analytics

by Gary Rainville

[Editor’s note: Nearly three dozen companies participated in the Emerging Companies Summit, held during NVIDIA’s GPU Technology Conference earlier this year. Below is one in a series of company profiles showcasing how startups are innovating with GPU technology.]

For many enterprises, Big Data can mean big confusion.

Fuzzy Logix, a North Carolina-based startup, is helping them make better sense of it and predict the future by deploying GPU-based technology.

Imagine a brokerage that has only milliseconds to decide whether a large investor’s big trade will be within margin requirements. Not just the requirements at the moment of the trade, but accounting for all projected trades throughout the rest of the day.

Or a call center determining which rep should field an incoming call based on the demographics of the caller’s location, past interactions and who will provide the best customer experience. And doing all this while also providing suggestions for the next likely purchase – before the second ring.

From marketing and finance to sales and customer service, there are literally thousands of applications for predictive analytics in enterprises. Fuzzy Logix is focused on making the technology adoptable and pervasive.Fuzzy Logix logo

“We do this,” says Michael Upchurch, chief operating officer at Fuzzy Logix, “by moving analytics out of the hands of specialists and specialized applications and making them broadly available to business decision makers.”

And they make it lightning-fast by using GPUs. “Working with our clients, we are consistently identifying general business challenges that can be best solved by leveraging GPU-based analytics,” says Upchurch.

To date, the company has ported the code base of more than 500 in-database analytics models to run in GPUs. With these essential building blocks at the ready, every Java, C/C++, .NET or other programmer in the world can build statistical models faster and spend their effort writing higher order models.

Since the models can be run using common programming languages, they are easily deployed to model runners, such as marketing and finance executives and salespeople in the field. These non-statisticians can run analytics on demand to meet their immediate needs, such as forecasting sales, optimizing bid prices for online ads and projecting customer turnover.

Fuzzy Logix also offers a deskside or rack-mountable appliance that taps into an enterprise’s computing infrastructure to grab a data file and then processes the information locally using GPUs.

The benefits to this approach are dramatic: it reduces the queues that occur when using CPU-based systems, lessens the demands on model builders and lets end-users make timely decisions because they can get answers they need fast.

How fast? The Fuzzy Logix appliance with four NVIDIA Tesla cards is able to perform 1 billion calculations in 13 milliseconds. This is speed a CPU-only system simply can’t match, performed at a fraction of the cost.

By providing this enormous processing power to businesses, and giving them control over an easily adopted analytics environment, Fuzzy Logix is making desktop supercomputing real.