Join The GPU-Fueled Analytics Revolution
MapD CEO Todd Mostak recently completed a European roadshow with Jim McHugh, NVIDIA vice president and general manager of AI Systems. They traveled to Munich, London and Paris to talk about the GPU-accelerated analytics ecosystem.
There was a lot of excitement around GPU analytics — especially with the increased adoption of AI and as data volumes grow. Here are the key takeaways:
Jim McHugh of NVIDIA noted that we’re entering a new “industrial revolution” that has been building through the eras of the PC, mobile and the cloud. While combining mobile with the cloud allows compute accessibility anywhere, it’s the combination of GPUs and deep learning that will produce truly game-changing insights.
“This opened up the modern world of AI,” McHugh said. “And it is going to spur so much progress that it will become the next industrial revolution.”
Life After Moore’s Law
Slow computing power has been the roadblock to data exploration. The long response times start to restrain the questions people ask.
A reversal of Moore’s Law was the last straw for CPU-based analytics as the 50 percent annual gains began to decline by 10 percent per year. This ushered in the era of GPU computing and accelerated analytics.
“We realized if we could offload what is really a parallel process from the CPU and put it to the GPU, we could see eye-opening acceleration of applications,” McHugh said. “What used to be talked about as real-time is now measured in sub-seconds — the kind of time we need to have game-changing solutions.”
The Promise of Accelerated Analytics
Now, any industry from advertising to financial services to telecom can benefit from accelerated analytics. McHugh shared examples like exploring New York City taxi data, seeing where items are going in the U.S. Postal Service and understanding cyber security efforts to track and prevent attacks
The partnership of MapD and NVIDIA fills the need for the accelerated workloads that can lead to more innovation. MapD and NVIDIA allow for a visualization of data and correlations that never existed before.
“Every part of the data migration, every part of the pipeline is being accelerated by solutions running on the GPU,” McHugh said. “We’re bringing the dreams of science fiction true.”