The presentation below is an educational resource that sets the stage for parallel programming with GPUs (graphics processing units) and was sponsored by the Center for Astrophysics and Supercomputing at Swinburne University of Technology. GPUs are becoming quite popular for the implementation of deep learning solutions.
I was very pleased to attend the GPU Technology Conference 2016 as the guest of host company NVIDIA on April 4-7 in Silicon Valley. I was impressed enough with the experience that I wanted to write this Field Report to give readers an in-depth perspective for what I saw.
MapD Technologies unveiled its GPU-powered database and visual analytics software platform that enables data analysts to interactively explore large data sets at high speed.
As the use of GPUs continues to rise in fields like deep learning, we thought it would be useful to readers not yet familiar with this technology to offer the “Introduction to GPU Computing” presentation below.
“NVIDIA will present an update on accelerated computing, in particular, the latest de- velopments in the platform. They will touch upon NVLink, OpenPOWER, ARM64, and new software updates and also cover the broad-sweeping impact that a new field of machine learning, called Deep Learning, is having on applications and domains.”
In this video from GTC 2014, Todd Mostak from MapD demonstrates the company’s GPU-powered in-memory relational database software for Big Data. The Cambridge, Mass., based startup has built a high-speed GPU in-memory database that brings interactivity to big data. It can, for example, track more than a billion tweets worldwide at a time – and provide real-time visual analysis of the data. MapD was also announced as the winner of the GPU Technology Conference’s Early Stage Challenge this year, and they will be coming home with a cool $100,000 check.
In this video from SC13, Thomas Graham and Todd Mostak from map-D demonstrate their super-fast GPU database technology. “We build unique databases that process huge datasets in milliseconds, performing up to 70 times faster than CPU-based solutions.”