Nvidia GPUs processing Big Data

The term GPU (Graphics Processing Unit) was popularized by Nvidia in 1999. In the GPU Technology Conference which was held last week in San Jose, Nvidia shared how GPUs are used to tackle big data analytics and advanced search for both for commercial and consumer applications. Top companies like Shazam, Salesforce.com and Cortexica use Nvidia Tesla GPUs to tackle big data analytics. These companies are expanding the use of GPUs beyond their role of processing massive data sets and complex algorithms for high performance computing and engineering applications. 

Shazam software can be installed on a mobile phone or a tablet and can be used to search and identify songs from its 27 million and growing track database. With more than 300 million users making more than 10 million song searches a day, 2 million new users joining the service every week, its database has been doubled over the last year. Tesla GPUs have been used to accelerate the search and matching process. These GPUs enable to maintain a low-cost server infrastructure. 

Salesforce.com - ranked as the most innovative company in America by Forbes helps major brands such as Cisco and Dell to monitor and analyze more than 500 million daily tweets for brand, product, service and support issues. Nvidia CUDA (Compute Unified Device Architecture) GPUs enable Salesforce.com to deliver insights to these companies by analyzing the keywords on how their products are being perceived by the customers almost 35 times faster than a comparable CPU-based system. 

Cortexica's mobile application is similar to Google Goggles but is restricted to products which can be searched and purchased online. This application makes it easy for customers to find and purchase goods they like. Consumers can photograph products using their mobile phone or tablet, and the app searches through an online database of apparel to instantly find similar matches (not exact match) available for purchase from online retailers. 

With GPU accelerators, Cortexica runs complex visual object recognition algorithms on a modest amount of server hardware to enable real-time searches against a database of millions of images.

"GPUs are fundamental to our success. GPU accelerators run our complex bio-inspired algorithms 30 times faster than CPUs, giving our users the best, fastest and most accurate consumer experience," said Iain McCready, chief executive officer at Cortexica.

Reference: http://www.youtube.com/watch?v=5uzvx1bAUFo