GPU computing and big data

April 05, 2013 − by Suzanne Elvidge − in Data mining − No Comments

Getting the most out of huge sets of data takes a lot of processing power, and it’s only over the last couple of decades that computer technology has caught up with researchers’ needs. One of the breakthroughs in this area was the invention of the graphics processing unit or GPU.

The US-based company NVIDIA first invented the GPU in 1999, to process three-dimensional graphics. However, these chips have since been exploited for analyzing big data and processing complex algorithms, because they have up to 2048 processing cores that can process large quantities of information in parallel. Combining GPUs with a CPU (central processing unit) in GPU computing means that processor-intensive calculations can be ‘offloaded’ to the GPU to run in parallel, while the computer code runs on the CPU.

“GPU accelerators provide great value to applications with lots of data or computations,” said Sumit Gupta, general manager of the Tesla accelerated computing business at NVIDIA. “A growing number of applications… have both. And that’s prompting their providers to turn to GPU accelerators as they scale up their infrastructure to meet growing demand.”

This flexibility and power makes GPUs an obvious choice for processing and comparing large quantities of medical research data. This allows researchers to make connections between genetic information and clinical data in controls and people with a certain disease, for example, allowing them to mining for biomarkers. This could lead to development of vital diagnostics, or reveal targets for life-saving treatments.

By combining massively parallel processing powered by an array of GPUs with array-based logic, GenoKey can provide cost-effective, fast, and powerful data analysis, with algorithms working up to 200 times faster than on equivalently-priced CPUs. This technology means that datasets from tens or hundreds of thousands of patients can be analyzed for significant patterns within minutes or hours rather than days or weeks.

”We are excited about the performance and scalability of our data mining technology executed on GPUs and trust this will be even more important for case-control analysis of NGS data with millions of genetic and clinical biomarkers,” says Gert L Møller, chief technology officer at GenoKey.



Post a Comment

Your email address will not be published. Required fields are marked *