Access and Feeds

MapD and GPUs: Fast Data: Speeding up Compute-Bound Database Queries

By Dick Weisinger

Big Data sets are only possible if there is some Big Processing available to handle the enormous volumes.  While the amount of data being captured is rising exponentially hardware and software engineers are rushing to develop software algorithms and processing hardware that can keep up with the load.

Increasingly GPU architectures are being used to solve exceptionally complex problems.  The misnamed graphical processing unit (GPU) is a computer processor that is fine-tuned to process complex mathematical operations, like matrix mathematics and linear algebra very quickly.  These chips were initially used primarily to work on graphic and image processing problems, but more recently they are finding success when applied to a variety of  scientific, analytics, engineering, consumer, and enterprise applications.

For example, consider database technology.  Recently speed gains have been seen by database solutions that are designed to run completely in-memory.  SAP HANA and Spark are examples of this.  In-memory databases are great when the problems you work on are memory bound.   But there’s another possible approach.  The San Francisco-based startup MapD has applied the use of GPUs to database processing.  The MapD solution attempts to use GPUs to speed things up processing for problems that are compute-bound.  The speed gains that they’ve seen have been huge, sometimes as much as 100 times the performance of standard CPU-based databases.

James Curtis, 451 Group’s Senior Analyst, said that “MapD has an innovative high-performance offering that combines GPU technology with the firm’s own columnar database and visualization frontend, making MapD well suited to help organizations drive real-time decision making.”

Todd Mostak, founder of MapD, told the Wall Street Journal that “I would say we’re ‘fast data.’”

Digg This
Reddit This
Stumble Now!
Buzz This
Vote on DZone
Share on Facebook
Bookmark this on Delicious
Kick It on
Shout it
Share on LinkedIn
Bookmark this on Technorati
Post on Twitter
Google Buzz (aka. Google Reader)

Leave a Reply

Your email address will not be published.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