Loading recent posts...

Jun 25, 2012

AMD Fusion Server Brings 252% of MATLAB Performance Using OpenCL

AFDS 2012 event is proving to be very interesting and is clearly has much more serious impact on the computing industry as last year.

Now we even have AMD APU based  professional servers from Penguin Computing fited with proper software that will take advantage of the iGPU inside AMD’s Fusion APUs and offer considerable better results. In this case the demonstration is done on a Penguin Computing Altus 2A00 server that you can read full story in Penguin Computings' Website.

The system is powered by AMD’s Llano APUs, but we can expect Trinity versions to appear soon. In this particular case, Penguin uses Accelereyes’ software called “Jacket” that allows MATLAB to run on but the x86 core and the iGPU at the same time. The iGPU is able to deliver 52% better performance, but as the speaker emphasizes, the server is actually running two threads at the same time offering a combined performance of 252% the performance of just the x86 cores.

The user will be able to run two different threads in MATLAB at the same time. One will use the x86 cores and the other will use the OpenCL capabilities of the iGPU with the help of Accelereyes’ “Jacket” software. The demo is impressive as practically the productivity of the system is increased almost three times while there is a single APU inside.

AMD's Sasa Marinkovic and AccelerEyes' John Melonakos demo Matlab on OpenCL and the AMD Trinity APU, from the Experience Zone at AFDS 2012. 

For more information, visit: http://bit.ly/AFDS-D and http://www.accelereyes.com/


George McKinney said...

Jacket is awesome and has really done well for our group in accelerating some radar processing codes. Glad to see it running on the APUs.

Post a Comment

Twitter Delicious Facebook Digg Stumbleupon Favorites More

Design by Free WordPress Themes | Bloggerized by Lasantha - Premium Blogger Themes | coupon codes