Information Technology News.


Coordinating core-to-core communications in hardware rather than software

Share on Twitter.

Sponsered ad: Get a Linux Enterprise server with 92 Gigs of RAM, 16 CPUs and 8 TB of storage at our liquidation sale. Only one left in stock.

Sponsered ad: Order the best SMTP service for your business. Guaranteed or your money back.

September 7, 2016

Working side-by-side with Intel, scientists at the North Carolina State University have designed a new framework that coordinates core-to-core communications in the hardware side of things rather than software.

The work the team is doing is important because to keep processing power scaling up, you want the overhead of multi-core computing processes to stay lower, something that's been a problem for quite some time.

Dubbed CAF (Core-to-core Acceleration Framework), the new implementation will be presented at the ACM's 2016 International Conference on Parallel Architectures and Compilation later this month in Israel.

The authors of the project are working on solving the various issues that arise using shared memory for core-to-core communications, cache misses, and a general loss of coherence throughout previous generations of the original Intel design.

To improve things in a significant manner, the researchers put together a queue management device (QMD) which gets closely integrated with the on-chip network.

The QMD's main job is to track communication requests between various cores without needing any queue management software, something that has never been done in the past.

Different workloads place various loads on the queues. In their abstract, the university researchers highlight 'pipelined packet processing' (PPP) as the kind of workload that suffers significant core-to-core communication overhead.

The scientists claim between 2 and 12 times better and faster performance improvements with their new QMD implementation.

While its overall processing power is limited to a certain degree, the team say the QMD has enough power to aggregate data from multiple cores which helps more basic computational functions by as much as 15 percent.

Source: North Carolina State University.

Sponsered ad: Get a Linux Enterprise server with 92 Gigs of RAM, 16 CPUs and 8 TB of storage at our liquidation sale. Only one left in stock.

Sponsered ad: Order the best SMTP service for your business. Guaranteed or your money back.

Share on Twitter.

IT News Archives | Site Search | Advertise on IT Direction | Contact | Home

All logos, trade marks or service marks on this site are the property of their respective owners.

Sponsored by Sure Mail™, Avantex and
by Montreal Server Colocation.

       © IT Direction. All rights reserved.