United Business Media EE Times


Search

HOMEMARKET INTELLIGENCE UNITFORUMSDESIGNNEW PRODUCTSCAREERSBLOGSCONTACTEVENTSSIGN UP!RSSMost Popular contentTrusted Sources

 

Multicore CPUs face slow road in comms
The culprit: complex, fragmented technology
Print this article Email this article Reprints RSS Digital Edition

EE Times


SANTA CLARA, Calif. — The transition to multicore processors in communications and networking systems is expected to be a slow one due to complex and fragmented nature of the underlying technology, predicted a technology analyst.

Processors with four or more cores will probably represent little more than 10 percent of the communications systems market in 2012, according to Linley Gwennap, principal analyst with The Linley Group (Mountain View, Calif.). He was speaking at a panel at this week's Multicore Association Expo here.

By contrast the use of single-core processors is still on the rise in embedded systems, peaking at about half the market over the period. The PowerQuicc, a unique heterogeneous architecture from Freescale Semiconductor that represented another large swath of the market, is on the decline as the company transitions to a simpler dual-core architecture, he said.

Gwennap projected that such dual-core designs could command as much as 20 percent of the market by 2012.


Click on image to enlarge.

"I think the deployment of multicore hardware has been and will be slower than people may have hoped, but there's a definite advantage to companies who can help us get over these barriers," Gwennap said.

The barriers are high, especially in software, he said.

Most of today's embedded software is written for single-core processors. Operating systems for symmetric multiprocessing architectures do not scale well beyond four cores. Asymmetric architectures require sophisticated load balancing, and hardware accelerators require complex mixtures of diverse APIs.

On the hardware side, multicore processors face several barriers to scaling their performance. Several elements can quickly become performance bottlenecks including central buses, single shared caches and single DRAM interfaces. In addition, multicore processors can also face costly cache misses due to the limited number of DRAM pages that can be open at a given moment.

"It will take a long time for some of today's multicore designs to work through the system," he said.

On the software front, chip vendors often find themselves doing some of the multicore programming for their systems customers, said Dave Lapp, a senior system architect at Freescale. "In general, customers don't want to learn [a new multicore] tool chain," he said..



Related Links:

  • Online course: Intro to multicore
  • Tool puts parallelism in embedded code
  • White paper: Multicore transition in comms and embedded



  •   Free Subscription to EE Times
    First Name Last Name
    Company Name Title
    Email address
      Click here for your Free Subscription to EETimes Europe
     
    CAREER CENTER
    Looking for a new job?
    SEARCH JOBS
    SPONSOR

    RECENT JOB POSTINGS
    CAREER NEWS
    DoD Recognizes University Scientists For Basic Research
    Annual awards to university faculty to conduct next-generation research projects were announced this week by the Defense Department.

    For more great jobs, career related news, features and services, please visit EETimes' Career Center.



    All White Papers »   

     
    Education and
    Learning


    Learn Now:












    Home | About | Editorial Calendar | Feedback | Subscriptions | Newsletter | Media Kit | Contact | Reprints|  RSS|   Digital|  Mobile
    Network Websites
    International
    Network Features




    All materials on this site Copyright © 2010 TechInsights, a Division of United Business Media LLC All rights reserved.
    Privacy Statement | Terms of Service | About