Results 1 to 10 of 10

Thread: Speed Boost: The Linux Kernel Can Run On Zero CPU Cores

  1. #1
    Join Date
    Jan 2007
    Posts
    14,906

    Default Speed Boost: The Linux Kernel Can Run On Zero CPU Cores

    Phoronix: Speed Boost: The Linux Kernel Can Run On Zero CPU Cores

    A patch emerged out of IBM yesterday so that the Linux kernel can now has optimal SMP efficiency by being able to support running off zero CPU cores...

    http://www.phoronix.com/vr.php?view=MTA4MTc

  2. #2
    Join Date
    Jan 2012
    Posts
    3

    Default

    .. very late for april fools!

  3. #3
    Join Date
    Mar 2011
    Posts
    378

    Default

    //EDIT: Whoops, wrong thread, sorry.

  4. #4
    Join Date
    Sep 2008
    Posts
    989

    Default

    Some kernel developers are wanting to remove the kernel's "tick" altogether, and not even have dynticks -- in other words, there will be no wake-ups even when applications are awake.

    Will NR_CPUS=0 make this easier to achieve?

  5. #5
    Join Date
    Nov 2009
    Posts
    379

    Default

    Lol, fun read

    I like this one from Boris:
    Btw, I just got confirmation from hw folk that we can actually give you
    hardware support for that code with an upcoming CPU which has NR_CPUS=0
    cores.

    Oh, and additionally, we can disable some of those so getting into the
    negative is also doable from the hw perspective, so feel free to explore
    that side of the problem too.

  6. #6
    Join Date
    Dec 2010
    Location
    MA, USA
    Posts
    1,345

    Default

    i'm not sure i really understand how this works. i don't compile the kernel so maybe that's why I feel out of the loop.

  7. #7
    Join Date
    Nov 2009
    Posts
    379

    Default

    Quote Originally Posted by schmidtbag View Post
    i'm not sure i really understand how this works. i don't compile the kernel so maybe that's why I feel out of the loop.
    You know how threads work and the CPU time is distributed between processes?

    On a single core each running process get a little time every now and then, which process and how much decides the scheduler (e.g. BFS, CFS etc.). This happens for each core on multicore platforms, so two processes or more can run at the same time.

    Now the Linux kernel supports a certain number of CPUs (I think 4096?) and with this patch even 0 or less

  8. #8
    Join Date
    Sep 2007
    Location
    Connecticut,USA
    Posts
    962

    Default

    Leave it to one prankster at IBM's research lab to come up with this wacky patch

    Now if anyone dared to apply the patch...

  9. #9
    Join Date
    Dec 2010
    Location
    MA, USA
    Posts
    1,345

    Default

    Quote Originally Posted by disi View Post
    You know how threads work and the CPU time is distributed between processes?

    On a single core each running process get a little time every now and then, which process and how much decides the scheduler (e.g. BFS, CFS etc.). This happens for each core on multicore platforms, so two processes or more can run at the same time.

    Now the Linux kernel supports a certain number of CPUs (I think 4096?) and with this patch even 0 or less
    Yes, I did know that, but that doesn't explain the benefit of 0 cores. Is it basically saying that these tasks would never run?

  10. #10
    Join Date
    Nov 2009
    Posts
    379

    Default

    Quote Originally Posted by schmidtbag View Post
    Yes, I did know that, but that doesn't explain the benefit of 0 cores. Is it basically saying that these tasks would never run?
    Yes, the email was sent on 01.04.2012

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •