Re: ULE vs. 4BSD in RELENG_7

Your tests with ffmpeg threads vs processes probably is triggering more
context switches due to lock contention in the kernel in the threads case.
This is also likely the problem with some super-smack tests. On each
context switch 4BSD has an opportunity to perfectly balance the CPUs. ULE
does not because it's too costly and hinders other workloads.

Thanks for the response Jeff, I appreciate it. Is this something that
the scheduler can recognize and auto-tune itself for? E.g. have it
recognize the scenario differences and do what 4BSD is doing for cases
such as ffmpeg and what ULE otherwise does in other circumstances? I
guess this would not come without its own overhead, which might defeat
the purpose anyway.

I don't doubt that we can improve things further. It will just have to
wait for another few weeks before I'm able to do much about it.

Thanks again, I do appreciate your work and help. So you're
anticipating nothing can be done at all, or did you mean at the moment
until you get your equipment back and get settled? Or is this
something that would need to be looked at in -current rather than
7-STABLE?

Regards,
Josh
_______________________________________________
freebsd-performance@xxxxxxxxxxx mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-performance
To unsubscribe, send any mail to "freebsd-performance-unsubscribe@xxxxxxxxxxx"

Relevant Pages

  • Re: Is this code MT-Safe?
    ... > (context switches are still possible when a critical section is entered) ... > a single process. ... > kernel call will be made and the thread can continue. ... > contention (another thread has already entered the same critical ...
    (comp.lang.cpp)
  • Re: ULE vs. 4BSD in RELENG_7
    ... context switches due to lock contention in the kernel in the threads case. ... such as ffmpeg and what ULE otherwise does in other circumstances? ... The higher idle time with ULE is a symptom of assinging work to the wrong place. ... I think I have a good idea of how to improve that cpu selection. ...
    (freebsd-performance)
  • Re: socket programming using MSG_WAITALL for receiving multicast packets
    ... kernel and user space. ... the scheduler agrees with you. ... Right, and that's exactly what the kernel will do, all by itself. ... If you test with no load, you will get lots of context switches. ...
    (comp.unix.programmer)
  • Re: Security issues with local filesystem caching
    ... put the policy out of the kernel. ... The additional context switches ... It costs a lot in latency, and you don't need a daemon in the first ...
    (Linux-Kernel)
  • Re: Floating point in kernel mode, Win2003 64-bit
    ... I guess there must be *no* difference between kernel and user mode threads, ... and the *whole* FPU must be saved during context switches. ... FPU state saving based on PSR.dfh/mfh, if the OS supports this feature. ...
    (microsoft.public.development.device.drivers)