Re: Linux 2.6.15-rc5: multiuser scheduling trouble

Linus Torvalds wrote:

Helge,
did this start at any particular point in time?


Not that I know of, flash games became popular the last weeks.
I can try some older kernels though.

Also, the most common case is that somebody has reniced the X server, which is just _wrong_. 

I wish it was that simple.  But according to "top", the xservers runs
with NI=0.  I believe I got rid of the X renicing at the time the scheduler
was improved.  Nobody played flash games at the time though.

I have tried running two tuxracers too, and found this to be somewhat
bursty.  Instead of the players enjoying half a cpu each (which
is bearable at sufficiently low resolution), they seem to get bursts of cpu
and short pauses.  I never bothered looking more into this as  the
pci card always hangs the machine after a while when doing 3D. Instead
I disabled DRI on that xserver.  Stability is more important.

Helge Hafting
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Relevant Pages

  • Re: [PATCH] Staircase scheduler v7.4
    ... > You know, if renicing X makes it smoother, then that is a good thing ... X needs large amounts of CPU and low latency in order to get ... up with a set of rules for CPU allocation that makes sure the X server ... send the line "unsubscribe linux-kernel" in ...
    (Linux-Kernel)
  • Re: [PATCH] Staircase scheduler v7.4
    ... >> You know, if renicing X makes it smoother, then that is a good thing ... X needs large amounts of CPU and low latency in order to get ... > up with a set of rules for CPU allocation that makes sure the X server ... send the line "unsubscribe linux-kernel" in ...
    (Linux-Kernel)
  • Re: Patch to add support for SGIs IOC4 chipset
    ... by drivers/ide/Kconfig (and optionally by init time message). ... If I can get rid of some easily, ... (you don't have to update declaration if you change corresponding function). ... send the line "unsubscribe linux-kernel" in ...
    (Linux-Kernel)
  • Re: sluggish system responsiveness under higher IO load
    ... I made some tests with renicing the IO ... (I probably should have been more verbose in my suggestion;) ... cpu usage at interactive status, and keep your application off of the ...
    (Linux-Kernel)
  • Re: sched_yield behavior
    ... >> CPU, to give to other tasks a chance to run. ... >> higher priority task, and the only of its priority list). ... > the exact nature of the real time constraint? ... send the line "unsubscribe linux-kernel" in ...
    (Linux-Kernel)
Loading