Re: powerd freeze with amd 5000 X2 but not with lower cpus

On Saturday 28 July 2007 20:21:02 Peter Jeremy wrote:
On 2007-Jul-28 19:03:54 -0300, JoaoBR <joao@xxxxxxxxxxxx> wrote:
so using ULE in 7 is ok ? ? ?

Yes.


I thought it is SCHED_SMP on 7 ... isn't it? if not forget my joke, I remember
a msg from Roberson saying the _SMP will substitute _ULE on 7 but I do not
use 7 so may be it changed later back to it's original name

ULE in 6.x is absolutely ok and it runs depending on situation faster than
4BSD with correct kernel and sysctl settings for it and it is perfectly
stable,

This is simply wrong. ULE in 6.x is known to have problems and is
unsupported. If the problems do not affect your particular workload
then fine. If you have _any_ problems whilst running with ULE in 6.x,
your problems will not be invstigated unless you can reproduce the
problem with the 4BSD scheduler.



interesting, what do you know? Do you have some data to share?

I don't know where you got your info but I have 50 X2 SMP amd64 running and
25 or so dual-opteron dualcores with SCHED_ULE absolute rockstable and faster
than 4BSD, as I mentioned under the circumstances I described before

I also have more 40 or so X2s and 60 or so dual and quad opteron dualcores
running 4BSD and before you tell me more blabla copied from newspapers and
other cha-cha sources better you come to me with data (DATA=numbers) from
*real* world

As Kris stated, reporting problems in 6.x when you are running ULE is
just wasting developer resources.

well I also don't know where you got this because I *never ever* claimed any
problem with ULE


Please stop implying that people should be using ULE in 6.x unless you
are willing to personally provide support for them.

I also do not know where you got this because I also *never ever* implied
using ULE, I simply say that I do *NOT* have any problem with it the way I
use it

So you please read the complete msgs and *try* to understand them before
answering with distortions and irrelevant conversations




--

João







A mensagem foi scaneada pelo sistema de e-mail e pode ser considerada segura.
Service fornecido pelo Datacenter Matik https://datacenter.matik.com.br
_______________________________________________
freebsd-stable@xxxxxxxxxxx mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "freebsd-stable-unsubscribe@xxxxxxxxxxx"

Relevant Pages

  • Re: cvs commit: src/sys/kern sched_ule.c
    ... Certainly if ULE were to implement ... >>What if a scheduler wants to keep a thread on TWO lists.. ... >>thread on BOTH teh percpu queue and an independent queue. ...
    (freebsd-current)
  • Re: cvs commit: src/sys/kern sched_ule.c
    ... >>ULE could implement fairness). ... >>implement fairness without any constraints of worrying about the BSD ... What if a scheduler wants to keep a thread on TWO lists.. ... thread on BOTH teh percpu queue and an independent queue. ...
    (freebsd-current)
  • Re: SCHEDULE and high load situations
    ... but I haven't tried it with Jeff's latest set of patches ... Since the run queue is a property of the scheduler, ... responsible for coming up with a notion of "load average". ... ULE, are we not getting enough work done, or both. ...
    (freebsd-current)
  • Re: ULE Scheduler available in 5.3-RELEASE?
    ... > "The ULE scheduler has been added as an additional scheduler. ... > This will not prevent problems due to overloaded disk subsystems, ... > queues which allow for CPU affinity, CPU binding, and advanced ...
    (freebsd-stable)
  • Re: cvs commit: src/sys/kern sched_ule.c (fwd)
    ... Jon Noack wrote: ... >>and this problem isn't exhibited with the ULE scheduler. ... However, Scott Long has ...
    (freebsd-current)