amd64 slower than i386 on identical AMD 64 system?
JoaoBR
joao at matik.com.br
Wed Mar 15 20:29:01 UTC 2006
On Wednesday 15 March 2006 15:46, Kris Kennaway wrote:
>
> OK. Just as a general point of common sense: when you find things
> that aren't enabled by default, there's almost always a good reason
> for this.
>
> Sometimes the reason is clear (driver conflicts with another driver,
> etc), but for secret poorly-documented options that change kernel
> behaviour it should ring warning bells that it might not be a good
> idea to just set and forget.
>
as well may exist good reason to check them out, if nobody checks this things
we never know about, right?
preemption and ipi_preemption are good theories (seems to me) and things like
that are worth to be tried even knowing that they are in early stage
anyway I am not blaming somebody nor the OS for anything instead I tried to
exchange experiences in first place.
>
> And above in the quoted text you said bad performance, so who knows
> what settings you had then :-) Maybe you're blaming something on
> polling that is really the fault of ULE.
>
ok but I am not that stupid to try a prototype scheduler and then blaming it
for my mess
>
> It really sounds like it could be a broken BIOS on your system (check
> for upgrades). AFAIK, dual-core systems are not known to have these
> problems in general.
>
well, that was my first thought too but makes no sense if the same happens on
several different brands, anyway I checked and tested different bios versions
on any board because the thing is important to me, I can not migrate 500
servers having this problem
also it is clearly X2 processor related since it doesn't appear when using a
normal athlon64, then you may say ACPI is to blame what could be the case of
Asus or Abit but the EPOX ACPI is very clean and appearently error free
My guess here is the shared memory access when running X2-SMP
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
More information about the freebsd-amd64
mailing list