pg.ddx.io pgsql-performance@postgresql.org mailing list archive
help / color / mirror / Atom feedRe: Adding more memory = hugh cpu load [solved]
6+ messages / 3 participants
[nested] [flat]
* Re: Adding more memory = hugh cpu load [solved]
@ 2011-10-11 18:02 alexandre - aldeia digital <adaldeia@gmail.com>
2011-10-11 18:05 ` Re: Adding more memory = hugh cpu load [solved] Claudio Freire <klaussfreire@gmail.com>
0 siblings, 1 reply; 6+ messages in thread
From: alexandre - aldeia digital @ 2011-10-11 18:02 UTC (permalink / raw)
To: pgsql-performance
Hi,
About 3 hours ago, the client contacted the Dell and they suggested 2
things:
1) Update the baseboard firmware (the only component that haven't
updated yesterday).
2) Change all memory chips to new others, instead of maintain the old
(16 GB) + new (32 GB).
After do this, until now, the load average does not surpasses 4 points.
The I/O wait returned to hit some values and the system % down to an
average of 5%.
Unfortunatly, we will never discover if the problem resides in baseboard
firmware or in RAM, but util now the problem was solved.
Thanks all for help !
Best regards
^ permalink raw reply [nested|flat] 6+ messages in thread
* Re: Adding more memory = hugh cpu load [solved]
2011-10-11 18:02 Re: Adding more memory = hugh cpu load [solved] alexandre - aldeia digital <adaldeia@gmail.com>
@ 2011-10-11 18:05 ` Claudio Freire <klaussfreire@gmail.com>
2011-10-11 20:02 ` Re: Adding more memory = hugh cpu load [solved] alexandre - aldeia digital <adaldeia@gmail.com>
2011-10-11 20:33 ` Re: Adding more memory = hugh cpu load [solved] Arjen van der Meijden <acmmailing@tweakers.net>
0 siblings, 2 replies; 6+ messages in thread
From: Claudio Freire @ 2011-10-11 18:05 UTC (permalink / raw)
To: alexandre - aldeia digital <adaldeia@gmail.com>; +Cc: pgsql-performance
On Tue, Oct 11, 2011 at 3:02 PM, alexandre - aldeia digital
<adaldeia@gmail.com> wrote:
> 2) Change all memory chips to new others, instead of maintain the old (16
> GB) + new (32 GB).
Of course, mixing disables double/triple/whatuple channel, and makes
your memory subsystem correspondingly slower.
By a lot.
^ permalink raw reply [nested|flat] 6+ messages in thread
* Re: Adding more memory = hugh cpu load [solved]
2011-10-11 18:02 Re: Adding more memory = hugh cpu load [solved] alexandre - aldeia digital <adaldeia@gmail.com>
2011-10-11 18:05 ` Re: Adding more memory = hugh cpu load [solved] Claudio Freire <klaussfreire@gmail.com>
@ 2011-10-11 20:02 ` alexandre - aldeia digital <adaldeia@gmail.com>
2011-10-11 20:08 ` Re: Adding more memory = hugh cpu load [solved] Claudio Freire <klaussfreire@gmail.com>
1 sibling, 1 reply; 6+ messages in thread
From: alexandre - aldeia digital @ 2011-10-11 20:02 UTC (permalink / raw)
To: ; +Cc: pgsql-performance
Em 11-10-2011 15:05, Claudio Freire escreveu:
> On Tue, Oct 11, 2011 at 3:02 PM, alexandre - aldeia digital
> <adaldeia@gmail.com> wrote:
>> 2) Change all memory chips to new others, instead of maintain the old (16
>> GB) + new (32 GB).
>
> Of course, mixing disables double/triple/whatuple channel, and makes
> your memory subsystem correspondingly slower.
> By a lot.
The initial change (add more memory) are maded by a technical person of
Dell and him told us that he use the same especification in memory chips.
But, you know how "it works"... ;)
^ permalink raw reply [nested|flat] 6+ messages in thread
* Re: Adding more memory = hugh cpu load [solved]
2011-10-11 18:02 Re: Adding more memory = hugh cpu load [solved] alexandre - aldeia digital <adaldeia@gmail.com>
2011-10-11 18:05 ` Re: Adding more memory = hugh cpu load [solved] Claudio Freire <klaussfreire@gmail.com>
2011-10-11 20:02 ` Re: Adding more memory = hugh cpu load [solved] alexandre - aldeia digital <adaldeia@gmail.com>
@ 2011-10-11 20:08 ` Claudio Freire <klaussfreire@gmail.com>
0 siblings, 0 replies; 6+ messages in thread
From: Claudio Freire @ 2011-10-11 20:08 UTC (permalink / raw)
To: alexandre - aldeia digital <adaldeia@gmail.com>; +Cc: pgsql-performance
On Tue, Oct 11, 2011 at 5:02 PM, alexandre - aldeia digital
<adaldeia@gmail.com> wrote:
> The initial change (add more memory) are maded by a technical person of Dell
> and him told us that he use the same especification in memory chips.
> But, you know how "it works"... ;)
Yeah, but different size == different specs
I do know how it works ;-)
^ permalink raw reply [nested|flat] 6+ messages in thread
* Re: Adding more memory = hugh cpu load [solved]
2011-10-11 18:02 Re: Adding more memory = hugh cpu load [solved] alexandre - aldeia digital <adaldeia@gmail.com>
2011-10-11 18:05 ` Re: Adding more memory = hugh cpu load [solved] Claudio Freire <klaussfreire@gmail.com>
@ 2011-10-11 20:33 ` Arjen van der Meijden <acmmailing@tweakers.net>
2011-10-11 20:46 ` Re: Adding more memory = hugh cpu load [solved] Claudio Freire <klaussfreire@gmail.com>
1 sibling, 1 reply; 6+ messages in thread
From: Arjen van der Meijden @ 2011-10-11 20:33 UTC (permalink / raw)
To: pgsql-performance
On 11-10-2011 20:05 Claudio Freire wrote:
> On Tue, Oct 11, 2011 at 3:02 PM, alexandre - aldeia digital
> <adaldeia@gmail.com> wrote:
>> 2) Change all memory chips to new others, instead of maintain the old (16
>> GB) + new (32 GB).
>
> Of course, mixing disables double/triple/whatuple channel, and makes
> your memory subsystem correspondingly slower.
> By a lot.
>
That really depends on the chipset/server. The current intel E56xx-chips
(and previous E55xx) basically just expect groups of 3 modules per
processor, but it doesn't really matter whether that's 3x2+3x4 or 6x4 in
terms of performance (unless the linuxkernel does some weirdness of
course). It at least won't disable triple-channel, just because you
added different size modules. Only when you get to too many 'ranks',
you'll see performance degradation. But that's in terms of clock speed,
not in disabling triple channel.
But as said, that all depends on the memory controller in the server's
mainboard or processors.
Best regards,
Arjen
^ permalink raw reply [nested|flat] 6+ messages in thread
* Re: Adding more memory = hugh cpu load [solved]
2011-10-11 18:02 Re: Adding more memory = hugh cpu load [solved] alexandre - aldeia digital <adaldeia@gmail.com>
2011-10-11 18:05 ` Re: Adding more memory = hugh cpu load [solved] Claudio Freire <klaussfreire@gmail.com>
2011-10-11 20:33 ` Re: Adding more memory = hugh cpu load [solved] Arjen van der Meijden <acmmailing@tweakers.net>
@ 2011-10-11 20:46 ` Claudio Freire <klaussfreire@gmail.com>
0 siblings, 0 replies; 6+ messages in thread
From: Claudio Freire @ 2011-10-11 20:46 UTC (permalink / raw)
To: Arjen van der Meijden <acmmailing@tweakers.net>; +Cc: pgsql-performance
On Tue, Oct 11, 2011 at 5:33 PM, Arjen van der Meijden
<acmmailing@tweakers.net> wrote:
> That really depends on the chipset/server. The current intel E56xx-chips
> (and previous E55xx) basically just expect groups of 3 modules per
> processor, but it doesn't really matter whether that's 3x2+3x4 or 6x4 in
> terms of performance (unless the linuxkernel does some weirdness of course).
> It at least won't disable triple-channel, just because you added different
> size modules. Only when you get to too many 'ranks', you'll see performance
> degradation. But that's in terms of clock speed, not in disabling triple
> channel.
>
There are too many caveats to doing that, including how they're
installed (in which slots), the timings, brand, internal structure,
and whatnot, with all those details not always published, only sitting
in the module's SPD.
In essence, mixing is more probable to mess your multichannel
configuration than not.
That's why matched kits are sold.
^ permalink raw reply [nested|flat] 6+ messages in thread
end of thread, other threads:[~2011-10-11 20:46 UTC | newest]
Thread overview: 6+ messages (download: mbox mbox.gz follow: Atom feed)
-- links below jump to the message on this page --
2011-10-11 18:02 Re: Adding more memory = hugh cpu load [solved] alexandre - aldeia digital <adaldeia@gmail.com>
2011-10-11 18:05 ` Claudio Freire <klaussfreire@gmail.com>
2011-10-11 20:02 ` alexandre - aldeia digital <adaldeia@gmail.com>
2011-10-11 20:08 ` Claudio Freire <klaussfreire@gmail.com>
2011-10-11 20:33 ` Arjen van der Meijden <acmmailing@tweakers.net>
2011-10-11 20:46 ` Claudio Freire <klaussfreire@gmail.com>
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox