Subject | Odp: [firebird-support] firebird performance on xeon cpu |
---|---|
Author | liviuslivius@poczta.onet.pl |
Post date | 2013-08-28T05:15:48Z |
Hi,
This is cache problem of win64 not firebird itself. i do not remember link but try with google firebird eat whole memory or somethink like this.
Regards,
Karol Bieniaszewski
----- Reply message -----
Od: "Alexis Diel" <alexis_diel@...>
Do: <firebird-support@yahoogroups.com>
Temat: [firebird-support] firebird performance on xeon cpu
Data: śr., sie 28, 2013 04:50
I'm actually having this same problem with one of my clients. He have two
servers with x64 Windows and 8Gb of RAM. One of them is xp, the other is
seven. I tested Firebird 2.5.2 x32 and x64... change various configuration
and no success yet to reduce the consume of memory... that after a while
get between 80 and 90% and reduce the performance of the system
Em 27/08/2013 22:25, "安占江" <anzhanjiang@...> escreveu:
[Non-text portions of this message have been removed]
This is cache problem of win64 not firebird itself. i do not remember link but try with google firebird eat whole memory or somethink like this.
Regards,
Karol Bieniaszewski
----- Reply message -----
Od: "Alexis Diel" <alexis_diel@...>
Do: <firebird-support@yahoogroups.com>
Temat: [firebird-support] firebird performance on xeon cpu
Data: śr., sie 28, 2013 04:50
I'm actually having this same problem with one of my clients. He have two
servers with x64 Windows and 8Gb of RAM. One of them is xp, the other is
seven. I tested Firebird 2.5.2 x32 and x64... change various configuration
and no success yet to reduce the consume of memory... that after a while
get between 80 and 90% and reduce the performance of the system
Em 27/08/2013 22:25, "安占江" <anzhanjiang@...> escreveu:
> **[Non-text portions of this message have been removed]
>
>
> 于 2013年08月27日 23:53, Leyne, Sean 写道:
> >
> >
> >
> > > I have resolved this question.
> > > on disk driver property labels, I checked the close window cache on
> > disk,
> > > which is unchecked by default.
> > > but I don't known the reason.
> >
> > Actually, changing that setting is VERY DANGEROUS, it could lead to
> > database corruption!
> >
> > The setting tells the OS to assume that the device has a protected
> > cache (write-back), that will ensure that all writes will be persisted
> > to storage, even when the power is lost.
> >
> > This setting should only applies (be used) with disk controllers with
> > battery protection or SSDs with power failure protection.
> >
> > In your case, a single HDD will have limited performance, period. 83
> > IO/sec is not exactly great performance.
> >
> > Further, are you aware of the Windows page cache issues on x64
> > systems? It can make the system grind. FB v2.5.2 has a fix to help
> > address the issue, what version are you running?
> >
> > Sean
> >
> >
> first, I am sorry for my poor english.
>
> I known what mean of this option.
> I imagine the main reason is that intel raid chipset has some issues
> with this option on win7 os.
> I try two way to validate my guess.
> 1, use a memory dist to create fdb.
> 2, change storage protocal from sata/raid to ide /sata on bios setting..
> the two ways are natural.
>
> the fb v2.1 && v2.5.2 have been tested.
>
> [Non-text portions of this message have been removed]
>
>
>
[Non-text portions of this message have been removed]