Subject | Re: [Firebird-general] About the "decrasing passion" of Firebird users. |
---|---|
Author | Alexandre Benson Smith |
Post date | 2014-02-11T22:40:41Z |
Ann,
Em 11/2/2014 20:13, Ann Harrison escreveu:
Em 11/2/2014 20:13, Ann Harrison escreveu:
On Tue, Feb 11, 2014 at 1:20 PM, Alexandre Benson Smith <iblist@...> wrote:
> Delaying SMP further was not an option, sorry,
> thus no way for v2.6.
No need to be sorry...
We have SMP with CS and SC....
Not to any meaningful degree. A shared page cache reduces disk activity,interprocess communication, memory usage ... nearly everything that makesa system slow. Yes, Classic and SuperClassic use multiple cores, but notefficiently.
Best regards,
Ann
__._,_._
I know...
And the Filesystem disk cache could reduce a lot the need of a shared page cache, as you already know.... And I am almost sure that you have from the top of your head the numbers of how the Filesystem cache could help (or not) CS process (or SC threads), It would be great if you share that kind of knowledge with us :) (yes it's another topic)
The interprocess communication is another history....
I know that a huge new design was needed to make Firebird really SMP friendly. The only point I was arguing is that we could have a shorter release cycle while those huge redesigns could be done without delaying the release of other features.
And I know that it would cause an overhead and that's the point that should be measured if it's doable or not.
If really apreciate all the work done, and if any part of my message sounded otherwise, that's because my English is very bad and that I did not expressed me as I wish. Again, I just talked about the diference of a shorter release cycle and the nigthly builds.