Page 1 of 1

PT2 Performance better than PT3. Why?

PostPosted: Mon Aug 04, 2008 11:22 am
by Lythande71
Hi,

the performance of showing player stats for a given period - say for a particular month - is up to 10 times faster with PT2 than with PT3. Regardles wich tab i choose.
I run several tests with different time periods, and PT2 is always remarkably faster. How could that be???

- both PT versions run with postgreSQL 8.3.
- data is the same for both versions (obv :roll: )
- approx: 500000 hands
- approx: 30000 players
- approx: 10000 tournaments (including tournament summaries)

I can only guess, what could be the cause:
- PT2-Option to only show players with at least X hands reduces the number of players in my example from 30000 to 5000
- I've created indices on some of the PT2-Database tables myself to speed up my own queries. I don't think this speeds up PT2 itself, coz these indices are not used by PT2 but i want to give u all information that might be relevant.

So, here are my questions:
- has anyone else made similar experiences comparing the performance of PT2 and PT3?
- what can i do to speed up PT3?

Please let me know if more information is needed to analyze this strange behavior.

Rgds,

Lythande71

Re: PT2 Performance better than PT3. Why?

PostPosted: Mon Aug 04, 2008 11:56 am
by WhiteRider
Are you using Beta 15? This has some database indexing improvement, but you need to cluster/vacuum/analyse the database from within PT3 first.

Re: PT2 Performance better than PT3. Why?

PostPosted: Tue Aug 05, 2008 9:12 am
by Lythande71
WhiteRider wrote:Are you using Beta 15? This has some database indexing improvement, but you need to cluster/vacuum/analyse the database from within PT3 first.


Yep, I'm using Beta 15. Yep, I did cluster/vacuum/ analyse several times.

I can add another guess:
- I'm using the "delete players with no hands or winnings" option a lot in PT2. Since PT3 imports all the data from the handhistories again, it could be that the number of players in the DB is remarkably higher than in PT2 especially because I used to play a lot MTTs with a huge number of players.

Could this be the root cause?
How can I delete players with no hands or winnings in PT3?

I'm quite familar with postgres and database modelling in general. I haven't looked into the new data model of PT3 yet. But if it helps I could give you detailed information about anything that is provided by postgres admin tool.

Rgds,

Lythande71

Re: PT2 Performance better than PT3. Why?

PostPosted: Tue Aug 05, 2008 9:48 am
by WhiteRider
Well, yes, I guess that will make quite a difference.
That option isn't available yet in PT3, but I'm sure it will be added. Both Selective Purge and Tournament summary entry are due to get an update soon, so maybe this will be one of the things that gets added.
Development Timeline.

Re: PT2 Performance better than PT3. Why?

PostPosted: Tue Aug 05, 2008 11:20 am
by Lythande71
Ok,

Can I manually run an SQL to purge those players?
If yes, could you post or sent that SQL?

thanks in advance

Re: PT2 Performance better than PT3. Why?

PostPosted: Wed Aug 06, 2008 4:23 am
by WhiteRider
It's not as simple as just removing reference to those players because the hands they played are tied to stats of other players, etc., so I'm afraid you'll have to wait for selective purge to be added.

highfalutin