speed issues

[checkits]

I just started using PT3 after uploading my hands (i.e., almost 700,000).

PT3 seems markedly slower than PT2 in terms of populating pages (e.g., my "summary" tab has been going for about 5-10 minutes). In PT2 this same (or very similar) summary tab would have taken about 10-20 seconds.

What am I doing wrong, because I can't imagine that PT3 would be slower than PT2 if operated correctly.

Please advise.

[keggler]

20 secs vs. 10 mins on the same pc spec?

network or localhost PT3 database?

get procexp from sysinterals and post back cpu, memory & i/o from the postgres processes and the pt3 processes...


susan

[wbuhr]

I just started using PT3 after uploading my hands (i.e., almost 700,000).

PT3 seems markedly slower than PT2 in terms of populating pages (e.g., my "summary" tab has been going for about 5-10 minutes). In PT2 this same (or very similar) summary tab would have taken about 10-20 seconds.

What am I doing wrong, because I can't imagine that PT3 would be slower than PT2 if operated correctly.

Please advise.

To me it seems that beta 14 is much slower than beta 13 and also PT2. I have a very small database with about 25 k hands and about 17 k players and from clicking the Pocker Tracker 3 button until the end of populating the time needed is about 6 minutes (local database). Machine is no extreme runner but a 3GHZ P4 with 2 Gig RAM and SATA drives.

[checkits]

20 secs vs. 10 mins on the same pc spec?

Maybe I exagerated -- maybe the difference is 20 seconds and 4 minutes. I didn't time it formally.

network or localhost PT3 database?

I'm on a server, and the PostGres db is local on the same server.

The server is a dual quad core w/ a bunch of ram (i forget how much).

get procexp from sysinterals and post back cpu, memory & i/o from the postgres processes and the pt3 processes...

I don't know how to do that. Where do I find sysinternals?

[keggler]

[quote="wbuhr]
To me it seems that beta 14 is much slower than beta 13 and also PT2. I have a very small database with about 25 k hands and about 17 k players and from clicking the Pocker Tracker 3 button until the end of populating the time needed is about 6 minutes (local database). Machine is no extreme runner but a 3GHZ P4 with 2 Gig RAM and SATA drives.[/quote]


have you clustered and vacuumed your db? and defrag'ed your hard-drives?

if you have, you need to get procexp as well as post back as i requested...


susan

[keggler]

http://www.sysinternals.com


susan

[wbuhr]

[

Thanks for your advice the defragmentation and clustering helped.
Starting Time for now is about 30 sec in summary for the PT3 and the postgress processes.

I try to keep in my mind that I have to defrag the hdd more often.

[checkits]

Ok. I defragged recently, but I'll do it again tonight as well as cluster/vacum.

get procexp from sysinterals and post back cpu, memory & i/o from the postgres processes and the pt3 processes...

Can you tell me how to "get procexp from sysinternals" and what to do w/ it. Sorry, I'm not very technically inclined.

[tarix]

20 secs vs. 10 mins on the same pc spec?

Are you suggesting this isn't the case for you?

I haven't timed it myself, but for me the exact same "database" is instant in PT2 and takes time on PT3. It is really difficult to get a timing because you are comparing instant to non-instant. Database maintenance seems to have no real effect on the performance for me. I can easily fit my entire database in the disk cache anyway.

[keggler]

20 secs vs. 10 mins on the same pc spec?

Are you suggesting this isn't the case for you?

I haven't timed it myself, but for me the exact same "database" is instant in PT2 and takes time on PT3. It is really difficult to get a timing because you are comparing instant to non-instant. Database maintenance seems to have no real effect on the performance for me. I can easily fit my entire database in the disk cache anyway.

sorry i have not the faintest idea what you're on about here with your words above... e.g., what does instant & non-instant mean?

i don't really want to get into an arguement here over this, but disk caches are used to stage i/o's for all sorts of activity on the drive; it's not a simple case of shoe-horning the whole db into it - anyhow, i thought disk caches were orders of MB's, 32 at most - you have a 32MB PT database? there's all sorts of other interactions here with db indexes, L2 CPU cache, and the o/s' control over main memory and paging...

i have no experience with PT2 so cannot really comment on speed comparisons, but I am led to believe the schemas are different between 2 & 3 and 3 holds more data in the db for a given hand, player, tournament, etc - so it seems to me, speed comparisons against the database are not realistic in any case.

my fairly common sense and simplistic view is to take the PT3 app fronting a very robust RDBMS and just make sure the customers here are running their systems sensibly, maybe the occasional tweak on the postgres conf files and count on the development team to write good code behind a well designed schema and not dwell on the past, with "this isn't as fast as this..." - it's time to move on...


susan