Upgraded to 0.25, same problem. When I import hands now all it does is show me winning, I wish my results were as good as it says as the graph is almost a straight line from the bottom left corner to the upper right corner.

Here is a screen shot of the new limits I have, still missing 400, 600 and now 1000 for limit games plus other limits are missing. I have never played in a game where the big bet would be $3, $5, $8 so I have no idea what those are from. And the limit games are stud8, razz, and badugi, mostly being stud8. But I also play rush nlhe, 'normal' nlhe, nlsngs, rush plo8, 'normal' plo8, and some limit holdem.

That sounds... odd.

Are you able to send me your hh archive for testing? i'll pm my email address shortly.

Great program, thanks so much for making this! My dream of pokerstars + linux is basically complete. But...

I am experiencing extremely high CPU usage from Python. I have a not-particularly-new machine (core 2 duo e7400) but still plenty fast, I could easily 20-table stars with hold em manager in a virtualized Windows environment.

Even just 4-tabling, the HUD is consuming >50% of CPU load and eventually crashing the whole system if I try to play 10 or more tables.

My details:

python 2.7
sqlite db
Ubuntu 11.04
Gnome desktop environment
Compiz
Pokerstars run under latest version of Wine

It's a pretty "vanilla" version of linux I'm running, I haven't really changed much. I believe I've installed all the Python dependencies like sqlalchemy. I've even tried to change things like setting "time between import" very high (1 minute) and I'm still having the same problems.

My database is 1.6 million hands, not sure if that's the problem. Can SQLite handle that or do I need to switch over to postgre/mysql?

My HUD is very minimal, 8 basic stats and I have turned off mucked cards.

Would it help to do the DB maintenance stuff like rebuilding indexes?

Any help is much appreciated. I tried on IRC but no one was available to answer the question. Thanks in advance.

You kinda need to hang around for a while (it took a touch over an hour for a response and you'd left by then)

With 1.6M hands i'm guessing the issue is probably the size of the HUD Cache.

As a test, could you try rebuilding the HUD Cache with only a couple of weeks worth of data and see if that shrinks your CPU usage please?

for triple draw games there is a program called mucked which shows the cards for pokerstars. this program doesnt work with fpdb because the hud display also has the table's name in it and it confused the other program.

is there a way to mess with the source and or any config and remove or change the name of the table in the hud window (the small thingy)??

at least as we wait for mucked cards support for draw games.

thanks!

ummm sorry the table name is in the window text. can you please tell me which file is that contained in and i would change the variable to null? im a coder so i think i can handle this.

If it is the window with the current list of tables it is in HUD_main.pyw

Changing this source in the exe version on windows won't work, you will need to run the sourcecode + dependencies installed.

call by the #fpdb on irq if you want any help getting the sourcecode running.

I'm not 100% sure this is what you actually want - I doubt(?) that the other program is reading gtk text from within an open window.

The code that actually sets the text in the "HUD Main Window" is in HUD_main.pyw

Changing tale.site and temp_key to " " or some other string will stop the text from appearing in that window.

See gimicks post re: .exe versions though.

Sorrow

thanks and sorry I didn't wait around in IRC, I'll try again today.

I rebuilt the cache to just the last few weeks and it doesn't make any difference. Python using 50+% of CPU while 4-tabling, with a very simple HUD and import time set to every 40 seconds. Any other ideas?

Should I try rebuilding index? Or maybe changing some preferences? I have already changed "stat_range" to T from A in preferences so that it doesn't try to use the whole DB.

One thing I've noticed is that "cachesessions" and "faststoreHUDcache" are turned off in my import preferences, do I need to change that?

here's a screenshot of prefs if this helps at all:





Thanks so much guys.

EDIT: BTW I am using the 0.25 from the .deb installation.

Are you seeing a difference when playing with the pulse time?

How long do the graphs take to generate?
Making sure Hero is set to T should minimise the size of the query.

Reindex is a good idea, but i'm not sure how long it will take (or if it will even complete) on a db that size.

Configuring MySQL may help things too.

Are you technically competent? i'd like to run a few things from the command line and see if we can't identify the issue. I suspect we can improve the index on the HudCache to perform better with large dbs.

cachesessions being off should be fine. faststoreHUDcache is now legacy and doesn't actually do anything.
Could you switch off callFbdbHud and 4 table and tell me if the CPU usage is still that high?

Sorrow.

Thanks for your reply.

If I turn off the HUD (callpfdbhud = False) then python doesn't really get used at all. I mean it shows up in the processes tray but never really gets above 5% CPU.

Graphs are total killers. I tried to graph ~100 hands, just $$, no showdown winnings, one site only. Python was using 70+% of CPU and it took a good solid 3-4 minutes to produce the graph.

Well I use linux so I guess I'm sort of technically competent? I am no programmer but I can certainly manage the terminal a bit.

Is SQlite the problem here? I know nothing about it, in the past I had my DB in PostgreSQL but when I d/l FPDB SQlite was already setup by default. So I just imported and started using it. Does SQlite have problems with really big DB?

Appreciate all the help. I will try IRC again this afternoon.

S0rrow was great and worked with me on IRC for a while today, but I'm still stuck. With 5 tables I'm using 75% of CPU on both cores, with 10 it's completely freezing my computer.

The system sees each part of the HUD as a separate window. So 1 6max pokerstars table = 8 windows: 1 for me, 5 for villains, 1 for FPDB HUD menu at top left, and then the pokerstars table itself. So that means 5-tabling is really 40-tabling.

This makes me wonder if the problem is something with GTK or Compiz?

I wondered if this was a Unity-specific problem and tried "ubuntu classic" option at the login (old vanilla GNOME). Same situation, huge python CPU usage.

I am going to mess around with Compiz setting a little and Wine settings too but I'm getting ready to give up pretty soon and just start over.

Oh I forgot to mention the size of my DB is 4.5 G for 1.6mil hands, does that seem very large or just normal?

Thanks again guys. I'll be back on IRC tomorrow one more time.

one more question, I've opened my fpdb.db3 file in a sql manager and I see this, look ok?

I've found a partial solution to this now, i'll clean things up and get you to test a bit later.

Sorrow

Using 75% on both cores? That's an interesting item, because we don't use anything in python that supports multiple core execution, unless they've added that to python somewhere automatically?

I'm trying to run the hud from stable 0.22 on Gentoo + Stars/wine and I'm having trouble to get it to find the tables:

Hud-errors shows
And in Hud-log I see:

Any help will be appreciated.

R.

Ooops, it turns out I was missing xwininfo on my system.

R.

Thanks for your reply on the previous page. I'm not surprised I was doing something stupid.

just an update, the IRC chat helped me get things working, but we had to switch to postgreSQL. It looks like SQLite has some limitations for very big DBs like mine (>1,500,000 hands) so if you have a giant DB just start with postgreSQL.

I am now able to run the HUD on basically any number of tables and python usage never goes about 10%. Thanks everyone who helped, and if you're a big grinder just start with PostgreSQL, the config is really quite simple.

is there anyway of getting fpdb working on ipoker yet???

I've made the software changes this evening to activate the mucked card display for draw games (27-3, 27-1, 5draw and badugi)

Join us on #IRC to get this working on your system, it requires a small change to the fpdb source and config file.

Otherwise this will be release in 0.26 whenever that comes out.

I/We appreciate the effort you put in with this Verstehen. The process of working through a database that size has helped us highlight several performance issues.
I've only received 1 hand history file for the iPoker network - a 7 Card stud H/L file, which imports fine.

I need more hand histories to expand support.

Nice work gimick.

I've only received 1 hand history file for the iPoker network - a 7 Card stud H/L file, which imports fine.

I need more hand histories to expand support.

where can I send you some... I have a bunch of NLHE

carl.gherardi@gmail.com

Please zip the files first.

At least 1 file with a known result is particularly useful (ie. file.txt has 111 hands and Hero lost 43.57) as it allows me to find the corner cases a lot more quickly

The HUD indication doesn't change when I switch to a new game...
aka the HUD remains with the NL HUD settings if this is the game i was playing when fpdb started up...

Is this a known bug or just a wrong set up on my side ?

Thanks a lot for the feedback

Eddy

PS : running v.025