FPHG available

[jukofyork]

[ QUOTE ]
Thanks jukofyork -- a much better performance on my P4 1.8GHZ!

[/ QUOTE ]
Try v0.04 and see if this helps even more (about 40%-50% speed increase). I only have a 1.4GHz and its down to about 10% CPU usage now... [img]/images/graemlins/smile.gif[/img]

Juk [img]/images/graemlins/smile.gif[/img]

[jukofyork]

[ QUOTE ]
Great work!!!
I think a graphic interface would be nice but it´s more important with speed and configurability. I think speed was the major issue in version 0.1 but this was a dramatic change in 0.3. Thanks!

[/ QUOTE ]
Try v0.04 as I have optimized it a little more and now CPU usage has become fairly insignificant (even on my crappy system!) [img]/images/graemlins/smile.gif[/img]

[ QUOTE ]
May I suggest a possibility to mine only tables with a certain amount of players. Statistics are not that interesting from a table with 3-4 players if you normally play full tables. This feature was used by PokerTracker and this feature is also available in Lunker.

[/ QUOTE ]
Yes, I can add this and will add this and the ability to re-add the 6max tags in a later version. Thanks for the suggestion (I always thought PT had some kind of option to import only if > N players, but this much have been taken out...) [img]/images/graemlins/smile.gif[/img]

Juk [img]/images/graemlins/smile.gif[/img]

[SamIAm]

[ QUOTE ]
(I always thought PT had some kind of option to import only if > N players, but this much have been taken out...)

[/ QUOTE ]Nope. It's still there. That's why I'm confused. [img]/images/graemlins/smile.gif[/img]
-Sam

[jukofyork]

[ QUOTE ]
[ QUOTE ]
(1) Added two more small optimizations which resulted in about a 40-50% speedup over version 0.03's code (now only using about 10% of CPU on my system scanning 10 tables).

[/ QUOTE ]

Holy crap I'm gonna start running FPHG while playing only to use it with SmartNotes. Awesome, very awesome.

[/ QUOTE ]
Tell us how you get one with it - any feedback helps [img]/images/graemlins/smile.gif[/img]

Juk [img]/images/graemlins/smile.gif[/img]

[jukofyork]

[ QUOTE ]
Amazing. Thanks. I definitely appreciate all the processor tweaks. (Now please get to work on lowering the load of the damn Party client. [img]/images/graemlins/smile.gif[/img] )

I've switched from my paid copy of PHT to FPHG. Just letting you know. [img]/images/graemlins/smile.gif[/img]

[/ QUOTE ]
Just to say sorry it took me a while to optimize it (v0.01 and v0.2 had no optimization effort put into them at all).

Alot of (experienced) coders will know though: their is never much point in optimizing stuff for the sake of it (ie: optimizing every line, using fancy data structures for the sake of it, jamming your code full of pointers, etc), as usually about 1% of your code uses 99% of CPU. Best plan is to wait and then try to optimize that 1% as much as possible, this way you get the biggest gains for the least effort and can keep 99% of the code nice and simple... [img]/images/graemlins/smile.gif[/img]

Sadly though, I think I have hit the point of diminishing returns now and only very small increases possible from now on unless I rethink the whole architecture and start to use dual-threaded string scanning system (which I might do if FPHG stays alive indefinitely, but its alot of work if it eventually gets banned or the method nullified via a client update...) [img]/images/graemlins/smile.gif[/img]

Juk [img]/images/graemlins/smile.gif[/img]

[SamIAm]

[ QUOTE ]
Alot of (experienced) coders will know though: their is never much point in optimizing stuff for the sake of it

[/ QUOTE ]Hey, I'm not going to give you flak for optimization. I'm the guy who spent hours and hours this week writing in a SCRIPTING LANGUAGE. We don't even have a compiler! [img]/images/graemlins/smile.gif[/img]
-Sam

[jukofyork]

[ QUOTE ]
Hey, I'm not going to give you flak for optimization. I'm the guy who spent hours and hours this week writing in a SCRIPTING LANGUAGE. We don't even have a compiler! [img]/images/graemlins/smile.gif[/img]

[/ QUOTE ]
Hey, I know u weren't giving me flak [img]/images/graemlins/smile.gif[/img] I just added that bit in the hope that some Comp Sci undergrads might read it... I can't count the number of times I seen people use fancy data structures and attempt to optimize every line of code, only to end up with an unreadable mess which is actually slower than had they just done post-optimization on some simple code... [img]/images/graemlins/smile.gif[/img]

A classic example from my FPHG would be jumping in using a Boyer-Moore scanner, a Hash Table or a Heap for the table list, Regular Expression library for the string matching, etc. It turned out about 30% of my speedup came from reordering the logical evaluation order of 2/3 if statements and removing a strlen() call from a for loop! [img]/images/graemlins/smile.gif[/img]

Juk [img]/images/graemlins/smile.gif[/img]

[jukofyork]

As I mentioned v0.04 can do "realtime" writing of logs now. I guess this can be used by botters, but it can also have legit uses, and PHT was writing in "realtime", so I added this feature.

<u>Please don't use this unless you have a reason too</u>, as it is slower and can in theory miss the end of hands (and write 1/2 finished hands). The non-realtime version will only write fully complete and 'sanity checked' hands...

Here are some notes on how to use the "realtime" feature and some ideas on parameters you could try using (I not using this myself, so can't say for sure what will be best...). I added this sub-thread so I don't get asked endless times in PMs how to use the "realtime" feature... [img]/images/graemlins/smile.gif[/img]

NOTE: The easiest way to enter the command line options for FPHG is just to edit the run.bat file I included in the zip.

FreePHG -r -d 500

Would be: Realtime, delay of 500ms

FreePHG -r -d 300 -p 1

Would be: Realtime, delay of 300ms, above average priority

FreePHG -r -d 100 -p 2

Would be: Realtime, delay of 100ms, high priority

NOTE: Just be careful if setting priority higher than 0 (normal), as can cause OS to go wrong. But I am assuming you would not want to miss any data (like PHT was doing), so uping FPHG's priority should make it work before all else.


If you want to write .txt files in realtime, then use the "-e" option:

FreePHG -r -d 500 -e txt

Would be: Realtime, delay of 500ms, extension of .txt


You will just have to experiment and see which parameters help and balance CPU use with possibly missed end of hands. Tell me how you get on if you use this option and post your results/parameter combinations here, as this will help others who want to try to use the realtime option... [img]/images/graemlins/smile.gif[/img]

Juk [img]/images/graemlins/smile.gif[/img]

[yellowbluebus]

Mine just hovers around 6%!

[jukofyork]

[ QUOTE ]
Mine just hovers around 6%!

[/ QUOTE ]
I know, it's sweet compared to the old versions - my crappy system is thanking me already! LOL [img]/images/graemlins/smile.gif[/img]

Peeps with new 3GHz+ machines and v0.04 must be getting nearly 0% CPU use I guessing? [img]/images/graemlins/grin.gif[/img]

I also just looked and it "seems" like I am getting alot more hands than I used to (I have a feeling the old near 100% CPU version was missing a ton of hands, but not sure yet...).

Juk [img]/images/graemlins/smile.gif[/img]