FPHG available

[jukofyork]

[ QUOTE ]
ive been running v.04 on my computer, but I just bought a new computer and downloaded v.06 and cannot get it to work for the life of me. I know very little about stuff like this so in laymans terms if possible. I tried setting it up like I had my old version set up and that didnt work. thanks for the help

[/ QUOTE ]
If you open up the .hhf files, are they being written to? If so, then if you run both v0.04 and v0.06 at the same time are the .hhf files any different?

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

[veganmav]

If anyone is having trouble using auto import, try the steps listed below, but instead of empire, use MultiPoker

it should work


1. Make a folder for mining somewhere.
2. Copy an execuatble into this folder and rename it EmpirePoker.exe (or another skin you don't use).
3. Copy FreePHG.exe into this folder also.
4. Open up some tables (see below).
5. Start running FreePHG.exe from your mining folder and you should start to see .hhf files appearing for the tables you are watching.
6. When you want to import these hands, point PT at this folder by pointing it at the EmpirePoker.exe (it will then think this folder is the empire folder and [auto-]import from their).

(Thanks the Adde for working out this method! )

[gr8vertical]

I have a basic question that hopefully someone can answer. Once I have PT read the .hff files from the folder that FPHG saved them to, can I delete the .hff files from the folder? Thanks

[jukofyork]

[ QUOTE ]
I have a basic question that hopefully someone can answer. Once I have PT read the .hff files from the folder that FPHG saved them to, can I delete the .hff files from the folder? Thanks

[/ QUOTE ]
Yes as they are all stored in PT after you import them, but I would save them anyway as it is possible for PT to corrupt it's DB and if you zip the old .hhf files, they don't take up much space.

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

[paksao]

Great job in making this available.


Cant get started, the tables do not get recognised. Says waiting for tables. Any ideas?

[jukofyork]

[ QUOTE ]
Great job in making this available.


Cant get started, the tables do not get recognised. Says waiting for tables. Any ideas?

[/ QUOTE ]
How long have you left it trying for?
Have you tried restarting Party and/or FPHG?

Try editing run.bat to look like this:

@echo off
freephg.exe -prof -debug >> fphg.log


Then use run.bat to run FPHG and zip/send the log file to me at: [email protected]

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

[jukofyork]

This version is all thanks to Mogobu. He has been working on the FPHG code, and has optimized it again so it should use alot less CPU than before (he has also compiled it on a better compiler than I was using, so this should also decrease CPU load).

I haven't had time to test this yet so I have left v0.06 up just incase their are any problems (you will see v0.07 is listed on the main page as the current test-version).

I'm just about to go away on vacation for the next 2/3 weeks so I wont have time to add any extra to FPHG until I gets back, but here is my ToDo list:

<u>Still ToDo:</u>
1. Make it so that .hhf files are split after N hands for people having problems importing into PT.
2. Fix the the missing newline which can be seen when viewing the hands in notepad.
3. Fix to warn user that "-o" option can't be used at same time as "-r".
4. Add a simple GUI to save all the CLI app questions.
5. Write the FAQ and instructions (to be posted on Ben's Wiki for easy editing by us all).

Have I missed anything?

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

[Mogobu The Fool]

Looks good to me.

I compiled it using Microsoft's Profile-Guided Optimization. I profiled against a "dirty memory" scenario where many tables had been running for an extended period of time, and used a version of the code which ran both RealTime and regular mode scans so that both logic branches got profiled and optimized.

Using the PGO resulted in 10 to 20% more scans per minute than just using the regular VS 2005 C++ compiler (with all relevant optimization turned on.) Using the VS 2005 C++ compiler improved performance over the initial version of the app before I made any changes, but it wasn't a huge improvement - I think it was about 10-15% faster.

Between my hacking away at the code and using the newer optimizing compiler, I'm seeing gains of well over 1,000% in regular mode, and more than double that in RealTime mode. I can run it with a 250ms delay and get roughly four scans per second without creating any serious load on machine. That makes it less likely to miss the end of a hand - the more frequently you scan, the more CPU load, but the less likely you are to miss the end of a hand.

I'm curious what kind of performance differences others notice on their machines, and whether the super-optimized version runs at all. Also, a little long-term run testing would be a good thing!

[FluffyTortus]

Well, the results from running it on my machine for several hours aren't quite the same.

I use a Dell laptop with a Celeron M 1.3Ghz CPU with 1mb L2 cache and 1.5Gb ram to grab hands from 12 SNGs on EmpirePoker. Version 0.06/0.06a used about 10% CPU when running with their default scanning delay of 1,000ms. Version 0.07 uses about 22% CPU with a delay of 250ms.

So maybe this is an improvement since it scans a lot more than previous versions, but the CPU usage is certainly noticable.
I have never used Realtime mode and I haven't tried altering the default scanning delay.

Do you think this has something to do with you optimizing for a different CPU or maybe the code is not optimized for Empire yet?

[Mogobu The Fool]

OK, that was my stupid mistake - I shouldn't have changed the default scanning delay. I meant that to be a temporary change for use while debugging. . . People are used to the current defaults. I will fix it in a "debug" release, along with other fixes as people report in. When Juk gets back, we'll get it posted.

I would like to know if it's because of optimizing based on profiling on MY machine, or what. . .

Can you try this?

Make a file called run.bat with the following in it:

<font class="small">Code:</font><hr /><pre>@echo off
FreePHG.exe -prof</pre><hr />

The "-prof" turns on profiling mode, which will report how many "scans per minute" your machine produces. The "scans per minute" line appears once in a while, and it starts with a lot of spaces, so you'll see it off on the right side of the DOS box in which it's running.

If you could fire this batch against .06 and against .07, I'd be curious to see what kind of numbers you turn up.

If you want to try running .07 with a 1,000 ms delay, use this:

<font class="small">Code:</font><hr /><pre>@echo off
FreePHG.exe -d 1000</pre><hr />

The "-d 1000" changes the delay to 1000. Don't use "prof" mode at the same time when running, because the profiling work slows things down a tiny bit.

Thanks!