I set CM3_DATA to point at my original data folder, CM3_TEMP to point at the current folder, and CM3_PREFS at my CM folder.
The basic structure for my bat files is:
The .... is my results parsing/processing code.Code:@echo off set CM3_TEMP=%CD% set CM3_DATA=C:\Program Files\cm0102\Data_org\ set CM3_PREFS=C:\Program Files\cm0102 :repeat_cycle cm0102bmquick.exe -load 00.sav ..... goto repeat_cycle exit
Also the parallellism is not that useful. Just makes the process quicker. Running several benchmarks with the same tactic is still tricky (because of timing).
I think we should remember the objective: create a good tactic. Just it. Not a perfect, not a unbeatable one. Just think about this: if you could find a unique and perfect tactic tha win any game, with any players, against any oponente, the game was ruined for you, except if you delete it. We search for the best tactic ever, but is the fact of we do not find it that make we like to play CM 0102 even incredibles 15 years after its release.
Never the less, the benchmark method is not a perfect away to test a tactic. The assistente manager make stupid changes of mentality, use conter atack and behind the ball with no reason (that is horrible to a off side trap and pressing on strategy), and do not use the best set up of inicial players. I am pretty sure that if you use your experience, use the right players, you will dominate any match with any of the, lest say, 50 best formations on OTL. You will lose the 5% of stupid matches because superkeepers, red cards, contusions, bad keepers, etc., and i doubt that any tactic (of the top 20, for example) would be significant best or worst in those bases. I think the challenge of OTL is to buildt the best tactic for THAT group, THAT league, THAT oponentes and THAT stupid assistant manager.
So... following your advice I've made the following .bat file:
What it does is run 5 instances one after another with a 2 second delay between each.Code:@echo off set _number=0 :start if %_number%==10 goto end set CM3_DATA=A:\Games\CM0102\Data set CM3_PREFS=A:\Games\CM0102 mkdir A:\Games\CM0102\temp\temp1 set CM3_TEMP=%CD%\temp\temp1 copy /y A:\Games\CM0102\cm0102_bm.exe A:\Games\CM0102\cm0102_bm_1.exe copy /y A:\Games\CM0102\a.sav A:\Games\CM0102\a(1).sav start /b /min cm0102_bm_1.exe -load a(1).sav timeout 2 mkdir A:\Games\CM0102\temp\temp2 set CM3_TEMP=%CD%\temp\temp2 copy /y A:\Games\CM0102\cm0102_bm.exe A:\Games\CM0102\cm0102_bm_2.exe copy /y A:\Games\CM0102\a.sav A:\Games\CM0102\a(2).sav start /b /min cm0102_bm_2.exe -load a(2).sav timeout 2 mkdir A:\Games\CM0102\temp\temp3 set CM3_TEMP=%CD%\temp\temp3 copy /y A:\Games\CM0102\cm0102_bm.exe A:\Games\CM0102\cm0102_bm_3.exe copy /y A:\Games\CM0102\a.sav A:\Games\CM0102\a(3).sav start /b /min cm0102_bm_3.exe -load a(3).sav timeout 2 mkdir A:\Games\CM0102\temp\temp4 set CM3_TEMP=%CD%\temp\temp4 copy /y A:\Games\CM0102\cm0102_bm.exe A:\Games\CM0102\cm0102_bm_4.exe copy /y A:\Games\CM0102\a.sav A:\Games\CM0102\a(4).sav start /b /min cm0102_bm_4.exe -load a(4).sav timeout 2 mkdir A:\Games\CM0102\temp\temp5 set CM3_TEMP=%CD%\temp\temp5 copy /y A:\Games\CM0102\cm0102_bm.exe A:\Games\CM0102\cm0102_bm_5.exe copy /y A:\Games\CM0102\a.sav A:\Games\CM0102\a(5).sav start /b /min cm0102_bm_5.exe -load a(5).sav timeout 30 echo %_number% set /a _number +=1 goto start :end exit
It also copies the benchmark exe and the save game and renames it.
It also runs the 5 instances 10 times before ending; resulting in 50 results.
I think I cracked it!
Last edited by Jesus; 01-03-16 at 10:33 PM.
To use it you have to use all Tapani benchmark stuff. Enjoy.
Last edited by Maragato; 02-03-16 at 01:25 PM.
Tapani may i ask you some 'silly' questions?
1)first of all i noticed that using your tactics my players dont get so much injuries.Is this a sign that the AI likes your formations?They also are in good shape after the match (condition)
2Do you have in mind while building a tactic how your players will move so they minimize their effort and so keep their stamina for pressing?
3)If you could be kind can u explain why do you start your wib/wob from wob first?Isnt the opposite possible and effective?
4)And for last a bold question if i may...Do u use some kind of programm that can simulate all AI tactics simultaneously with their variations and arrows , the one above the other, so u could find the most free spots to place your players on every box?
With all my respect its nice having you here and i ll be glad if you have the time and chance to give me some answers
I honestly can't get my head around this bit. Does this mean make individual changes on different squares of the wib/wob (for example, 451_a1 moves MCs in the top left of the WIB, 451_a2 moves MCs in the top middle of the wib, etc) or does this mean make individual changes of the same square in different combinations to see which is best?Do only one change, and save it with a new name. Like 451_b1. Then you go back to your candidate, and do another one change. Save it like 451_b2 or smth.
It's great to see Tapani back on the forum! It's also great to see Monte Carlo simulations in action for CM! The methodology is solid. The only downside is that you may get stuck at local maxima while we search for global maxima. We really don't know what the "points surface" looks like. It might be worthwhile to generate a surface based on some of the classic formations and go from there. I wish I had time to conduct the analysis but thx to Tapani we have tools to get there. Hopefully, someone here will take the lead. I will be very curious to see follow-up work here.
Huh, after so long away from the forums, it's interesting to see people finally moving away from Iodine.
What's more interesting is that tactic creation has become a game on its own, arguably more fun than playing CM0102 in a "normal" way.
Creating a decent tactic which will win you everything is really, really easy with just basic wibwob. Using one of the top tactics, howevrer, sucks the fun out of the game as they render players' ability irrelevant.
Creating superb tactics for the testing leagues does seem challenging. It's a nice evolution of the game.
I glady want to use it but i dont understand s*** sorry maybe my english is bad and maybe i am stupid as F*** because my native language is Dutch and never have taken the time to invest some in these topics. I appreciate the effort that has been put in this topic can anybody help me out that maybe are speaking dutch its for me easy to communicate?
Tapani what code did you inject to get your cm0102_bm.exe to export the results at the end of each season in to one file?
I want to edit the filename/location that is exported by I can't find any reference to the command within the .exe
Hi guys, I'm trying to run some tactic benchmarks and am getting duplicate results seemingly at random times, one tactic I run 10 times has no repeats then the next has about 4! Is there anything I can change in the below code from the benchmark file to include enough of a delay? I've tried changing 'sleep 3' to values of 5,10,20 and 30 with absolutely no difference.
Code:@echo off set nof_threads=5 for /L %%i in (0,1,5) do mkdir dup%%i set CM3_PREFS=%CD% set i=0 :run_next set CM3_TEMP=%CD%\dup%i% start /b cm0102.exe -load PSV.sav sleep 3 set /a i+=1 if %i% LSS %nof_threads% goto run_next
Hopefully this will give you some guidance