According to Robert Houdardt, he reckoned about 500 games would be suffice to test his older and newer Houdini versions.
At last, after about two weeks of nonstop testing, my PC have finished a 500 games match (20 mins on each side) between the two versions.
Here is the results with ponder on:
Out of the 500 games, Houdini 1.5a won 119 games, Houdini 2.0 won 143 games and 238 games were drawn.
Each engine won roughly about a quarter of the games and the remaining games were drawn.
Thus won one, loose one, draw two, etc.
Houdini 1.5a scored 238 points (about 47%) and Houdini 2.0 scored 262 points (about 52%).
The ELO difference was 15, just what Robert Houdart said about the new Houdini 2.0!
Guess which chess engine is now my new favorite and number one?
Houdini 2.0 won Houdini 1.5a (32 bit) in 500 games
Re: Houdini 2.0 won Houdini 1.5a (32 bit) in 500 games
can we have the games?
Re: Houdini 2.0 won Houdini 1.5a (32 bit) in 500 games
Oh yeah Pieterhb, remember that if you don't provide any games, your test isn't any different from running this bash script that doesn't play any games!
Re: Houdini 2.0 won Houdini 1.5a (32 bit) in 500 games
I ran many different tests the past couple of weeks.
Will have a look tonight to see if I still have the files (hope so) and then I will upload it for those who are interested in it.
I have seen plenty of brilliant games being played by the two chess engines and would love to share it with you.
Will have a look tonight to see if I still have the files (hope so) and then I will upload it for those who are interested in it.
I have seen plenty of brilliant games being played by the two chess engines and would love to share it with you.