Uri Blass wrote:based on this forum the author of stockfish recommended the optimistic version.
Correct in a way. Here is what Marco posted:
Stockfish 1.2 Optimistic is probably the strongest and has an aggressive play but sometime it is weak with lower rated engines.
Stockfish 1.2 Default is my favorite, it has a more solid play and makes less mistakes and my guess is that at longer time controls should be good.
Yes, it is.
The point here is that I was NOT able to test al longer time control and with an opponent's pool as Graham and others do.
I have tested at blitz time and in my tests optimistic is stronger. But because the difference between the two versions is that optimistic has a more aggressive futility pruning setup (hence the name) and I had no data at longer time controls AND the previous versions of Stockfish where weaker then expected at these time controls I tried to be conservative and raccomended the default.
Now that "default" has proven at least not worst then previous versions, I am more optimistic also myself and believe "optimistic" could be stronger.
As example in the new version of Stockfish, now in development, I am using the "optimistic" setup.
I believe that the main problem is not pruning but bugs
In one of the games stockfish optimistic even blundered in the first move of the game
I think that stockfish should clear the hash when it starts games so results can be reproducable.
Uri Blass wrote:
I believe that the main problem is not pruning but bugs
Unfortunatly these kind of bugs are very difficult to kill down because are not reproducible.
Anyhow I would think it is an accident that optimistic shown that bugs and default not because the ONLY difference is in a couple of code lines in the razoring...but with software you never know for sure....
Uri Blass wrote:
I believe that the main problem is not pruning but bugs
Unfortunatly these kind of bugs are very difficult to kill down because are not reproducible.
Anyhow I would think it is an accident that optimistic shown that bugs and default not because the ONLY difference is in a couple of code lines in the razoring...but with software you never know for sure....
The first question is if stockfish one cpu produce deterministic output.
unfortunately I see that it is impossible to change the threads parameter under fritz8 gui and the reason is that the window of parameters is too big and I do not see all of it and the threads parameter is probably at the bottom of the window when I do not see it.
Uri Blass wrote: the window of parameters is too big and I do not see all of it and the threads parameter is probably at the bottom of the window when I do not see it.
The only way is to enter the highest resolution your graphics card will permit (1280x 1024 is enough), make the changes you want in the Engine parameters window and save the configuration as a new engine.
Uri Blass wrote:
I believe that the main problem is not pruning but bugs
Unfortunatly these kind of bugs are very difficult to kill down because are not reproducible.
Anyhow I would think it is an accident that optimistic shown that bugs and default not because the ONLY difference is in a couple of code lines in the razoring...but with software you never know for sure....
The first question is if stockfish one cpu produce deterministic output.
unfortunately I see that it is impossible to change the threads parameter under fritz8 gui and the reason is that the window of parameters is too big and I do not see all of it and the threads parameter is probably at the bottom of the window when I do not see it.
Uri
Version for Fritz/Chessbase with reduced uci parameters screen which fits widescreen lcd displays.
Uri Blass wrote:The first question is if stockfish one cpu produce deterministic output.
Using Jim Ablett's (16-9) version (fitting parameter screen), I see that with 1 thread, infinite analysis is deterministic (everytime I test, I have to relaunch the application and set to 1 thread, since this is the only way to clear hash with Stockfish or Glaurung)
Uri Blass wrote:The first question is if stockfish one cpu produce deterministic output.
Using Jim Ablett's (16-9) version (fitting parameter screen), I see that with 1 thread, infinite analysis is deterministic (everytime I test, I have to relaunch the application and set to 1 thread, since this is the only way to clear hash with Stockfish or Glaurung)
Glaurung 2.2 and Stockfish 1.2 both have, on the UCI parameters screen, a "Clear Hash" button you can click to clear the hash.
Uri Blass wrote:
I believe that the main problem is not pruning but bugs
Unfortunatly these kind of bugs are very difficult to kill down because are not reproducible.
Anyhow I would think it is an accident that optimistic shown that bugs and default not because the ONLY difference is in a couple of code lines in the razoring...but with software you never know for sure....
The first question is if stockfish one cpu produce deterministic output.
unfortunately I see that it is impossible to change the threads parameter under fritz8 gui and the reason is that the window of parameters is too big and I do not see all of it and the threads parameter is probably at the bottom of the window when I do not see it.
Uri
Version for Fritz/Chessbase with reduced uci parameters screen which fits widescreen lcd displays.
Another way that i use is to tick "auto hide" the toolbar and that gives you much more room at the bottom. Not quite enough at first- but you can keep bumping it up at the top and finally force it up enough to see what you need at the bottom. I have done this with a number of engines, and i bet it would work for you here.
Uri Blass wrote:
The first question is if stockfish one cpu produce deterministic output.
Yes it does, otherwise it would be impossible to develop.
I always test the cleanup/code rewrite patches benchmarking on a position's set, at fixed depth and with one thread only and checking if node count doesn't change, as example with the command line:
stockfish bench 100 1 13 default depth
Regarding the parameters, I start to think they are too much!!!!
To visually shrink them I am thinking of adding one more called "advanced options" (unchecked by default) that let Stockfish to show additional parameters only on request.