EXchess v7.26 released

[dchoman]

EXchess v7.26 beta is now available from my website.

https://sites.google.com/site/experimentalchessprogram/

Things have been very busy this year, so not a lot of changes, but a few. In my testing it is 30-40 elo stronger than v7.18 in self-play. Probably half of that against other engines.

- Dan

[Graham Banks]

Thanks Dan.

[Sylwy]

Sorry !

Silvian R

See, please:

http://www.talkchess.com/forum/viewtopi ... 023#570023

[dchoman]

Hi Ruxy,

I really appreciate this test, but I think it might be too early to claim for sure that v7.26 is weaker than v7.18. The difference you find is 56 elo, but the error bars on a test like this (120 games) are very large and could lead to this difference by chance.

However, I cannot easily test the w32 version, so it is possible there is something going on with the 32 bit version that is unusual. When I run the w32 version on my 64-bit windows it plays the same as the 64 bit version, but only slower (as expected). I don't have a 32 bit machine to test it on, but I will try to take a closer look. For the 64 bit version under MacOS X, v7.26 was something like 52 +/- 12 elo stronger than v7.18 in a head-to-head match of 500+ games that I ran at the time of release (TC of 1 min/game). Versus other opponents, I certainly expect less of a gain than that, and I don't know what will happen at longer TC, so that is another possibility for the difference we see.

Thanks for letting me know what you found in your test!

- Dan

[Sylwy]

Hi Ruxy,

I really appreciate this test, but I think it might be too early to claim for sure that v7.26 is weaker than v7.18. The difference you find is 56 elo, but the error bars on a test like this (120 games) are very large and could lead to this difference by chance.

However, I cannot easily test the w32 version, so it is possible there is something going on with the 32 bit version that is unusual. When I run the w32 version on my 64-bit windows it plays the same as the 64 bit version, but only slower (as expected). I don't have a 32 bit machine to test it on, but I will try to take a closer look. For the 64 bit version under MacOS X, v7.26 was something like 52 +/- 12 elo stronger than v7.18 in a head-to-head match of 500+ games that I ran at the time of release (TC of 1 min/game). Versus other opponents, I certainly expect less of a gain than that, and I don't know what will happen at longer TC, so that is another possibility for the difference we see.

Thanks for letting me know what you found in your test!

- Dan

Just now 7.26b & 7.18b versions play in a tournament together with others 10 engines. Will be much more games !

Have a nice weekend !

Silvian R

[JohnW]

Should I be able to create a UCI engine in the Fritz gui with this? I can select the .exe but it doesn't populate with the author so it doesn't allow me to complete the uci creation. thank you..

[Sylwy]

Should I be able to create a UCI engine in the Fritz gui with this? I can select the .exe but it doesn't populate with the author so it doesn't allow me to complete the uci creation. thank you..

Hey man !

EXchess 7.26b is a pure Winboard engine ! Use (please ) the Wb2Uci.exe & .eng to create an UCI engine under Fritzy GUI ! (Fritz GUI supports only the native ChessBase communication protocol and the UCI communication protocol) !

Have a nice day & night !

Silvian R

NR: a bonus information : EXchess has a very particular structure of the Wb2Uci.eng file !!!!! Otherwise............

[JohnW]

ok, great. Thank you..

[Graham Banks]

Hi Ruxy,

I really appreciate this test, but I think it might be too early to claim for sure that v7.26 is weaker than v7.18. The difference you find is 56 elo, but the error bars on a test like this (120 games) are very large and could lead to this difference by chance.

However, I cannot easily test the w32 version, so it is possible there is something going on with the 32 bit version that is unusual. When I run the w32 version on my 64-bit windows it plays the same as the 64 bit version, but only slower (as expected). I don't have a 32 bit machine to test it on, but I will try to take a closer look. For the 64 bit version under MacOS X, v7.26 was something like 52 +/- 12 elo stronger than v7.18 in a head-to-head match of 500+ games that I ran at the time of release (TC of 1 min/game). Versus other opponents, I certainly expect less of a gain than that, and I don't know what will happen at longer TC, so that is another possibility for the difference we see.

Thanks for letting me know what you found in your test!

- Dan

Only 105 games in the CCRL 40/40 list so far, but at present, EXChess 7.26b 64-bit is 30 elo ahead of EXchess 7.18b 64-bit.

[Sylwy]

Hi Ruxy,

I really appreciate this test, but I think it might be too early to claim for sure that v7.26 is weaker than v7.18. The difference you find is 56 elo, but the error bars on a test like this (120 games) are very large and could lead to this difference by chance.

However, I cannot easily test the w32 version, so it is possible there is something going on with the 32 bit version that is unusual. When I run the w32 version on my 64-bit windows it plays the same as the 64 bit version, but only slower (as expected). I don't have a 32 bit machine to test it on, but I will try to take a closer look. For the 64 bit version under MacOS X, v7.26 was something like 52 +/- 12 elo stronger than v7.18 in a head-to-head match of 500+ games that I ran at the time of release (TC of 1 min/game). Versus other opponents, I certainly expect less of a gain than that, and I don't know what will happen at longer TC, so that is another possibility for the difference we see.

Thanks for letting me know what you found in your test!

- Dan

Only 105 games in the CCRL 40/40 list so far, but at present, EXChess 7.26b 64-bit is 30 elo ahead of EXchess 7.18b 64-bit.

Hello Graham !

All my tests were with the w32 versions. I haven't a 64-bit OS !

Silvian R