Changelog:
-Fixed the OwnBook checkbox bug
-Matching Stockfish beta having timestamp 1586274421 - continuation histories when in check.
For Windows and Android (NOT UPDATED FOR ANDROID), NO LINUX, NO MACOS.
This distribution indluces:
Windows:
- Stockfish_Polyglot_2020-04-24_32bit.exe FOR 32BIT WINDOWS SYSTEMS
- Stockfish_Polyglot_2020-04-24_32bit_general.exe FOR GENERAL 32BIT WINDOWS SYSTEMS
- Stockfish_Polyglot_2020-04-24_32bit_general.exe FOR OLD MACHINES HAVING 32BIT WINDOWS S
- Stockfish_Polyglot_2020-04-24_32bit_ppc-32.exe FOR POWER COMPUTERS HOSTING WINDOWS SYSTEMS 32BIT
- Stockfish_Polyglot_2020-04-24_x64_BMI2.exe FOR 64BIT WINDOWS SYSTEM AND INTEL PROCESSORS
- Stockfish_Polyglot_2020-04-24_x64_general.exe FOR GENERAL 64BIT WINDOWS SYSTEMS
- Stockfish_Polyglot_2020-04-24_x64_modern.exe FOR MODERN MACHINES ON 64BIT WINDOWS SYSTEMS (POPULATION COUNT ALGORYTHM INCLUDED)
- Stockfish_Polyglot_2020-04-24_x64_ppc.exe FOR POWER COMPUTERS HOSTING WINDOWS SYSTEMS 64BIT (POPULATION COUNT ALGORYTHM INCLUDED)
- Stockfish_Polyglot_2020-04-24_x64_x64.exe FOR 64BIT WINDOWS SYSTEMS
Compiles by Pavel C.
Android:
- Stockfish_Polyglot_2020-04-24_arm7-pgo for 32bit Andoid systems
- Stockfish_Polyglot_2020-04-24_arm8-pgo for 64bit Android systems
Compiles by Skynet (OutSkirts forum)
CHANGES SINCE THE LAST UPDATE:
See the README file inside.
Thank you again Kramnik
- However, I must say I have not found a version of SF Polyglot... able to play BIN books without "best book move"
- I understand (partially) you say the polyglot code is same as Brainfish, and it should work the same; but it doesnt
- As a hint... Brainfish code is 2 bin book; SF is 4 bin book.
- AFAIK, this is a typical initialization problem already fixed in other SF derivatives
Best regards
PD: I took the risk of starting a tournament using this engine for a mix of CTG and BIN books...
--- but I will stop tournament due to this bug.
Thanks for your effort but it seems one has to register at least at one of the offered mirror- sites?
Nope.. Just choose the available link for your country and download the engine. No registry needed.
This time it worked, thanks again.
Yet, without wanting to disrespect your work, the MagicTacticSolver seems strange to me as for most positions I tried it with. Evals jump up and down and are mainly fully wrong. Tactical hot shots are found now and then, yet there are more tactical best move positions, that are solved easily by SF dev. and aren't by TacticSolver.
Should I give examples or is it to be used for certain kind of positions only?
peterm wrote:
Yet, without wanting to disrespect your work, the MagicTacticSolver seems strange to me as for most positions I tried it with. Evals jump up and down and are mainly fully wrong. Tactical hot shots are found now and then, yet there are more tactical best move positions, that are solved easily by SF dev. and aren't by TacticSolver.
Should I give examples or is it to be used for certain kind of positions only?
I'm not the author of the engine and i have no idea about the engine's criteria, i simple updated the engine. Perhaps the engine needs some tuning regarding "jumping" evaluation, but again, i have no idea how to fix this.
The number of books don't have anything to do with this matter: being global variables I wouldn't even be able to compiled if I've left something out. I guess the problem is somewhere else, most probably in the very own polyglot libraries, untouched from the Brainfish source apart just adding two more global variables (Book3 and Book4), all the remaining code of these libraries (polybook.cpp and polybook.h) are generic functions, and if you would know something about programming you would know they behaves generically for any calling, provided that all the parameters are satisfied.
Alex1976 (Skynet?...) or RomeoA28 can confirm it.
I never had the time to dig more on this issue.
tacos40 wrote:Thank you again Kramnik
- However, I must say I have not found a version of SF Polyglot... able to play BIN books without "best book move"
- I understand (partially) you say the polyglot code is same as Brainfish, and it should work the same; but it doesnt
- As a hint... Brainfish code is 2 bin book; SF is 4 bin book.
- AFAIK, this is a typical initialization problem already fixed in other SF derivatives
Best regards
PD: I took the risk of starting a tournament using this engine for a mix of CTG and BIN books...
--- but I will stop tournament due to this bug.
kramnik wrote:The number of books don't have anything to do with this matter: being global variables I wouldn't even be able to compiled if I've left something out. I guess the problem is somewhere else, most probably in the very own polyglot libraries, untouched from the Brainfish source apart just adding two more global variables (Book3 and Book4), all the remaining code of these libraries (polybook.cpp and polybook.h) are generic functions, and if you would know something about programming you would know they behaves generically for any calling, provided that all the parameters are satisfied.
Alex1976 (Skynet?...) or RomeoA28 can confirm it.
I never had the time to dig more on this issue.
tacos40 wrote:Thank you again Kramnik
- However, I must say I have not found a version of SF Polyglot... able to play BIN books without "best book move"
- I understand (partially) you say the polyglot code is same as Brainfish, and it should work the same; but it doesnt
- As a hint... Brainfish code is 2 bin book; SF is 4 bin book.
- AFAIK, this is a typical initialization problem already fixed in other SF derivatives
Best regards
PD: I took the risk of starting a tournament using this engine for a mix of CTG and BIN books...
--- but I will stop tournament due to this bug.
I updated the engine till yesterday's patch "continuation histories when in check". https://github.com/official-stockfish/Stockfish/commit/bb5589b829b79d7c60a820d4b1634dccbc4bbb3f
where BBMI was fixed, in case it can help; link below