blitzkibitz - Project Hosting on Google Code

Earlier this year Nov 28, 2009 r14 (* converted some files to Windows end of line style (CR LF) ...) committed by valentin.stanciu - * converted some files to Windows end of line style (CR LF) * use IDDFS in main.cpp as well * other minor edits * converted some files to Windows end of line style (CR LF) * use IDDFS in main.cpp as well * other minor edits Nov 28, 2009 r13 (* changed spaces inserted in r11 to tabs) committed by valentin.stanciu - * changed spaces inserted in r11 to tabs * changed spaces inserted in r11 to tabs Nov 28, 2009 r12 (* nicer makefile and -msse flag) committed by valentin.stanciu - * nicer makefile and -msse flag * nicer makefile and -msse flag Nov 28, 2009 r11 (Added Iterative Deepening and new makefile) committed by claudiu.bogdan.mihail - Added Iterative Deepening and new makefile Added Iterative Deepening and new makefile Nov 05, 2009 r10 (* changed indentation from white spaces to tabs * all .cpp f...) committed by valentin.stanciu - * changed indentation from white spaces to tabs * all .cpp files include just it's associated .h file and the header file includes the rest * modified rand.cpp to respect the same coding standards * translated some strings to English * changed indentation from white spaces to tabs * all .cpp files include just it's associated .h file and the header file includes the rest * modified rand.cpp to respect the same coding standards * translated some strings to English Nov 03, 2009 r9 (Added C code for LSB and USE_ASM macro for switching between...) committed by claudiu.bogdan.mihail - Added C code for LSB and USE_ASM macro for switching between asm and regular C code Added C code for LSB and USE_ASM macro for switching between asm and regular C code Nov 03, 2009 r8 (Improved makefile and translated comments) committed by claudiu.bogdan.mihail - Improved makefile and translated comments Improved makefile and translated comments Nov 03, 2009 r7 (Coding style: * changed some brackets * converted some tabs ...) committed by valentin.stanciu - Coding style: * changed some brackets * converted some tabs to spaces Coding style: * changed some brackets * converted some tabs to spaces Nov 03, 2009 issue 2 (Translate all internal strings to English) Status changed by valentin.stanciu - Not completely fixed, this also includes comments or function documentation! Status: Started Not completely fixed, this also includes comments or function documentation! Status: Started Nov 02, 2009 r6 (English strinsg and variables issure fixed) committed by claudiu.bogdan.mihail - English strinsg and variables issure fixed English strinsg and variables issure fixed Nov 02, 2009 issue 2 (Translate all internal strings to English) Status changed by claudiu.bogdan.mihail - Status: Fixed Status: Fixed Oct 22, 2009 r5 (Code Review) committed by claudiu.bogdan.mihail - Code Review Code Review Jun 24, 2009 issue 2 (Translate all internal strings to English) Labels changed by valentin.stanciu - Labels: Priority-High −Priority-Medium Labels: Priority-High −Priority-Medium Jun 24, 2009 issue 8 (Automatic speed testing) reported by valentin.stanciu - Implement some way to measure the overall speed of the program. While partial component testing can be done before we integrate it into the engine, sometimes attenuating factors come into play and we need to measure if it makes a speed improvement or not (and how bit it is) We could make the tester evaluate each intermediate board of a recorded game and report the summed up time it took to evaluate the whole game. But don't let the engine make a move. This way, the recorded game stays the same even if we change the scoring logic. Just evaluate, make recorded move, evaluate, make recorded move, etc.. We can enhance it to remember the time of each move and create fancy statistics. But later on. Implement some way to measure the overall speed of the program. While partial component testing can be done before we integrate it into the engine, sometimes attenuating factors come into play and we need to measure if it makes a speed improvement or not (and how bit it is) We could make the tester evaluate each intermediate board of a recorded game and report the summed up time it took to evaluate the whole game. But don't let the engine make a move. This way, the recorded game stays the same even if we change the scoring logic. Just evaluate, make recorded move, evaluate, make recorded move, etc.. We can enhance it to remember the time of each move and create fancy statistics. But later on. Jun 24, 2009 issue 4 (Create config file for settings) commented on by valentin.stanciu - Transposition Table size should also be kept here Transposition Table size should also be kept here Jun 24, 2009 issue 7 (Implement UnmakeMove) Owner changed by valentin.stanciu - Owner: --- Owner: --- Jun 24, 2009 issue 7 (Implement UnmakeMove) reported by valentin.stanciu - Implement the reverse of MakeMove. This way we won't need to instance another Board class at every step and just keep one instance that we make/unmake moves. The performance benefit won't be so great because of the already compiler optimizations and when (if) we switch to threads we will have to keep at least one instance per thread. This is also an option in xboard protocol that we need to implement to make it feature complete. Implement the reverse of MakeMove. This way we won't need to instance another Board class at every step and just keep one instance that we make/unmake moves. The performance benefit won't be so great because of the already compiler optimizations and when (if) we switch to threads we will have to keep at least one instance per thread. This is also an option in xboard protocol that we need to implement to make it feature complete. Jun 24, 2009 issue 6 (Evaluation speed optimization) Labels changed by claudiu.bogdan.mihail - Labels: Type-Enhancement −Type-Defect Labels: Type-Enhancement −Type-Defect Jun 24, 2009 issue 6 (Evaluation speed optimization) reported by claudiu.bogdan.mihail - Instead of calculating the material value for each player every time in the evaluation ( by iterating through all pieces ), it would be a good idea to keep the board's material value inside the Board class and simply updating it. That way it will shave some time off, after several million evaluations. Instead of calculating the material value for each player every time in the evaluation ( by iterating through all pieces ), it would be a good idea to keep the board's material value inside the Board class and simply updating it. That way it will shave some time off, after several million evaluations. Jun 24, 2009 issue 5 (Maintain a Board Zobrist hash in MakeMove) reported by valentin.stanciu - Currently we recalculate the hash of a board every time we need it. This wastes to much computation time. We can maintain it in MakeMove. This is the advantage of using Zobrist after all. Currently we recalculate the hash of a board every time we need it. This wastes to much computation time. We can maintain it in MakeMove. This is the advantage of using Zobrist after all. Jun 24, 2009 issue 4 (Create config file for settings) reported by valentin.stanciu - We should use a file like "blitzkibitz.cfg" to store parameters for the engine. For example MAX_DEPTH should be taken from there. We should use a file like "blitzkibitz.cfg" to store parameters for the engine. For example MAX_DEPTH should be taken from there. Jun 21, 2009 BlitzKibitz5.exe (BlitzKibitz running at 5-6 plies (faster, but not as smart)) file uploaded by valentin.stanciu - Labels: Type-Executable OpSys-Windows Labels: Type-Executable OpSys-Windows Jun 21, 2009 issue 2 (Translate all internal strings to English) Labels changed by valentin.stanciu - Labels: Maintainability Labels: Maintainability Jun 21, 2009 issue 3 (King valid move performace problem) reported by valentin.stanciu - Whenever a king can move to an opponent attacked square it take a lot more time to process. To reproduce: set the maximum depth and maximum limit of depth to 1 and place a piece so that it attacks squares where the engine's king can move. You will see it takes a few seconds to decide on the move. At 1 depth the decision should be instantaneous! Whenever a king can move to an opponent attacked square it take a lot more time to process. To reproduce: set the maximum depth and maximum limit of depth to 1 and place a piece so that it attacks squares where the engine's king can move. You will see it takes a few seconds to decide on the move. At 1 depth the decision should be instantaneous! Jun 21, 2009 issue 2 (Translate all internal strings to English) Owner changed by valentin.stanciu - Owner: --- Owner: --- Jun 21, 2009 issue 2 (Translate all internal strings to English) reported by valentin.stanciu - We must translate all internal text to English. This includes comments or function documentation as well as internal variable names and macros. ex: The Move class should not have a member "sah". It should be renamed "check". We must translate all internal text to English. This includes comments or function documentation as well as internal variable names and macros. ex: The Move class should not have a member "sah". It should be renamed "check". Jun 21, 2009 issue 1 (Properly deal with draw by repetition) reported by valentin.stanciu - When 2 chess engines battle against each other, they might enter a board configuration where they continue to repeat moves in a circular manner, thus the board never advancing. These cases can spawn for more than 2 board states, for example: A->B->C->A (3 board states) These cases need to be properly analyzed and dealt with. Either by making a move that exists the cycle or by offering a draw by repetition. When 2 chess engines battle against each other, they might enter a board configuration where they continue to repeat moves in a circular manner, thus the board never advancing. These cases can spawn for more than 2 board states, for example: A->B->C->A (3 board states) These cases need to be properly analyzed and dealt with. Either by making a move that exists the cycle or by offering a draw by repetition. Jun 21, 2009 BlitzKibitz.exe (Windows x86 executable) file uploaded by valentin.stanciu - Labels: Type-Executable OpSys-Windows Labels: Type-Executable OpSys-Windows Jun 21, 2009 r4 (- translated some text to English - xboard play bugfix - bet...) committed by valentin.stanciu - - translated some text to English - xboard play bugfix - better implementation of xboard protocol (can now properly end a game) - translated some text to English - xboard play bugfix - better implementation of xboard protocol (can now properly end a game) Jun 21, 2009 BlitzKibitz.zip (Windows x86 executable with opening tables) file uploaded by valentin.stanciu - Labels: Type-Archive OpSys-Windows Labels: Type-Archive OpSys-Windows Jun 21, 2009 BlitzKibitz.exe (Windows x86 executable) file uploaded by valentin.stanciu - Labels: Type-Executable OpSys-Windows Labels: Type-Executable OpSys-Windows Jun 21, 2009 r3 (Can now play through winboard even without the openings.bk f...) committed by valentin.stanciu - Can now play through winboard even without the openings.bk file Can now play through winboard even without the openings.bk file Jun 21, 2009 BlitzKibitz.exe (Windows x86 executable) file uploaded by valentin.stanciu - Labels: Type-Executable OpSys-Windows Labels: Type-Executable OpSys-Windows Jun 21, 2009 openings.zip (Opening tables) file uploaded by valentin.stanciu Jun 21, 2009 BlitzKibitz.zip (Windows x86 executable with opening tables) file uploaded by valentin.stanciu - Labels: Type-Archive OpSys-Windows Featured Labels: Type-Archive OpSys-Windows Featured Jun 19, 2009 r2 (Initial commit) committed by valentin...@gmail.com - Initial commit Initial commit Jun 19, 2009 Project blitzkibitz created by valentin.stanciu - C/C++ chess engine C/C++ chess engine

issue 8 (Automatic speed testing) reported by valentin.stanciu - Implement some way to measure the overall speed of the program. While partial component testing can be done before we integrate it into the engine, sometimes attenuating factors come into play and we need to measure if it makes a speed improvement or not (and how bit it is) We could make the tester evaluate each intermediate board of a recorded game and report the summed up time it took to evaluate the whole game. But don't let the engine make a move. This way, the recorded game stays the same even if we change the scoring logic. Just evaluate, make recorded move, evaluate, make recorded move, etc.. We can enhance it to remember the time of each move and create fancy statistics. But later on.

Implement some way to measure the overall speed of the program. While partial component testing can be done before we integrate it into the engine, sometimes attenuating factors come into play and we need to measure if it makes a speed improvement or not (and how bit it is) We could make the tester evaluate each intermediate board of a recorded game and report the summed up time it took to evaluate the whole game. But don't let the engine make a move. This way, the recorded game stays the same even if we change the scoring logic. Just evaluate, make recorded move, evaluate, make recorded move, etc.. We can enhance it to remember the time of each move and create fancy statistics. But later on.

Earlier this year