329 lines
13 KiB
Plaintext
329 lines
13 KiB
Plaintext
1. Introduction
|
|
---------------
|
|
|
|
Glaurung is a free UCI chess engine. It is not a complete chess
|
|
program, but requires some UCI compatible GUI (like XBoard with
|
|
PolyGlot, eboard, José, Arena, Sigma Chess, Shredder, Chess Partner,
|
|
or Fritz) in order to be used comfortably. Read the documentation for
|
|
your GUI of choice for information about how to use Glaurung with your
|
|
GUI.
|
|
|
|
Glaurung 2 is a completely rewritten version of Glaurung. Apart from
|
|
the parallel search code, almost no code is shared with Glaurung
|
|
1.2.1, the previous stable version. The new program is clearly
|
|
stronger than the old, but has a less attractive style of play,
|
|
because there are still a few major holes in its evaluation function
|
|
(most notably space and development).
|
|
|
|
This version of Glaurung supports up to 8 CPUs, but has not been
|
|
tested thoroughly with more than 2. The program tries to detect the
|
|
number of CPUs on your computer and set the number of search threads
|
|
accordingly, but please be aware that the detection is not always
|
|
correct. It is therefore recommended to inspect the value of the
|
|
"Threads" UCI parameter, and to make sure it equals the number of CPU
|
|
cores on your computer.
|
|
|
|
|
|
2. Files
|
|
--------
|
|
|
|
This distribution of Glaurung consists of the following files:
|
|
|
|
* Readme.txt, the file you are currently reading.
|
|
|
|
* Copying.txt, a text file containing the GNU General Public
|
|
License.
|
|
|
|
* src/, a subdirectory containing the full source code, including a
|
|
Makefile that can be used to compile Glaurung on Unix-like
|
|
systems. For further information about how to compile Glaurung
|
|
yourself, read section 4 below.
|
|
|
|
* MacOSX/, a subdirectory containing excutables for Apple Macintosh
|
|
computers running Mac OS X 10.4 (Tiger) and newer. There are two
|
|
executables, one for OS X 10.4, and one for OS X 10.5. The
|
|
executable for OS X 10.4 will work in 10.5 as well, but the one
|
|
for 10.5 is faster.
|
|
|
|
* LinuxX86/, a subdirectory containing 32-bit and 64-bit x86 GNU/Linux
|
|
executables.
|
|
|
|
* Windows/, a subdirectory containing 32-bit and 64-bit Windows
|
|
executables.
|
|
|
|
* polyglot.ini, for using Glaurung with Fabien Letouzey's PolyGlot
|
|
adapter.
|
|
|
|
|
|
3. Opening books
|
|
----------------
|
|
|
|
This version of Glaurung has experimental support for PolyGlot opening
|
|
books. For information about how to create such books, consult the
|
|
PolyGlot documentation. The book file can be selected by setting the
|
|
UCI parameter "Book File".
|
|
|
|
A book file contributed by Salvo Spitaleri can be found on the
|
|
Glaurung web page.
|
|
|
|
|
|
4. Compiling it yourself
|
|
------------------------
|
|
|
|
On Unix-like systems, it should usually be possible to compile
|
|
Glaurung directly from the source code with the included Makefile.
|
|
The exception is computer with big-endian CPUs, like PowerPC
|
|
Macintoshes. Some of the bitboard routines in the current version of
|
|
Glaurung are endianness-sensitive, and won't work on a big-endian CPU.
|
|
Ensuring that the line with #define USE_32BIT_ATTACKS" near the top
|
|
of bitboard.h is commented out should solve this problem.
|
|
Commenting out the line with "#define USE_32BIT_ATTACKS" near the
|
|
|
|
There is also a problem with compiling Glaurung on certain 64-bit
|
|
systems, regardless of the endianness. If Glaurung segfaults
|
|
immediately after startup, try to comment out the line with
|
|
"#define USE_FOLDED_BITSCAN" near the beginning of bitboard.h and
|
|
recompile.
|
|
|
|
Finally, even if Glaurung does work without any changes on your
|
|
computer, it might be possible to improve the performance by changing
|
|
some of the #define directives in bitboard.h. The default settings
|
|
are optimized for 64-bit CPUs. On 32-bit CPUs, it is probably better
|
|
to switch on USE_32BIT_ATTACKS, and to use BITCOUNT_SWAR_32 instead of
|
|
BITCOUNT_SWAR_64. For computers with very little memory (like
|
|
handheld devices), it is possible to conserve memory by defining
|
|
USE_COMPACT_ROOK_ATTACKS.
|
|
|
|
|
|
|
|
5. History
|
|
----------
|
|
|
|
2007-05-06: Glaurung 2 - epsilon
|
|
--------------------------------
|
|
|
|
The first public release, and the first version of my new program
|
|
which is able to match the old Glaurung 1.2.1 on a single CPU. Lots
|
|
of features and chess knowledge is still missing.
|
|
|
|
2007-05-10: Glaurung 2 - epsilon/2
|
|
----------------------------------
|
|
|
|
This version is very close to 2 - epsilon. The major changes are:
|
|
|
|
* A number of compatibility problems which appeared when trying to
|
|
compile Glaurung 2 - epsilon on various operating systems and CPUs
|
|
have been solved.
|
|
|
|
* Fixed a major bug in the detection of rooks trapped inside a
|
|
friendly king.
|
|
|
|
* Added knowledge about several types of drawn endgames.
|
|
|
|
* Fixed a few FRC related bugs. FRC now works, but because of
|
|
serious holes in the evaluation function the program plays very
|
|
badly.
|
|
|
|
* A slightly more sophisticated king safety evaluation.
|
|
|
|
2007-06-07: Glaurung 2 - epsilon/3
|
|
----------------------------------
|
|
|
|
The first public version with support for multiple CPUs. Unless you
|
|
have a dual-core (or better) computer, use Glaurung with a PolyGlot
|
|
book, or runs games with ponder on, you may want to skip this version,
|
|
which is almost certainly no stronger than 2 - epsilon/2 when running
|
|
on a single CPU. The main changes compared to the previous version
|
|
are:
|
|
|
|
* Parallel search, with support for 1-4 CPUs. The program currently
|
|
always allocates a separate pawn hash table and material hash
|
|
table for four threads, which is a pure waste of RAM if your
|
|
computer has just a single CPU. This will be fixed in a future
|
|
version.
|
|
|
|
* Fixed a bug in book randomization. When using Polyglot books, the
|
|
previous version would always select exactly the same move in the
|
|
same position after a restart of the program. Thanks to Pavel
|
|
Háse for pointing this out.
|
|
|
|
* Fixed a UCI pondering bug: Glaurung no longer instantly prints its
|
|
best move when the maximum depth is reached during a ponder
|
|
search, as the previous version did. According to the UCI
|
|
protocol, it is not allowed to print the best move before the
|
|
engine has received the "stop" or "quit" command.
|
|
|
|
* Additional search information: The new version displays hash
|
|
saturation and the current line(s) of search.
|
|
|
|
* Several minor bug fixes and optimizations in the search and
|
|
evaluation.
|
|
|
|
2007-06-08: Glaurung 2 - epsilon/4
|
|
----------------------------------
|
|
|
|
A bugfix release, with only a single important change:
|
|
|
|
* Fixed a very serious pondering bug. As pointed out by Marc
|
|
Lacrosse, the previous version would lose on time in almost every
|
|
single game with pondering enabled. The new version handles
|
|
pondering correctly (or so I hope). When playing with ponder
|
|
off, the new version is identical to version 2 - epsilon/3.
|
|
|
|
2007-06-25: Glaurung 2 - epsilon/5
|
|
----------------------------------
|
|
|
|
Another minor update, including the following improvements and bug
|
|
fixes:
|
|
|
|
* As Werner Schüle discovered, the previous version would sometimes
|
|
stop thinking and lose on time right before delivering checkmate
|
|
(which is of course a very unfortunate moment to lose on time).
|
|
I haven't been able to reproduce Werner's problem on my computer
|
|
(probably because I run a different OS), but I have fixed the bug
|
|
which I suspect caused the time losses. I hope the time losses
|
|
will no longer occur with 2 - epsilon/5.
|
|
|
|
* The program is now slightly less resource-hungry on computers
|
|
with less than 4 CPU cores: The previous version would always
|
|
allocated separate pawn and material hash tables for four
|
|
threads, even when running on a single-core CPU. The new version
|
|
only allocates pawn and material hash tables for the threads
|
|
which are actually used.
|
|
|
|
* A minor reorganization of the memory layout has made the parallel
|
|
search about 10% more efficient (at least on my computer, but the
|
|
results are likely to vary considerably on different systems).
|
|
|
|
* The Intel Mac OS X binary is much faster than before, thanks to
|
|
the Intel C++ compiler (previous versions were compiled with
|
|
GCC).
|
|
|
|
* A few other very minor bug fixes and enhancements.
|
|
|
|
2007-11-21: Glaurung 2.0
|
|
------------------------
|
|
|
|
The first stable (or so I hope) and feature-complete version of
|
|
Glaurung 2. The following are the main changes compared to the
|
|
previous version:
|
|
|
|
* The license has been changed from GPL version 2 to GPL version 3.
|
|
|
|
* MultiPV mode.
|
|
|
|
* Support for the "searchmoves" option in the UCI "go" command.
|
|
This means that it is possible to ask Glaurung to exclude some
|
|
moves from its analysis, or to restrict its analysis to just a
|
|
handful of moves selected by the user. This feature must also be
|
|
supported by the GUI under which Glaurung is run. Glaurung's own
|
|
GUI does currently not support this feature.
|
|
|
|
* Chess960 support now works. The program still plays this game
|
|
very badly, because of lack of opening knowledge.
|
|
|
|
* Much more aggressive pruning in the last few plies of the main
|
|
search.
|
|
|
|
* Somewhat better scaling on multi-CPU systems, and support for up
|
|
to 8 CPUs.
|
|
|
|
* Lots of new UCI parameters.
|
|
|
|
* Improved time managment, especially in games with pondering on
|
|
(i.e. when the engine is allowed to think when it's the
|
|
opponent's turn to move).
|
|
|
|
* Some evaluation improvements, and some new basic endgame
|
|
patterns.
|
|
|
|
* The program should no longer crash if the game lasts longer than
|
|
1000 plies.
|
|
|
|
* Many minor bug fixes and other tiny improvements throughout the
|
|
code.
|
|
|
|
* More generously commented code, and numerous cosmetic changes in
|
|
coding style.
|
|
|
|
2007-11-22: Glaurung 2.0.1
|
|
--------------------------
|
|
|
|
* Fixed (or so I hope) a bug which would occasionally cause one of
|
|
the search threads to get stuck forever in its idle loop.
|
|
|
|
2008-05-14: Glaurung 2.1
|
|
------------------------
|
|
|
|
This version contains far too many changes to list them all, but most
|
|
of them are minor and cosmetic. The most important and noticable
|
|
changes are a lot of new UCI parameters, and many improvements in the
|
|
evaluation function. The highlights are:
|
|
|
|
* Extensive changes in the evaluation function. The addition of
|
|
king safety is the most important improvement, but there are also
|
|
numerous little improvements elsewhere in the evaluation. There
|
|
is still much work left to do in the evaluation function, though.
|
|
Space and development are still missing, and the tuning is likely
|
|
to be very poor. Currently, the program is optimized for an
|
|
entertaining style rather than maximum strength.
|
|
|
|
* More accurate forward pruning. The previous version used the
|
|
null move refutation move to improve the pruning accuracy by
|
|
means of a very simple trick: It did not allow pruning of any
|
|
moves with the piece captured by the null move refutation move.
|
|
In Glaurung 2.1, this has been enhanced: It does not allow
|
|
pruning of moves which defend the destination square of the null
|
|
move refutation move, nor of moves which block the ray of the
|
|
piece in the case that the moving piece in the null move
|
|
refutation move is a slider.
|
|
|
|
* More conservative use of LMR at PV nodes. The previous version
|
|
searched the first 6 moves with full depth, 2.1 by default
|
|
searches the first 14 moves with full depth (but there is a new
|
|
UCI parameter for configuring this). I am not at all sure
|
|
whether this is an improvement. More thorough testing is
|
|
required.
|
|
|
|
* Feedback from the evaluation to the search. The search passes an
|
|
object of type 'EvalInfo' to the eval, and the eval fills this
|
|
struct with various potentially useful information (like the sets
|
|
of squares attacked by each piece type, the middle game and
|
|
endgame components of the eval, etc.). At the moment, almost
|
|
none of this information is actually used by the search. The
|
|
only exception is that the evaluation function is now used to
|
|
adjust the futility pruning margin in the quiescence search.
|
|
|
|
* Less extensions. This hurts the programs performance a lot in most
|
|
test suites, but I hope it improves the branching factor in deep
|
|
searches.
|
|
|
|
* A very long list of new UCI parameters, especially for tuning the
|
|
evaluation.
|
|
|
|
|
|
6. Terms of use
|
|
---------------
|
|
|
|
Glaurung is free, and distributed under the GNU General Public License
|
|
(GPL). Essentially, this means that you are free to do almost exactly
|
|
what you want with the program, including distributing it among your
|
|
friends, making it available for download from your web site, selling
|
|
it (either by itself or as part of some bigger software package), or
|
|
using it as the starting point for a software project of your own.
|
|
|
|
The only real limitation is that whenever you distribute Glaurung in
|
|
some way, you must always include the full source code, or a pointer
|
|
to where the source code can be found. If you make any changes to the
|
|
source code, these changes must also be made available under the GPL.
|
|
|
|
For full details, read the copy of the GPL found in the file named
|
|
Copying.txt.
|
|
|
|
|
|
7. Feedback
|
|
-----------
|
|
|
|
The author's e-mail address is tord@glaurungchess.com
|
|
|