larsaur/arcanum

UCI chess engine

arcanum's Issues

Strength difference

Hi Lars,

In your estimation what is the difference in strength using-not using NNUE?

NNUE usage

If the NNUE file is not present the engine crashes even if UseNNUE is false.

If it is present the engine plays OK but how can I check whether NNUE is indeed used or not?

Gabor Szots
CCRL testing group

Missing dll files in Arcanum

I'm hoping to include Arcanum in my Amateur Series, but it is missing dll files.

Graham (CCRL)

Version number display

I suggest including version number in the 'id name Arcanum' line.

uci
id name Arcanum
id author Lars Murud Aurud
option name Hash type spin default 32 min 1 max 8196
option name ClearHash type button
option name UseNNUE type check default true
uciok

Arcanum 2.0 lose on time

Thanks for sharing Arcanum!
Arcanum 1.12 works fine but Arcanum 2.0 lose on time!

GUI: Arena 3.5.1
OS: Windows 10 Pro
TimeControl: 10s+0.1 per game

can not compile on Linux

i tried to compile your v2.0 code on Linux .. it seems a valid binary is created, but the program crashes : Segmentation fault (core dumped). I did some research and tests, changing some concerning code lines, but i failed .. however, the program runs without problems when i use a debugger !? I will explain below.

at first i encountered a basic error :

src/tuning/fenGen.cpp:1:10: fatal error: tuning/fengen.hpp: No such file or directory
    1 | #include <tuning/fengen.hpp>
      |          ^~~~~~~~~~~~~~~~~~~
compilation terminated.
make: *** [makefile:62: build/src/tuning/fenGen.o] Error 1
make: *** Waiting for unfinished jobs....
src/uci.cpp:10:10: fatal error: tuning/fengen.hpp: No such file or directory
   10 | #include <tuning/fengen.hpp>
      |          ^~~~~~~~~~~~~~~~~~~
compilation terminated.
make: *** [makefile:62: build/src/uci.o] Error 1

it seems the file fengen.hpp is expected, but it's called fenGen.hpp (mark the capital letter G) .. so i renamed that file into all lowercase letters and now compilation succeeds : on Windows, filenames are case-insensitive (isn't it?), so you won't notice this simple mistake .. i'll suggest to always use lowercase letters for ALL filenames.

when running in CuteChess GUI the program crashes !
executing in terminal gives this output :

$ ./Arcanum.exe 
uci
id name Arcanum dev_build
id author Lars Murud Aurud
option name Hash type spin default 32 min 0 max 8196
option name ClearHash type button
option name SyzygyPath type string default <empty>
option name NNUEPath type string default arcanum1.fnnue
uciok
ucinewgame
position startpos
go wtime 423000 btime 423000 winc 3000 binc 3000
info depth 1 nodes 22 score cp 136 pv d2d4
info depth 2 nodes 104 score cp 78 pv d2d4 e7e6
info depth 3 time 1 nodes 747 score cp 154 nps 747000 pv d2d4 d7d5 g1f3
Segmentation fault (core dumped)

however, when i use a debugger (gdb or lldb) this does NOT happen !? :

$ gdb ./Arcanum.exe 
GNU gdb (Ubuntu 12.1-0ubuntu1~22.04) 12.1
Copyright (C) 2022 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.
Type "show copying" and "show warranty" for details.
This GDB was configured as "x86_64-linux-gnu".
Type "show configuration" for configuration details.
For bug reporting instructions, please see:
<https://www.gnu.org/software/gdb/bugs/>.
Find the GDB manual and other documentation resources online at:
    <http://www.gnu.org/software/gdb/documentation/>.

For help, type "help".
Type "apropos word" to search for commands related to "word"...
Reading symbols from ./Arcanum.exe...
(No debugging symbols found in ./Arcanum.exe)
(gdb) run
Starting program: /home/roelof/Compiled/Arcanum-2.0/build/Arcanum.exe 
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
uci
id name Arcanum dev_build
id author Lars Murud Aurud
option name Hash type spin default 32 min 0 max 8196
option name ClearHash type button
option name SyzygyPath type string default <empty>
option name NNUEPath type string default arcanum1.fnnue
uciok
ucinewgame
position startpos
go wtime 423000 btime 423000 winc 3000 binc 3000
[New Thread 0x7ffff5f35640 (LWP 9626)]
info depth 1 nodes 22 score cp 136 pv d2d4
info depth 2 nodes 104 score cp 78 pv d2d4 e7e6
info depth 3 time 2 nodes 747 score cp 154 nps 373500 pv d2d4 d7d5 g1f3
info depth 4 time 8 nodes 2111 score cp 71 nps 263875 pv d2d4 d7d5 b1c3 c7c6
info depth 5 time 32 nodes 7821 score cp 139 hashfull 1 nps 244406 pv b1c3 d7d5 e2e4 d5d4 c3d5
info depth 6 time 66 nodes 41473 score cp 80 hashfull 9 nps 628378 pv e2e4 d7d5 e4e5 a7a5 d2d4 b8c6
info depth 7 time 113 nodes 100084 score cp 129 hashfull 22 nps 885699 pv e2e4 d7d5 e4e5 c8f5 b1c3 d5d4 c3b5
info depth 8 time 170 nodes 168267 score cp 109 hashfull 39 nps 989805 pv e2e4 d7d5 e4e5 c8f5 d2d4 c7c6 f1d3 g8h6
info depth 9 time 338 nodes 365260 score cp 110 hashfull 83 nps 1080650 pv e2e4 d7d5 e4e5 c8f5 d2d4 b8c6 b1c3 c6b4 f1d3
info depth 10 time 845 nodes 937603 score cp 100 hashfull 217 nps 1109589 pv e2e4 d7d5 e4d5 g8f6 d2d4 e7e6 d5e6 c8e6 b1c3 f8b4
info depth 11 time 1612 nodes 1827542 score cp 100 hashfull 382 nps 1133710 pv e2e4 d7d5 e4d5 d8d5 g1f3 a7a6 b1c3 d5e6 d1e2 b8c6 e2e6 c8e6
info depth 12 time 3915 nodes 4362625 score cp 116 hashfull 712 nps 1114335 pv e2e4 d7d5 e4d5 c7c6 d2d4 d8d5 b1c3 d5d6 g1f3 g8f6 f3e5 a7a5
info depth 12 time 12272 nodes 13577894 score cp 93 hashfull 983 nps 1106412 pv e2e4 e7e5 g1f3 b8c6 b1c3 g8f6 f1c4 f6e4 c4f7 e8f7 c3e4 d7d5 e4g5 f7e8 d2d4
bestmove e2e4
[Thread 0x7ffff5f35640 (LWP 9626) exited]

this is weird, we should expect the opposite .. once i had a similar Issue with the Monza chess engine, see mourabitiziyad/Monza-Chess#1 .. here i describe in detail what i found .. it can be a 'race condition' : the debugger causes the execution to be a bit slower, and THEREFOR there's no crash .. regarding the Monza engine i found a solution by adjusting the TT size, but tinkering the concerning code didn't help .. i also adjusted the makefile in many ways, just to see the difference, but to no avail.

compiling v1.12 went without such error, i hope you can solve this Issue and/or reproduce it .. i guess my explanation could help.
NOTE: i have only 8 Gb RAM, normally i set max 256 Mb Hash for any chess engine and max 2 threads.

[ i'm on Xubuntu 22.04 ]