Arasan 23.0

[jdart]

Trying to compile it under Windows using the makefile I get this message: Makefile:41: *** missing separator. Stop.

I searched the web and found that this could be caused by using spaces in the makefile instead of tabs. However, whatever I do the message remains.

There are two Makefiles. Makefile.win in Github (= Makefile in the Windows source zip) is for Microsoft NMAKE. Makefile in the repo and in the Linux/Mac source tarball is for Gnu Make. They are both working for me.

[Carlos777]

[The problem for non AVX2 compiles remains the same, only 86K NPS. And the 32 bit version is even faster 101K NPS.

That's a known issue: no SIMD support for non-AVX2 architectures. On the to-do list. For right now I just recommend disabling NNUE if not on AVX2.

Very old HW here (Phenom II x6)
I'll keep using the classic version for my tournaments. This is single thread, ponder off, 1'+0.5", random openings (limited to 5 moves,) both versions play the same position with reversed colors.

Code: Select all

                             Score     (Tie)  12345678901234567890123456789012345678901234567890
-------------------------------------------------------------------------------------------------
 1: Arasan 23.0.1 Classic  36.0 / 50  504.00  ==1=1===111==1100=100=1111111=0110110111=1=111111=  (+29 -7 =14)
 2: Arasan 23.0.1 NNUE     14.0 / 50  504.00  ==0=0===000==0011=011=0000000=1001001000=0=000000=  (+7 -29 =14)
-------------------------------------------------------------------------------------------------
50 games: +19 -17 =14

[jdart]

I searched the web and found that this could be caused by using spaces in the makefile instead of tabs. However, whatever I do the message remains.

You are using which compiler? and which Makefile?

[Gabor Szots]

I searched the web and found that this could be caused by using spaces in the makefile instead of tabs. However, whatever I do the message remains.

You are using which compiler? and which Makefile?

I use the windows makefile. I first tried GCC 10.3 (mingw32-make.exe) then NMAKE. With the former I got that missing separator message, with the latter I get this:

NMAKE : fatal error U1077: '"c:\Program Files (x86)\Microsoft Visual Studio\2019\BuildTools\VC\Tools\MSVC\14.29.30133\bin\Hostx64\x64\cl.EXE"' : return code '0x2'
Stop.

[jdart]

I don't know why NMAKE wouldn't work. I use it all the time. I don't think the supplied Makefile is defective.

Note per the build instructions:

Arasan can also be built using the Cygwin development tools, which simulate a Linux environment under Windows. Use the Linux source package for building with Cygwin. I do not currently recommend or support MingW.

Cygwin does work, but you need to ensure "bc" (from the Math package) is installed.

[Werner]

Nice game by Arasan

[pgn]
[Event "6600K"]
[Site "I56600K"]
[Date "2021.09.08"]
[Round "49"]
[White "Arasan 23.0NN x64 1CPU"]
[Black "Seer 2.3.0NN x64 1CPU"]
[Result "1-0"]
[ECO "C19"]
[PlyCount "32"]
[EventDate "2021.??.??"]
[TimeControl "40/480:40/480:40/480"]

1. e4 e6 2. d4 d5 3. Nc3 Bb4 4. e5 c5 5. a3 Bxc3+ 6. bxc3 Ne7 7. a4 {0.12/23 14
} b6 {-0.51/16 9} 8. Bb5+ {0.13/21 18} Bd7 {-0.37/19 9} 9. Qg4 {0.18/23 17}
Bxb5 {0.03/20 9} 10. axb5 {0.26/25 13} O-O {0.04/20 9} 11. Ne2 {0.10/25 15} Qe8
{-0.13/21 27} 12. Bg5 {0.14/25 13} Nf5 {0.00/20 9} 13. Ng3 {0.28/26 18} Qxb5 {
1.73/21 17} 14. Bf6 {11.09/29 14} g6 {-18.56/21 47} 15. Nxf5 {21.25/33 13} Rd8
{-21.16/21 43} 16. Qg5 {#7/12 0} Qc4 {adjudication -64.48/34 7 Arena
Entscheidung} 1-0[/pgn]

[pohl4711]

7000 games testrun of Arasan 23.0.1 avx2 finished. First testrun of Arasan: All 7000 games were OK (no crashes, timelosses), but after each game a warning message by cutechess-cli: "EngineProcess: Process is destroyed while engine is still running" - very annoying.

https://www.sp-cc.de

[peter]

I have just now pushed an update that fixes the issue with the default NNUE option values under UCI.

The console app does output a message on startup saying what net it is loading but it's not prefixed with "info" so may not be visible in the GUI. You should see it in debug output, though.

Thanks for the update but just for feedback, I tried 23.01 in Shredder GUI, not output about net loaded there neither in console (just started the avx2.exe by double click) regards

[jdart]

I tried 23.01 in Shredder GUI, not output about net loaded there neither in console (just started the avx2.exe by double click) regards
Peter.

I have done some experimentation running under the Shredder GUI and I do see the network being loaded in that environment.

However, there are some other issues with Shredder. Games seem to be saved correctly by Shredder but the log output and game history maintained by Arasan don't look right. I have tested mostly under XBoard mode for Arasan. It looks like Shredder has an imperfect implementation of the protocol, for example I don't see "result" being sent to the engine. I'll try UCI too and see if that is better.

[peter]

I'll try UCI too and see if that is better.

Tried UCI only, Stefan Meyer-Kahlen being the father of this protocol has for sure the best implementation of it in his GUI (which for me at the moment ist Deep Shredder 13).
And as for output of NNUE loaded I just know it to show so at all engines, I tried so far.
No message of loaded NNUE in console neither (starting the engine by double click at avx2.exe).
Just to be clear, I don't really mind all of that, it's just for feedback regards