Slow Chess Blitz Classic

[Roland Chastain]

Made a quick engines tournament with CuteChess Cli. I noticed two little things. First, Slow Chess "loses" several games because of the missing "q" character in promotion moves, which (unfortunately) is considered by CuteChess as an illegal move. Second thing, CuteChess repeats often the warning "Illegal PV move <tt> from SlowChessClassic 1.4". Both issues happens with many engines. The warning can simply be ignored. The promotion move declared illegal is more embarrassing. Maybe you could do something for that?

[Guenther]

I'll compile a no popcount tonight.
The UCI pv might not be correctly formatted will check that. For UCI support I tested using Arena, unfortunately don't have the Shredder or Fritz GUI to test, but can find some other UCI tests. Otherwise I used Winboard for any regular testing, but probably every engine is UCI now.
If any more issues let me know can look into them tonight too.
I am curious where it would end up on rating lists.

Thanks Jon. The no popcount compile now works here.

As I already used SlowChess in the past in RWBC I knew how to set hash in the 'Schess.dft' config file,
but at a quick look I found no hint for new users neither in the readme nor at your site.
I guess it should be mentioned somewhere that e.g. '24' corresponds to 192MB hash and '25' to 384MB and so on.

I also noticed that setting inside the GUI is still at max of 192MB (the old times), so inside the GUI you cannot
set more that that. Knowing what I wrote above I tried 384 and 768MB and they worked for SlowChess.

[jonk]

Roland : yikes, my UCI sendmove was checking the the src sq piece after the move instead of before and thinking it wasn't a pawn move, so that was making it not send promotion type. Just now I updated the exe in the 1.4 zip on website to fix this issue, so if you re-download it should work now. Since it sounds like CuteChess Cli is stricter test of UCI than Arena, I downloaded it and ran a quick 20 game match without issue. I didn't notice it saying illegal <tt> in pv, so maybe that was already fixed.

Guenther : Good to hear that the no popcount works, will keep it around then. I will add higher hash sizes in next version. And look into more UCI configuration options and maybe redo the schess.dft config format, I'm actually not sure what the numbers in it do anymore =)

[Gabor Szots]

Also not sure why not finding opening book with arena, that means probably not finding bitbases either. Expects working directory same as exe, and sub-directories same as in the .zip with bitbases and openingBook.

Thanks for the hint, Jon. The problem was indeed that I did not keep the directory structure of the download package. Now it's OK. Also, the PV is displayed during search with the updated exe.

[Guenther]

...

Guenther : Good to hear that the no popcount works, will keep it around then. I will add higher hash sizes in next version. And look into more UCI configuration options and maybe redo the schess.dft config format, I'm actually not sure what the numbers in it do anymore =)

Jon, I found another issue, Cutechess doesn't like the SAN output (only tried it as xboard engine).
Is there a way to disable it or do you need to change it in the source code?

Anyhow I suggest testing the great CuteChess GUI (and cmd interface) too besides Winboard/Xboard.
https://github.com/cutechess/cutechess
(latest compilation is from 2017 but source code is still developed and newer compiles can be found in the forum
or you can compile yourself - never gave me any trouble)
Those are the only GUIs I prefer as a (professional) software tester.

Code: Select all

[Event "Test_20190905"]
[Site "RWBC Win7U-64 Q8200 2.33Ghz Nvidia GT 710"]
[Date "2019.09.05"]
[Round "1"]
[White "SlowChessBlitz_Classic14-64"]
[Black "Shield_21-64"]
[Result "0-1"]
[ECO "A05"]
[Opening "Reti Opening"]
[PlyCount "6"]
[Termination "illegal move"]
[TimeControl "40/120"]

1. Nf3 {book} Nf6 {book} 2. b3 {book} g6 {book} 3. Bb2 {book}
Bg7 {book, White makes an illegal move: d2-d4} 0-1

[Event "Test_20190905"]
[Site "RWBC Win7U-64 Q8200 2.33Ghz Nvidia GT 710"]
[Date "2019.09.05"]
[Round "1"]
[White "Shield_21-64"]
[Black "SlowChessBlitz_Classic14-64"]
[Result "1-0"]
[ECO "A05"]
[Opening "Reti Opening"]
[PlyCount "7"]
[Termination "illegal move"]
[TimeControl "40/120"]

1. Nf3 {book} Nf6 {book} 2. b3 {book} g6 {book} 3. Bb2 {book} Bg7 {book}
4. g4 {-0.01/16 3.11s, Black makes an illegal move: h7-h6} 1-0

[Event "Test_20190905"]
[Site "RWBC Win7U-64 Q8200 2.33Ghz Nvidia GT 710"]
[Date "2019.09.05"]
[Round "2"]
[White "SlowChessBlitz_Classic14-64"]
[Black "Shield_21-64"]
[Result "0-1"]
[ECO "A45"]
[Opening "Queen's pawn game"]
[PlyCount "6"]
[Termination "illegal move"]
[TimeControl "40/120"]

1. d4 {book} Nf6 {book} 2. g3 {book} e6 {book} 3. Bg2 {book}
c5 {book, White makes an illegal move: Ng1-f3} 0-1

[Event "Test_20190905"]
[Site "RWBC Win7U-64 Q8200 2.33Ghz Nvidia GT 710"]
[Date "2019.09.05"]
[Round "2"]
[White "Shield_21-64"]
[Black "SlowChessBlitz_Classic14-64"]
[Result "1-0"]
[ECO "A45"]
[Opening "Queen's pawn game"]
[PlyCount "7"]
[Termination "illegal move"]
[TimeControl "40/120"]

1. d4 {book} Nf6 {book} 2. g3 {book} e6 {book} 3. Bg2 {book} c5 {book}
4. Nf3 {0.00/15 2.09s, Black makes an illegal move: d7-d5} 1-0

[BrendanJNorman]

Seems really strong and really positional in playing style...reminds me of Schooner's return actually.

I was right. This engine just comprehensively outplayed both Deep Junior 11.2 and Fizbo 1.3.1.

I'd say it's somewhere around 2900 CCRL just from first impressions of its play.

Nice work to add 300 or so Elo to the engine - especially with such an un-computer-like style.

Testing vs my beloved Rodent 4 next.

[jonk]

The 1.4 download will now use the same format for sending Winboard moves as it does UCI. (I downloaded CuteChess earlier after Roland's comment and fixed the uci promo bug he mentioned, but didn't test xboard mode before. Now with latest I was able to do a short test tournament in either mode without issue. )

Somewhere around 2900 sounds like a reasonable estimate to me. I didn't want to guess since results vary so much. (Even against the same program, but Slow Classic seemed to do much better against some programs than others. Which might be from running too few test games or the time controls/setup I used too.)

I wasn't sure if anyone else would consider it un-computer like, I view it as human-like because I hand tuned the eval & search and occasionally adjusted it for single positions from lost games, so I see it as how I'd play if I had computer search super-power. Most of the random ideas I tried seemed successful, but probably only because the starting point was so much weaker than current programs. The strength increase is also partially due to better practices now known ( like crank up late move reductions, futility pruning, move pruning, etc. I still haven't really figured that out I seem to get small fractional improvements in depth instead of the crazy depths of other programs. Oh and singular extension I think helps noticeably too.)

[jonk]

One more fix for cutechess UCI, ponder was broken but should work now in the 1.4 download.( Slow Chess didn't send a ponder check option on startup so it looked like that made cutechess never tell it to go ponder.)

[PK]

I must say I'm impressed. Watched a couple of bullet games between Rodent and SlowChess, and Your engine seems a perfect opponent for Rodent, showing where my brainchild goes wrong. When Rodent attacks, SlowChess keeps its cool. Where Rodent thinks it has a won endgame, SlowChess proves the poor mouse wrong. Just add some search depth, and You will have a nice, calm, +3000 Elo engine.

[Max]

Hi Jonathan,
is there some hope for a mac version of Slow Chess Blitz Classic please?

Using wine is atm no option, because your engine crashed immediately after sending uciok.

Code: Select all

$ wine slow64.exe 
uci
id name SlowChess Blitz Classic 1.4b
id author Jonathan Kreuzer
option name Hash type spin default 24 min 2 max 300
option name Low Priority type check default false
option name Clear Hash type button
option name Ponder type check
option name OwnBook type check default true
option name Debug Log type check default false
uciok
wine: Unhandled page fault on read access to 0x00000020 at address 0x1400440a6 (thread 002e), starting debugger...
Unhandled exception: page fault on read access to 0x00000020 in 64-bit code (0x00000001400440a6).
Register dump:
 rip:00000001400440a6 rsp:0000000004a3fdc0 rbp:0000000004a3ffd0 eflags:00010246 (  R- --  I  Z- -P- )
 rax:0000000000000000 rbx:0000000000000001 rcx:0000000000000000 rdx:0000000140063328
 rsi:000000007bccc268 rdi:000000007bccc268  r8:00000001400e66b0  r9:0000000004a3fd80 r10:000063f6dd6a644d
 r11:0000000000000000 r12:0000000000000000 r13:0000000000000000 r14:0000000000073010 r15:0000000000000000
Stack dump:
0x0000000004a3fdc0:  0000000000000000 0000000000000000
0x0000000004a3fdd0:  0000000000000005 0000000000000005
0x0000000004a3fde0:  0000000000073010 0000000140036486
0x0000000004a3fdf0:  0000000000000000 0000000000000000
0x0000000004a3fe00:  0000000000073010 0000000000000000
0x0000000004a3fe10:  0000000140036458 000000007bc70db0
0x0000000004a3fe20:  0000000000000000 0000000000000000
0x0000000004a3fe30:  0000000000000000 0000000000000000
0x0000000004a3fe40:  0000000000000000 ffffffffffffffff
0x0000000004a3fe50:  000000007bc8ed77 000000007bc323de
0x0000000004a3fe60:  0000000000000000 0000000140036458
0x0000000004a3fe70:  0000000004a3ffd0 0000000004a3fe20
Backtrace:
=>0 0x00000001400440a6 in slow64 (+0x440a6) (0x0000000004a3ffd0)
  1 0x0000000140036486 in slow64 (+0x36485) (0x0000000004a3ffd0)
0x00000001400440a6: movq	0x0000000000000020(%rax),%rcx
Modules:
Module	Address					Debug info	Name (29 modules)
PE	        681d0000-        681d4000	Deferred        wsock32
PE	        6cb10000-        6cb14000	Deferred        ws2_32
PE	        6cb50000-        6cb54000	Deferred        iphlpapi
PE	        6cb80000-        6cb84000	Deferred        advapi32
PE	        6cc20000-        6ccfa000	Deferred        user32
PE	        6ce80000-        6ce84000	Deferred        gdi32
PE	        6cff0000-        6cff4000	Deferred        version
PE	        6d010000-        6d023000	Deferred        comdlg32
PE	        6d0a0000-        6d90b000	Deferred        shell32
PE	        6da80000-        6da84000	Deferred        shlwapi
PE	        6db30000-        6db33000	Deferred        shcore
PE	        6db60000-        6db88000	Deferred        ole32
PE	        6dd40000-        6dd44000	Deferred        rpcrt4
PE	        6de00000-        6de40000	Deferred        comctl32
PE	        6df60000-        6df63000	Deferred        usp10
PE	        6dfa0000-        6dfa4000	Deferred        imm32
PE	        6dfd0000-        6dfd6000	Deferred        winspool
PE	        6e0e0000-        6e0e4000	Deferred        winemac
PE	        6e6a0000-        6e6a4000	Deferred        uxtheme
PE	        6e7e0000-        6e7e4000	Deferred        localspl
PE	        6e800000-        6e803000	Deferred        spoolss
PE	        6e820000-        6e824000	Deferred        wineps
PE	        6e970000-        6e973000	Deferred        api-ms-win-core-synch-l1-2-0
PE	        6e980000-        6e983000	Deferred        api-ms-win-core-fibers-l1-1-1
PE	        6e990000-        6e993000	Deferred        api-ms-win-core-localization-l1-2-1
PE	        7b410000-        7b46a000	Deferred        kernel32
PE	        7bc10000-        7bc14000	Deferred        ntdll
ELF	        7c400000-        7c404000	Deferred        <wine-loader>
PE	       140000000-       140668000	Export          slow64
Threads:
process  tid      prio (all id:s are in hex)
0000000e services.exe
	00000020    0
	0000001b    0
	00000013    0
	00000010    0
	0000000f    0
00000011 winedevice.exe
	00000018    0
	00000017    0
	00000016    0
	00000012    0
00000019 plugplay.exe
	0000001d    0
	0000001c    0
	0000001a    0
0000001e winedevice.exe
	00000028    0
	00000027    0
	00000026    0
	00000025    0
	00000022    0
	00000021    0
	0000001f    0
00000023 explorer.exe
	0000002b    0
	0000002a    0
	00000029    0
	00000024    0
0000002c (D) Z:\Users\max\Downloads\SlowChessClassic-1.4\slow64.exe
	0000002e    0 <==
	0000002d    0
System information:
    Wine build: wine-4.0.2
    Platform: x86_64
    Version: Windows 7
    Host system: Darwin
    Host version: 17.7.0