Roce 0.0380

View previous topic View next topic Go down

Roce 0.0380

Post  Patrick Buchmann on Tue Jul 08, 2008 7:55 pm

Hi,

I had problems with the new Roce 0.0380.

When it plays vs a human, it works fine, but in tournament it does'nt play a single move after the last books move.

My system: PIV, XP SP2.

Regards,
Patrick

Patrick Buchmann

Number of posts : 30
Registration date : 2008-05-22

View user profile http://pagesperso-orange.fr/lefouduroi/computerchess.htm

Back to top Go down

Re: Roce 0.0380

Post  Admin on Tue Jul 08, 2008 9:43 pm



Patrick Buchmann wrote:Hi,

I had problems with the new Roce 0.0380.

When it plays vs a human, it works fine, but in tournament it does'nt play a single move after the last books move.

My system: PIV, XP SP2.

Regards,
Patrick

Hi Patrick,

Roce 0.0380 plays fine here vs other engines, using PolyGlot 1.4w2/WB, WinXP.

Best wishes,
Leo

Admin
Admin

Number of posts : 19
Registration date : 2008-05-17

View user profile http://wbec-ridderkerk.forumotion.com

Back to top Go down

Re: Roce 0.0380

Post  Patrick Buchmann on Tue Jul 08, 2008 10:08 pm

Admin wrote:


Hi Patrick,

Roce 0.0380 plays fine here vs other engines, using PolyGlot 1.4w2/WB, WinXP.

Best wishes,
Leo

Hi Leo,

I would use Poce 0.0380 as UCI Engine.

Thanks,
Patrick

Patrick Buchmann

Number of posts : 30
Registration date : 2008-05-22

View user profile http://pagesperso-orange.fr/lefouduroi/computerchess.htm

Back to top Go down

Re: Roce 0.0380

Post  Roman Hartmann (Roce) on Sat Jul 12, 2008 10:23 pm

Hi Patrick,
I read here and also in your forum that you have some problems with my new version. Trying to figure out what's causing the problems I just installed Arena on a 2nd computer but still it works as expected here. I thought it could be related to my new time allocating method. To make the code more portable I changed all the time related code, maybe that's causing the problems on your computer. None of my computer seems to have any problems though.

Another cause for the problems you have might be that you downloded a buggy version which I uploaded on my homepage by accident. This version was replaced really quickly but still there is a chance that you're testing that version. To see if you have this broken version just doubleclick on it and type analyze in the window and if it doesn't start to calculate it's the broken version you have.

best regards
Roman

PS: I'm testing here a new version with hash-tables (finally) and I'm going to release that really soon as it seems to be about 30-40 points stronger than 0.037. So I won't mind if you (or Leo) don't test 0.038 at all. Still I would like to know what's causing the problems you have.

Roman Hartmann (Roce)

Number of posts : 5
Registration date : 2008-07-12

View user profile http://www.rocechess.ch/rocee.html

Back to top Go down

Re: Roce 0.0380

Post  Patrick Buchmann on Sat Jul 12, 2008 11:25 pm

Roman Hartmann (Roce) wrote:
Another cause for the problems you have might be that you downloded a buggy version which I uploaded on my homepage by accident. This version was replaced really quickly but still there is a chance that you're testing that version. To see if you have this broken version just doubleclick on it and type analyze in the window and if it doesn't start to calculate it's the broken version you have.

Hi Roman,

I think I have the right version (05/07/08 14:25).

As I wrote on my forum, it's very strange. On Fritz GUI and Shredder Classic no problem. But on Arena 1.1, after the last book move, Roce doen't realy start. After 1 minute, it starts (the clock is reinitialized!). And do with each games.

Regards,
Patrick

Patrick Buchmann

Number of posts : 30
Registration date : 2008-05-22

View user profile http://pagesperso-orange.fr/lefouduroi/computerchess.htm

Back to top Go down

Re: Roce 0.0380

Post  Roman Hartmann (Roce) on Sun Jul 13, 2008 2:14 pm

Hi Patrick,
thanks for the additional details. I have to confess that I didn't test with a book yesterday but I did that today (by using the Arena Main-Book from Harry Schnapp).
But still I have no problems here. I tried several time controls but after the book moves it would start immediately to calculate on a move without any delay.
Are you maybe playing with ponder on? That might cause this behaviour as the code for pondering is in alpha-stage. Another reason for the problem could be that the roce.cfg file is not in the same directory as the binary. Roce will read that file by startup and if it's not there issue a warning which might confuse Arena.
Another -rather unrealistic- possibility is that one of the compiler switches I used is causing problems on some CPUs (I think I used -Ox instead of -O2 for the compile).

best regards
Roman

Roman Hartmann (Roce)

Number of posts : 5
Registration date : 2008-07-12

View user profile http://www.rocechess.ch/rocee.html

Back to top Go down

Re: Roce 0.0380

Post  Patrick Buchmann on Sun Jul 13, 2008 2:43 pm

Hi Roman,

I use a PIV, Win XP SP2. I don't use ponder on.

It's very strange, when I play against Roce, it replays immediatly. The problem only occurs in engines tournament.

Regards,
Patrick

Patrick Buchmann

Number of posts : 30
Registration date : 2008-05-22

View user profile http://pagesperso-orange.fr/lefouduroi/computerchess.htm

Back to top Go down

Re: Roce 0.0380

Post  Roman Hartmann (Roce) on Sun Jul 13, 2008 5:30 pm

When changing my time allocating sheme I also implemented some routines to output how much time was allocated. That could be another source for the problems.
I removed that debugging stuff now and maybe you can check if it still behaves strangely (I also used the old compiler switches just in case).

link: http://www.rocechess.ch/roce/roce38x.zip

This compile uses also hash-tables btw (I didn't want to edit my old code).

best regards
Roman

Roman Hartmann (Roce)

Number of posts : 5
Registration date : 2008-07-12

View user profile http://www.rocechess.ch/rocee.html

Back to top Go down

Roce38x

Post  Patrick Buchmann on Sun Jul 13, 2008 10:51 pm

Hallo Roman,

Hat sich nichts verändert.

Hier die log von Arena:

Arena 1.1
6640**-------------NewGame!!!-------------
138078********======== Tournoi 1 ========
139578********Arrête module 1 Zct 3.2451
139921********Arrête module 2 Zappa 1.1
142312********- - - - Appariement 1/64 : Roce 0.0380x - Abrok 5.0 - - - -
142328********Démarre module 1 Roce 0.0380x
142671*1*-----------------------Démarre module 1 Roce 0.0380x------------------------
142671*1*Configured Engine 1 Type: UCI
142812*1*Engine 1 dir: C:\cbase\Engines.UCI\Roce
142843*1*Engine 1 commandline: C:\cbase\Engines.UCI\Roce\roce38x.exe
143109<1:Roce version: 0.0380 - Roman's Own Chess Engine
143109<1:Copyright (C) 2003-2007 Roman Hartmann, Switzerland. All rights reserved.
143109<1:loading: Roce.cfg
143250>1:uci
143296<1:roce:
143343<1:ROCE-Roman's Own Chess Engine
143406<1:rhartmann@bluewin.ch
143406<1:have fun
143453<1:id name Roce 0.0380
143500<1:id author Roman Hartmann
143546<1:option name Hash type spin default 32 min 4 max 512
143562<1:option name useInternalOpeningBook type check default false
143562<1:option name ponder type check default false
143562<1:option name Quiescent Search type check default true
143562<1:option name Logging type check default true
143562<1:option name NullMove Reduction type spin default 2 min 0 max 4
143562<1:uciok
143765>1:setoption name Hash value 64
143765>1:setoption name useInternalOpeningBook value false
143765>1:setoption name ponder value false
143765>1:setoption name Quiescent Search value true
143765>1:setoption name Logging value true
143781>1:setoption name NullMove Reduction value 2
143781>1:isready
144796<1:ok logging
144796<1:k: 0
144796<1:readyok
147625********Démarre module 2 Abrok 5.0
147968*2*-------------------------Démarre module 2 Abrok 5.0-------------------------
147968*2*Configured Engine 2 Type: UCI
148140*2*Engine 2 dir: C:\cbase\Engines.UCI\abrok
148171*2*Engine 2 commandline: C:\cbase\Engines.UCI\abrok\Abrok_5_0.exe hash 32
148468<2:feature done=0
148609>2:uci
148656<2:id name Abrok 5.0
148656<2:id author Roman Korba
148703<2:option name Hash type spin default 32 min 8 max 256
148765<2:option name Ponder type check default true
148812<2:option name OwnBook type check default true
148812<2:uciok
149015>2:setoption name Hash value 64
149015>2:setoption name Ponder value false
149015>2:isready
150062<2:readyok
153906>1:isready
153921<1:readyok
154000>2:isready
154031<2:readyok
154390>1:stop
154453>2:stop
154515>2:setoption name Ponder value false
154515********»»»»Partie 1 de 2: Roce 0.0380x - Abrok 5.0««««
154515**-------------NewGame!!!-------------
155578*1*Start calc, move no: 0
155656*1*HS-Book.abk:
155656*1*c2c4
156109*2*Start calc, move no: 1
156109*2*HS-Book.abk:
156109*2*e7e6
156484*1*Start calc, move no: 2
156500*1*HS-Book.abk:
156500*1*d2d4
156953*2*Start calc, move no: 3
156968*2*HS-Book.abk:
156968*2*f7f5
157546*1*Start calc, move no: 4
157546*1*HS-Book.abk:
157546*1*b1c3
158000*2*Start calc, move no: 5
158015*2*HS-Book.abk:
158015*2*f8b4
158500*1*Start calc, move no: 6
158515*1*HS-Book.abk:
158515*1*c1d2
158843*2*Start calc, move no: 7
158843*2*HS-Book.abk:
158843*2*b7b6
159421*1*Start calc, move no: 8
159437*1*HS-Book.abk:
159437*1*e2e3
159859*2*Start calc, move no: 9
159875*2*HS-Book.abk:
159875*2*c8b7
160328*1*Start calc, move no: 10
160328*1*HS-Book.abk:
160328*1*f2f3
160609*2*Start calc, move no: 11
160625*2*HS-Book.abk:
160625*2*b4d6
160890*1*Start calc, move no: 12
160906*1*HS-Book.abk:
160906*1*g1h3
161187*2*Start calc, move no: 13
161187*2*HS-Book.abk:
161187*2*g8f6
163031*1*Start calc, move no: 14
163046*1*HS-Book.abk:
163046*1*f1d3
163421*2*Start calc, move no: 15
163437*2*HS-Book.abk:
163437*2*b8c6
163906*1*Start calc, move no: 16
163921*1*HS-Book.abk:
163921*1*d1e2
164218*2*Start calc, move no: 17
164234*2*HS-Book.abk:
164234*2*e8g8
164750*1*Start calc, move no: 18
164781*1*HS-Book.abk:
164781*1*a1d1
165062*2*Start calc, move no: 19
165078*2*HS-Book.abk:
165078*2*a7a6
165390*1*Start calc, move no: 20
165421*1*HS-Book.abk:
165421*1*e1g1
165703*2*Start calc, move no: 21
165734*2*HS-Book.abk:
165734*2*d8e8
166031*1*Start calc, move no: 22
166046*1*HS-Book.abk:
166046*1*a2a3
166671*2*Start calc, move no: 23
166687*2*HS-Book.abk:
166687*2*e8h5
166984*1*Start calc, move no: 24
167000*1*HS-Book.abk:
167000*1*b2b4
167359*2*Start calc, move no: 25
167375*2*HS-Book.abk:
167375*2*g7g5
167671*1*Start calc, move no: 26
167687*1*HS-Book.abk:
167687*1*c4c5
167984*2*Start calc, move no: 27
168015*2*HS-Book.abk:
168015*2*d6e7
168312*1*Start calc, move no: 28
168328*1*HS-Book.abk:
168328*1*d3c4
168625*2*Start calc, move no: 29
168656*2*HS-Book.abk:
168656*2*g8h8
168937*1*Start calc, move no: 30
168984>1:ucinewgame
168984>1:isready
169000<1:Illegal move: stopreadyok
219046>1:position startpos moves c2c4 e7e6 d2d4 f7f5 b1c3 f8b4 c1d2 b7b6 e2e3 c8b7 f2f3 b4d6 g1h3 g8f6 f1d3 b8c6 d1e2 e8g8 a1d1 a7a6 e1g1 d8e8 a2a3 e8h5 b2b4 g7g5 c4c5 d6e7 d3c4 g8h8
219046>1:go wtime 300000 btime 300000
219062<1:info depth 1 seldepth 2 nodes 130 nps 1 score cp 129 time 0 pv c5b6 c7b6
219093<1:info depth 2 seldepth 5 nodes 1280 nps 1 score cp 129 time 0 pv c5b6 c7b6
219109<1:info depth 3 seldepth 5 nodes 11514 nps 719625 score cp 115 time 16 pv c5b6 d7d5 b4b5 a6b5 c4b5 c7b6
219156<1:info depth 4 seldepth 9 nodes 66205 nps 607385 score cp 115 time 109 pv c5b6 d7d5 b4b5 a6b5 c4b5 c7b6
219640<1:info depth 5 seldepth 8 nodes 414880 nps 738221 score cp 117 time 562 pv c5b6 d7d5 c4d3 c7b6 h3f2
221312<1:info depth 6 seldepth 6 nodes 1719080 nps 764036 score cp 115 time 2250 pv c5b6 c7b6 c4b3 h8g8 d4d5 c6d8
224687<1:info depth 7 nodes 4415569 score cp 115 nps 784990 currmove c4a6 currmovenumber 4
226000<1:info depth 7 nodes 5540799 score cp 115 nps 796893 currmove e3e4 currmovenumber 7
227781<1:info depth 7 nodes 7047554 score cp 115 nps 808298 currmove b4b5 currmovenumber 10

Ab 169000<1:Illegal move: stopreadyok bleibt die Engine für ca. eine Minute stehen und dann erst geht es weiter.

MfG,
Patrick

Edit: Sorry I write in German pale
The smilie in the logfile is directly writen by the forum What a Face

Patrick Buchmann

Number of posts : 30
Registration date : 2008-05-22

View user profile http://pagesperso-orange.fr/lefouduroi/computerchess.htm

Back to top Go down

Re: Roce 0.0380

Post  Roman Hartmann (Roce) on Sun Jul 13, 2008 11:45 pm

Hi Patrick,
thanks for the debug-listing. The 'Illegal move: stopreadyok' is particulary interesting.

I guess I know what's causing the problem. Arena seems to send the "stop" command even though there is nothing to be stopped anymore as the "stop" command was sent already before. I implemented a workaround for this now. I hope Arena doesn't send other unexpected things as well as I only covered the case when the "stop" command is sent even though there is nothing to be stopped. Not that I'm blaming Arena for this problem though.

link to fixed engine: http://www.rocechess.ch/roce/roce38x.zip

Btw, the debug looks a little bit different when I start a tournament with Arena here. But probably I'm using a different version of Arena. I'm using version Arena 1.99beta5 (Build 2286).

best regards
Roman

PS: Just had a look into the standard and it's definitely my fault. That's what the UCI-Standard says
* if the engine receives a command which is not supposed to come, for example "stop" when the engine is
not calculating, it should also just ignore it.
Seems I was a bit sloppy there ...

Roman Hartmann (Roce)

Number of posts : 5
Registration date : 2008-07-12

View user profile http://www.rocechess.ch/rocee.html

Back to top Go down

It works!

Post  Patrick Buchmann on Mon Jul 14, 2008 1:14 am

Hi Roman,

The version 13/07/08 22:34 works now with Arena 1.1 Very Happy

Thanks for your job.

But now it's a new engine with hashtables support. No ? Perhaps this version must call Roce 0.0381.

Regards,
Patrick

Patrick Buchmann

Number of posts : 30
Registration date : 2008-05-22

View user profile http://pagesperso-orange.fr/lefouduroi/computerchess.htm

Back to top Go down

Re: Roce 0.0380

Post  Roman Hartmann (Roce) on Mon Jul 14, 2008 1:50 pm

Patrick Buchmann wrote:Hi Roman,

The version 13/07/08 22:34 works now with Arena 1.1 Very Happy

Thanks for your job.

But now it's a new engine with hashtables support. No ? Perhaps this version must call Roce 0.0381.

Regards,
Patrick

Hi Patrick,
You're right, of course. But this new versions still needs some testing before I'm going to release it. I'm not too unhappy with the hash-table implementation as it plays definitely better in the endgame but now with the hash-tables null-move is causing some problems. Sometimes it will play "funky" moves now.

best regards
Roman

Edit: For those who are running tournaments with Arena here is a link to a version that works under Arena. The difference to version 0.038 is that it has hash-tables and doesn't use a verification search. The problem with null-move is probably fixed (this problem is not present in the official 0.038 only in 0.038x). Note that this is not an official release even though I gave it a new number (0.0381 as suggested by Patrick).
link: http://www.rocechess.ch/roce/roce381.zip

Roman Hartmann (Roce)

Number of posts : 5
Registration date : 2008-07-12

View user profile http://www.rocechess.ch/rocee.html

Back to top Go down

Re: Roce 0.0380

Post  Sponsored content


Sponsored content


Back to top Go down

View previous topic View next topic Back to top

- Similar topics

 
Permissions in this forum:
You cannot reply to topics in this forum