Page 4 of 5

Re: New uci engine: Rofchade

Posted: Wed Sep 12, 2018 9:25 pm
by Scally
Hi Ronald,

Thanks for the binary, it runs but I get a segmentation fault on stopping the search via ‘quit’ when running go infinite.

Al@PicoChess:/opt/picochess/engines/armv7l $ ./rofchade
RofChade Version 1.0 b104
by Ronald Friederich

enter 'help' to see the commands
uci
id name RofChade Version 1.0 b104
id author Ronald Friederich
option name Hash type spin default 256 min 1 max 32768
option name Threads type spin default 1 min 1 max 64
option name Ponder type check default false
uciok
go infinite
info score cp 23 depth 11 seldepth 14 nodes 134199 time 495 hashfull 0 pv e2e4 e7e5 g1f3 b8c6 b1c3 g8f6 f1d3 d7d5 e1g1 c8e6 h2h3
info score cp 10 depth 12 seldepth 14 nodes 353096 time 1096 hashfull 0 pv e2e4 e7e5 g1f3 b8c6 b1c3 g8f6 f1d3 d7d5 e4d5 f6d5 c3d5 d8d5
info score cp 24 depth 13 seldepth 15 nodes 506329 time 1473 hashfull 0 pv e2e4 e7e5 g1f3 b8c6 b1c3 g8f6 f1c4 f8b4 c3d5 f6e4 d5b4 c6b4 f3e5
quitinfo score cp 24 depth 14 seldepth 17 nodes 798972 time 2170 hashfull 0 pv e2e4 e7e5 g1f3 b8c6 b1c3 g8f6 f1c4 f8b4 c3d5 f6e4 c4d3 f7f5 d5b4 c6b4

info score cp 24 depth 15 seldepth 19 nodes 1263399 time 3243 hashfull 0 pv e2e4 e7e5 g1f3 b8c6 b1c3 g8f6 f1c4 f8b4 c3d5 f6e4 c4d3 f7f5 d5b4 c6b4 f3e5
Segmentation fault

Thanks for the gcc Compile, I look forward to the source code when I’ll try other compile methods.


Cheers,

Al.

Re: New uci engine: Rofchade

Posted: Sat Nov 10, 2018 10:34 am
by mibere
Hello Ronald,

first of all, thanks for your engine :)

I'm using HIARCS Chess Explorer v1.9.2 on macOS. In that GUI your engine (regular version) outputs something like
RofChade Version 1.0 b104 (124,9M Knoten, 2 nps)
On every move it strangely shows "2 nps". Maybe you can have a look at it?

Re: New uci engine: Rofchade

Posted: Sat Nov 10, 2018 1:53 pm
by Ronald
Hi,

You' re welcome :D

I am not familiar with the Hiarcs gui, so I don't know exactly what part of the output you are showing. I do know that Rofchade doesn't output nps in its info string so it has to be calculated by the GUI based on total nodes and time. Is it possible to show a full output of the engine to the GUI, something like:
info score cp 23 depth 11 seldepth 14 nodes 134199 time 39 hashfull 0 pv e2e4 e7e5 g1f3 b8c6 b1c3 g8f6 f1d3 d7d5 e1g1 c8e6 h2h3
info score cp 10 depth 12 seldepth 14 nodes 353096 time 102 hashfull 0 pv e2e4 e7e5 g1f3 b8c6 b1c3 g8f6 f1d3 d7d5 e4d5 f6d5 c3d5 d8d5
etc..

Re: New uci engine: Rofchade

Posted: Sat Nov 10, 2018 2:22 pm
by mibere
Ronald wrote:
Sat Nov 10, 2018 1:53 pm
I am not familiar with the Hiarcs gui, so I don't know exactly what part of the output you are showing.
Bildschirmfoto 2018-11-10 um 15.13.46.png
Bildschirmfoto 2018-11-10 um 15.13.46.png (35.27 KiB) Viewed 1987 times

The nps output of Stockfish/Hiarcs/Junior in the GUI is correct.

Re: New uci engine: Rofchade

Posted: Sat Nov 10, 2018 2:42 pm
by Fuddur
Hi, Will rofChade be available for android?
It is a request if you provide us an android version.
rofChade is great chess engine!
Waiting for reply!!!

Re: New uci engine: Rofchade

Posted: Sat Nov 10, 2018 5:06 pm
by Ronald
I suppose it has to do with the fact that rofchade doesn't return the nps field in the info score part. All the GUI's I use calculate the nps themselves, because nodes and time are also given in the info score part. I don't have plans to add the nps to the info score part yet.
Hi, Will rofChade be available for android?
It is a request if you provide us an android version.
rofChade is great chess engine!
Waiting for reply!!!
Thanks :D
I own an Iphone myself, so an android version of rofchade is not so high on my priority list :lol: . The current "official" version 1.0 is a Piece Square Table only version, my main concern for now is the development of a good evaluation function.

I don't know anything about android, on Windows I compile rofchade with MSVC, it should be possible to compile C++ with MSVC for android but I don't know if that's easy or a bumpy road.. Somewhere on my todo list is an action to make te sourcecode of version 1.0 available after some cleanup, maybe somebody else can create an android version with that.

Re: New uci engine: Rofchade

Posted: Sat Nov 10, 2018 5:35 pm
by CMCanavessi
Ronald, I see that TCEC has a dev version of Rofchade with a partial evaluation now, and it's SUPER STRONG! Are you planning on releasing a public version soon or is it still far from complete, even if it's much stronger than 1.0?

Re: New uci engine: Rofchade

Posted: Sun Nov 11, 2018 1:37 pm
by Ronald
It does pretty well :D .
Rofchade will participate in the next programmers tournament of CSVN on december 1st and 2th, I am experimenting with eval to get a good version for that and was not busy with creating a new version yet.

To maybe join TCEC was a surprise so I created an intermediary version for TCEC overnight, which does pretty well so far. Time control is very short for this tournament, with so many threads getting all the workers running for every move may become relative time consuming, so perhaps engines which do this more efficiently are doing better with this time control. With longer time controls the starting time will be irrelevant. I don't know if it will do as well with longer time controls also, because I haven't done real testing at longer time controls yet.

If I am satisfied with the result of the CSVN version, I'll probably will release version 2.0 soon after.

Re: New uci engine: Rofchade

Posted: Sun Nov 11, 2018 1:48 pm
by Fuddur
[/quote]

where can I find source code of rofChade ?

Re: New uci engine: Rofchade

Posted: Sun Nov 11, 2018 2:55 pm
by Damir
Why do you always keep asking about the source ?
Do you want to clone the engine ?