updates chess engines

[tmokonen]

And, the recently released engine SCTR, by Can Cetin & Dogac Eldenk:

https://github.com/nitrocan/sctr/releases

I have been testing it against the likes of Rodent III and ChessBrainVB, and it is competitive with those engines.

[Guenther]

And, the recently released engine SCTR, by Can Cetin & Dogac Eldenk:

https://github.com/nitrocan/sctr/releases

I have been testing it against the likes of Rodent III and ChessBrainVB, and it is competitive with those engines.

I am not sure that one is kosher...

[tmokonen]

What is it similar to?

[Graham Banks]

And, the recently released engine SCTR, by Can Cetin & Dogac Eldenk:

https://github.com/nitrocan/sctr/releases

I have been testing it against the likes of Rodent III and ChessBrainVB, and it is competitive with those engines.

I am not sure that one is kosher...

It is open source. What are your concerns, Guenther?

[Guenther]

Hmm, seems that an engine called DirtyBit, by Andrew Backes, has been overlooked:

https://github.com/andrewbackes/dirty-bit/releases

I played a bit with the latest release. It's a fixed hash size, about 775 MB. I've only run a few test games with it:

4-0-0 versus Isa (2186 ELO)
3-0-0 versus EveAnn (2337 ELO)
1-4-1 versus Ethereal (2631 ELO)

So anyway, I guess we get a very rough ballpark estimate of its strength.

It had its own web page at one point, but it is now only viewable through the Wayback Machine:

http://web.archive.org/web/201603121230 ... y-bit.com/

Thanks! Never heard of this one, Tony.

[Guenther]

Hmm, seems that an engine called DirtyBit, by Andrew Backes, has been overlooked:

https://github.com/andrewbackes/dirty-bit/releases

I played a bit with the latest release. It's a fixed hash size, about 775 MB. I've only run a few test games with it:

4-0-0 versus Isa (2186 ELO)
3-0-0 versus EveAnn (2337 ELO)
1-4-1 versus Ethereal (2631 ELO)

So anyway, I guess we get a very rough ballpark estimate of its strength.

It had its own web page at one point, but it is now only viewable through the Wayback Machine:

http://web.archive.org/web/201603121230 ... y-bit.com/

Thanks! Never heard of this one, Tony.

I will add the wayback link below instead, because it has same content as above, except that the css file is saved with it too.

http://web.archive.org/web/201601110154 ... y-bit.com/

It seems the author was very interested to get a rating at CCRL, but strangely he never announced a release somewhere, except at the later created github site in 2015/06?
@Graham: did he ever try to contact CCRL before he stopped working on Dirty-Bit?

About Dirty-Bit

• Project Goals
  
  Engine:
  ★ Goal: Create a program that plays legal chess using bitmap based internal data structure to represent the gamestate.
  ★ Goal: Create a program that can beat other chess playing programs, namely Exacto 0.d.
  ☆ Goal: Obtain a CCRL rating of at least 2000 ELO.
  ☆ Goal: Play at the Master level. I am estimating this to be approximately 2300 ELO on the CCRL scale.
  ☆ Goal: Rank within the top 100 programs on CCRL.

Current Features

Dirty-Bit has most well known search heuristics and techniques in its code. However, some of the features still require tuning and have been disabled.

• Enabled Features:
  
  Principle Variation Search (PVS)
  Quiescent Search (with delta pruning)
  Null-Move Pruning
  Late Move Reductions
  Piece Square Tables
  Search Extensions
  MVV/LVA
  Static Exchange Evaluation
  Hash Tables (Transposition Table and Pawn Table)
  Iterative Deepening
  Internal Iterative Deepening
  Aspiration Windows
  Magic Bitboards (9 bit / 12 bit)
  Lazy move generation[/code]
  
  Disabled Features:
  
  Killer Moves - Currently corrupts the move list
  Futility Pruning - Somehow increases nodes search!?
  
  Planned Features
  
  Other than tuning the above disabled features, the following need to be implimented before release:
  
  Multithreaded Search (DTS)
  More elaborate evaluation
  
  These features will most likely be added after release since they will not be used in CCRL games:
  
  Pondering
  Opening Book
  Positional Learning

Change Log

Code: Select all

0.37 	9-12 Magics
More eval tweaks
added movestogo (UCI) command to go()

0.36 	Mobility eval

0.35 	Pawn Hash Tables and more pawn eval.

0.31 	R=3

0.3 	Rewritten LMR (set to 4/3)
Killers disabled (since they are buggy at the moment)
Rewritten SEE (SEE3, set for ALL captures, not just non-winners)
Still using Check extensions

0.25 	SEE and move ordering changes
IID implimented, but disabled (needs tuning!)
Time control fixes
Check extensions
50 move rule bug fixed
LMR disabled
34 cp for apsiration window

0.2 	Rewritten fail hard PVS using nextmove()
LMR above ply 4 and below move 5
Aspiration window of 50, fails expand by 100 then go infinite
Simple Eval
No check extensions
No to move bonus

0.12 	No Check extensions, no LMR
(Changed move list to use NextMove() and lazy killers)

0.11 	Check extensions

0.1 	Lazy movegen PVS + killer moves
Aspiration windows
Simple eval: material and PSTs
No check extensions, no LMR

0.01 	Plays legal chess

[nitrocan]

And, the recently released engine SCTR, by Can Cetin & Dogac Eldenk:

https://github.com/nitrocan/sctr/releases

I have been testing it against the likes of Rodent III and ChessBrainVB, and it is competitive with those engines.

I am not sure that one is kosher...

Hey Guenther, I'm one of the authors of SCTR. Can you explain why you think SCTR is not kosher? Thanks.

[Graham Banks]

@Graham: did he ever try to contact CCRL before he stopped working on Dirty-Bit?

Can't recall off hand, but we certainly tested Exacto.