Crystal

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

Vinvin
Posts: 5228
Joined: Thu Mar 09, 2006 9:40 am
Full name: Vincent Lejeune

Re: Crystal

Post by Vinvin »

I would be glad to someone who would make a Windows (moderm) compile !

Thanks to Joseph for this gift !
Stockfish has become less and less interesting for analyze because it miss so many obvious moves :-(
Dann Corbit
Posts: 12537
Joined: Wed Mar 08, 2006 8:57 pm
Location: Redmond, WA USA

Re: Crystal

Post by Dann Corbit »

Here is my version:
https://drive.google.com/open?id=1xEqAs ... MqEaN2D_0y

Unfortunately, I make some changes for my own purposes. You can figure them out with diff.
If you want to use this version, you will probably want to turn off writing the analysis to disk UCI logging option.

There is a 64 bit binary for ancient CPUs and one for SSE CPUs, but I do not have a BMI build in it.
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.
User avatar
Marek Soszynski
Posts: 581
Joined: Wed May 10, 2006 7:28 pm
Location: Birmingham, England

Re: Crystal

Post by Marek Soszynski »

Many thanks Dann!
Marek Soszynski
Vinvin
Posts: 5228
Joined: Thu Mar 09, 2006 9:40 am
Full name: Vincent Lejeune

Re: Crystal

Post by Vinvin »

Marek Soszynski wrote: Sat Aug 17, 2019 8:56 am Many thanks Dann!
+1 :D
User avatar
MikeB
Posts: 4889
Joined: Thu Mar 09, 2006 6:34 am
Location: Pen Argyl, Pennsylvania

Re: Crystal

Post by MikeB »

jhellis3 wrote: Wed Aug 14, 2019 10:16 pm I had intended to "release" this quite some time ago, but I never seemed to reach a point where I consider it finished (and still have not).

Nevertheless, I thought I would go ahead and post the branch here for those who might be interested.

Crystal remains quite similar to Stockfish, though the departure is greater than MateFinder ever was, and will likely continue to grow over time. The major discriminating factor being the underlying philosophy and reasoning behind the changes.

A few highlights:

0 Default Base Contempt
Dynamic Contempt is a toggle and defaults to false
Analysis Contempt defaults to false
Full key TT
Max Depth 192 (though SF now supports even higher than this using the same trick)
Old depth iteration & display. WYSIWYG as much as possible.
Some small move gen changes
Move legality check moved to movepick
Mate search cutoff
Less LMR near root
Syzygy TB scoring (displayed score reflects the number of pieces remaining on the board and the DTZ)
Emphasis on maximizing the quality of the PV (much less pruning)

Future work will focus on enhanced draw detection, static eval tricks, and greater tactical awareness, though I have no timeline for this. I had hoped to have made more progress than this before release but...

Don't have any builds (and since my CPU is quite old they wouldn't be any good anyway), but the repo is located here:
https://github.com/jhellis3/Stockfish/tree/crystal

Default bench should be: 7215170
Thanks Joe!
Image
maxdeg
Posts: 229
Joined: Thu Mar 10, 2016 6:17 pm
Location: Canada

Re: Crystal

Post by maxdeg »

Marek Soszynski wrote: Sat Aug 17, 2019 8:56 am Many thanks Dann!
+1 :D :D
jhellis3
Posts: 546
Joined: Sat Aug 17, 2013 12:36 am

Re: Crystal

Post by jhellis3 »

I have updated Crystal; changes include preliminary game cycle detection and slightly increased tactical awareness.

Sauce: https://github.com/jhellis3/Stockfish/tree/crystal

Bench: 7410759
Paloma
Posts: 1167
Joined: Thu Dec 25, 2008 9:07 pm
Full name: Herbert L

Re: Crystal

Post by Paloma »

Why no binaries?

Most people (more than 95%) can't compile.
Spliffjiffer
Posts: 416
Joined: Thu Aug 02, 2012 7:48 pm
Location: Germany

Re: Crystal

Post by Spliffjiffer »

this is a 64bit compile popcnt (modern)...i cant provide any other...its from the completely unchanged source code by J.Ellis from the link he gave above....hopefully it fits to your system:

https://drive.google.com/open?id=1gNDUW ... ZHFuNgxAjR

its much easier though when u compile Stockfishes (or crystal-SF eg as well) by yourself in the future...therefore i uploaded the SF-BYO (Build Your Own) 2017a edition that i use myself...there is no virus eighter in the 1. nor in the 2. link i gave !!
to use SF-BYO, do this:
1. download it
2. create a new folder on your DESKTOP <-- !!!....renane that folder to eg "180825" (i think 190825 would not work because of a bug !?)...18 is the year, 08 is the month and 25 the day.
3.download the src.code u wish to compile eg J.Ellis crystal from his link above
4.in such downloads there is a folder named "src"...go into this folder and copy the whole content and paste it into the folder u created at the dektop as described before (use the name "180825" just to look if it works proberly)
5.doubbleclick or run the StockfishBYO.exe with administrator-rights
6.a window should open that writes u that a folder named "180825" was found on your desktop...if u did as i said everything from now on goes automatically....it will make a first compile, then tuning (according to your machine nearly perfectly) and then does a last compile...after that the windows closes and a .rar file and a folder was created named "Stockfish" on your desktop
7. open your gui and install the engine as ure used to from this "Stockfish" folder on the desktop...you are done, hf :-)

https://drive.google.com/open?id=1exgcp ... lUgcgx20Yt
Wahrheiten sind Illusionen von denen wir aber vergessen haben dass sie welche sind.
User avatar
Scally
Posts: 232
Joined: Thu Sep 28, 2017 9:34 pm
Location: Bermondsey, London
Full name: Alan Cooper

Re: Crystal

Post by Scally »

Here’s a Raspberry Pi binary with uci file:

https://github.com/ScallyBag/AlsPicoEng ... /requested


Cheers,

Al.