Slow Chess Blitz

[Jon Kreuzer]

http://www.3dkingdoms.com/chess/SlowBlitz0.zip

I thought I'd put up a download for this version instead of letting it gather dust. (I haven't done any chess programming for a month, other than finally adding support for en passant in FEN copy and paste a few days ago.)

This is an intermediate version, the idea was to completely change the eval (Make the eval more bloated while increasing lazy eval) and the search too (by adding a tactical quiescence search that includes checks and other moves.) I haven't done nearly all this though, so I don't claim this isn't mostly still the same Slow Chess. A lot of the King Safety and Passed Pawn scoring are new at least. Many other smaller eval changes and search changes (ie. finds some backrow mates in the QSearch now, etc.)

This one also has personality support (add -agg, -con or -dep to the command line, "slow.exe -agg", these are almost surely weaker than default.) I haven't made a personality editor yet, but next version will have full support of this.

I make no playing strength claims, but it's different enough it might be fun to test.

[Ross Boyd]

Congrats on the new release Jon.

I predict a strong improvement... checks in qs, king safety and passed pawn scoring helped me a lot. Slowchess is going to be a real handful...

Watch out Winboarders!

Ross

[Normand]

Hello ,

shall i replace SlowChess with this new version on the Olympiades 2004 tournament ?...
What do you think about this replacement ?

Regards

Nicolas

[Pallav Nawani]

Congrats on the new release Jon.

I predict a strong improvement... checks in qs, king safety and passed pawn scoring helped me a lot. Slowchess is going to be a real handful...

Watch out Winboarders!

Ross

Hello Ross,

I've been meaning to add checks in qsearch to my program for some time now. Can you say how much ELO benefit it gives on the avarage?

Regards,
Pallav

[Jon Kreuzer]

The King Safety and passed pawn scoring are newer and slower, but not necessarily better =)
There are some types checks in QS, (and check replies anytime the king is in check in the QS), but they were never completely implemented because I wanted to write a checking move generation function, but was too lazy to actually do it. I know it will be better for playing strength to be more general rather than doing all the crazy special purpose code I have (which doesn't take any longer to write surprisingly, but misses too many cases, and is more likely to be buggy.)

I couldn't guess the ELO change of anything myself. I'm most interested in finding some moves really fast, which a Tactical Q-search can do. Slow Chess solves this position with about a +2 score in 4 seconds (AMD 2700+), I don't know if Slow Chess 2.82 would get a + score in 10 minutes.
(3r2k1/2Nr1ppp/p5q1/5b2/1b1n4/2RPQ1BP/BP3PP1/4R1K1 w - - ;bm Nd5)

About which version to use: If you want me to choose, I'd always recommend the newest version, whether it's the strongest or not. I do assume Blitz probably isn't weaker than 2.94.

[Uri Blass]

The King Safety and passed pawn scoring are newer and slower, but not necessarily better =)

Then what is the explanation for the big improvement that is reported in
http://www.talkchess.com/forums/1/message.html?397314

Uri

[Tord Romstad]

I've been meaning to add checks in qsearch to my program for some time now. Can you say how much ELO benefit it gives on the avarage?

I think this depends to a great extent on how the rest of your program works. In general, it seems that checks in the qsearch works better combined with a null move reduction factor of 3 than with lower factors. I also suspect that checks in qsearch are more important for relatively slow engines than for extremely fast engines, but I have no data to back up this claim.

As always, you just have to try and see whether they work for you. But don't give up too soon. Even if your initial results are not promising, it might be possible to improve by finding good ways to reduce the number of checks searched.

Tord

[Dan Honeycutt]

Checks in qsearch is good fodder for discussion, but we should be in Programming. I've opened a thread.

Dan H.

[Jon Kreuzer]

I do hope Blitz is improved. I like its play better in terms of general evaluation and King Safety (still after the rewrite I see some of the same type of misevaluations, I guess I naturally write King Safety evals to be similar.) This is just from watching some games though, I wouldn't want to declare it better in games without extensive testing. I've kinda given up on testing and estimating playing strength myself.

I did test 20 games against Ruffian (at 2 minutes + 2 secs per game), Slow was up 2 points after 10 games, and even after 20, so I thought it was a clear improvement, and started another match of 40 new games. Slow scored about 22%, so I decided it might not be better. After seeing other people's results I assume it is an improvement on 2.94, but not as much as some results initially indicate. I am glad that Slow Chess Blitz rocks though, I was hoping it would =)

[Anonymous]

http://www.3dkingdoms.com/chess/SlowBlitz0.zip

I thought I'd put up a download for this version instead of letting it gather dust. (I haven't done any chess programming for a month, other than finally adding support for en passant in FEN copy and paste a few days ago.)

This is an intermediate version, the idea was to completely change the eval (Make the eval more bloated while increasing lazy eval) and the search too (by adding a tactical quiescence search that includes checks and other moves.) I haven't done nearly all this though, so I don't claim this isn't mostly still the same Slow Chess. A lot of the King Safety and Passed Pawn scoring are new at least. Many other smaller eval changes and search changes (ie. finds some backrow mates in the QSearch now, etc.)

This one also has personality support (add -agg, -con or -dep to the command line, "slow.exe -agg", these are almost surely weaker than default.) I haven't made a personality editor yet, but next version will have full support of this.

I make no playing strength claims, but it's different enough it might be fun to test.

There isn't any documentation included in the SlowChess Blitz download. I assume "Schess.dft" is the configuration file. How to configure SlowChess Blitz, hash settings, etc. ?

This is what is contained in "Schess.dft" ...

----------
Human
127.0.0.1
5
5
28
0
0
1
21
1
1
1
0
0
1
1
1
TestBook.scb
1
1
1
1
TestBook.scb
1
----------

Regards
Dave

[Dr.Wael Deeb]

Hi Dave,
Just open the GUI of Slow Chess Blitz and under the menu hasing choose your suitable hash size,it will save the settings when you close it's GUI!
Regards,
Dr.Wael Deeb