Top 3 engines have TB implementations obeying Bible morals

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

Moderators: hgm, Rebel, chrisw

Joerg Oster
Posts: 937
Joined: Fri Mar 10, 2006 4:29 pm
Location: Germany

Re: Top 3 engines have TB implementations obeying Bible mora

Post by Joerg Oster »

Here the results of a first sanity check over 100 games each.
Conditions: 500 ms per move, 16 MB Hash, Cutechess-cli 1.0 (no resign adjudication)
50 start positions randomly chosen from your End_789_13_17.epd (with -repeat option set)

SF-NoSyzygy vs. SF-6Syzygy

Code: Select all

Finished game 100 (SF-Master-6Syzygy vs SF-Master-NoSyzygy): 1-0 {White mates}
Score of SF-Master-NoSyzygy vs SF-Master-6Syzygy: 32 - 36 - 32  [0.480] 100
Elo difference: -13.90 +/- 56.59
Finished match

SF-NoSyzygy vs. Zurichess-Neuchatel

Code: Select all

Finished game 100 (Zurichess-Neuchatel vs SF-Master-NoSyzygy): 1/2-1/2 {Draw by fifty moves rule}
Score of SF-Master-NoSyzygy vs Zurichess-Neuchatel: 44 - 11 - 45  [0.665] 100
Elo difference: 119.11 +/- 51.17
Finished match

SF-6Syzygy vs. Zurichess-Neuchatel

Code: Select all

Finished game 100 (Zurichess-Neuchatel vs SF-Master-NoSyzygy): 1/2-1/2 {Draw by fifty moves rule}
Score of SF-Master-NoSyzygy vs Zurichess-Neuchatel: 44 - 11 - 45  [0.665] 100
Elo difference: 119.11 +/- 51.17
Finished match
Tomorrow I will check the games for any peculiarities.
If everything is fine, I will run the 1,000 games comparisons.
Jörg Oster
brtzsnr
Posts: 433
Joined: Fri Jan 16, 2015 4:02 pm

Re: Top 3 engines have TB implementations obeying Bible mora

Post by brtzsnr »

Hi, Kai!

Can I have the dataset, too? The last one (Draw 5-men) I got from you uncovered some search issues for which I've never thanked you :)

I'm actually testing endgame bitbases for Zurichess [1]. So far after adding KPK, KRK, KQK and KRKR I've seen 0 Elo improvement. I'd like to see if egtb make sense for positions closer to the endgame.

[1] https://bitbucket.org/zurichess/endgame/src is the current generator. Not very fast, at ~2 secs for a 3-men table.
User avatar
Laskos
Posts: 10948
Joined: Wed Jul 26, 2006 10:21 pm
Full name: Kai Laskos

Re: Top 3 engines have TB implementations obeying Bible mora

Post by Laskos »

brtzsnr wrote:Hi, Kai!

Can I have the dataset, too? The last one (Draw 5-men) I got from you uncovered some search issues for which I've never thanked you :)

I'm actually testing endgame bitbases for Zurichess [1]. So far after adding KPK, KRK, KQK and KRKR I've seen 0 Elo improvement. I'd like to see if egtb make sense for positions closer to the endgame.

[1] https://bitbucket.org/zurichess/endgame/src is the current generator. Not very fast, at ~2 secs for a 3-men table.
I will send you the link via PM.