Komodo 12.2 Announcement

[Gian-Carlo Pascutto]

Hi, do you have a link to the post of the most common 7 man suggestions

http://talkchess.com/forum3/viewtopic.p ... 17#p769925

[Hurnavich]

Many thanks for that

[kasinp]

[Hi, do you have a link to the post of the most common 7 man suggestions
[/quote]

I did some analysis on a database of human and computer games (over 6 million).
Total number of times a 7-men position was reached was 616,969.

6-1 endgames: 5,999
5-2 endgames: 80,275
4-3 endgames: 530,695

Here are the 30 most commonly found; together they account for over 80% of all cases.
The top 10 account for 58.5% of all cases.

497,376 80.6%

1 KRPPKRP 153,091 24.8%
2 KPPPKPP 63,395 10.3%
3 KBPPKBP 24,923 4.0%
4 KNPPKNP 20,204 3.3%
5 KRPPPKR 20,144 3.3%
6 KNPPKBP 19,617 3.2%
7 KBPPKNP 17,783 2.9%
8 KBPPKPP 15,470 2.5%
9 KNPPKPP 13,554 2.2%
10 KRPPKPP 13,052 2.1%
11 KQPPKQP 12,302 2.0%
12 KBPPKRP 12,082 2.0%
13 KRBPKRP 11,080 1.8%
14 KPPPKBP 9,392 1.5%
15 KRNPKRP 9,210 1.5%
16 KRPPKBP 9,074 1.5%
17 KPPPKRP 8,564 1.4%
18 KQPPKPP 8,339 1.4%
19 KPPPKNP 8,151 1.3%
20 KNPPKRP 7,567 1.2%
21 KRPPKNP 6,025 1.0%
22 KPPPKQP 5,228 0.8%
23 KPPPPKP 5,075 0.8%
24 KRPPKRB 4,330 0.7%
25 KRPPKQP 3,904 0.6%
26 KRPPKRN 3,785 0.6%
27 KBPPPKP 3,575 0.6%
28 KRBPKRB 2,865 0.5%
29 KBPPPKB 2,807 0.5%
30 KNPPPKP 2,788 0.5%

PK

[mjlef]

Mark:
My idea was that very few people download full 7-man TBs--just too big to get or store on disk; maybe too big to access quickly on disk. Therefore, what your TB guy should think about is having Komodo access 6-man TB on disk, as it always has; and access 7-man TB from an online repository, e.g., the one maintained by Chess OK, vendor of Aquarium.
If your team can do that, then Komodo becomes a must-have product for correspondence players.

online accessing 7 piece tablebases would be way too slow. If you want a free 7 piece tablebase program, the Android Lomonosov 7 App has them.

We do plan to eventually support 7 piece Syzygy, but locally. The internet is too slow to access millions of positions a second.

[schack]

Any idea when 12.2.1 will be released?

[mjlef]

Any idea when 12.2.1 will be released?

We are running tests on it now. If they pass then soon. If not, I have more work to do. Looking good so far.

[schack]

Thanks!

[Nordlandia]

Thanks for another Komodo update Mark and Larry

Considering H6 is approaching 12 months since release. Is it wise to tune down contemt from default to 1 or 0 vs latest Komodo?

Houdini 6 is almost 1 year old. Maybe default contempt is not optimal against latest K release. Is K contempt of zero ideal against H6? (K is new while H6 is 1 year old) Any thoughts?

[lkaufman]

Thanks for another Komodo update Mark and Larry

Considering H6 is approaching 12 months since release. Is it wise to tune down contemt from default to 1 or 0 vs latest Komodo?

Houdini 6 is almost 1 year old. Maybe default contempt is not optimal against latest K release. Is K contempt of zero ideal against H6? (K is new while H6 is 1 year old) Any thoughts?

H6 and K12 are close enough in strength to use zero Contempt. Maybe 1 or 2 might be miscroscopically better, but it would be probably less than one elo and so impossible to determine. Probably Contempt should be lower than 16 for MCTS in most situations.

[Nordlandia]

Sounds like zero is good all-around choise for both parties.