Texel 1.05

[Charly]

Hi,

Many thanks for this new release ! I will test as soon as possible in my leagues.

A humoristic suggestion : maybe you could help the "author of Taipan", the "new super mega strongest engine in the world coming from nowhere" to learn programmation ?

[petero2]

The tablebase implementation is non-standard, and while I have mentioned it in the programming forum before, it is still probably not widely known. Here is a quote from the readme file:

Code: Select all

Tablebases
----------

Texel can use endgame tablebases to improve game play and analysis in the
endgame. Both Gaviota tablebases (only .cp4 compression) and Syzygy tablebases
are supported, and both tablebase types can be used simultaneously.

For game play Syzygy tablebases are recommended because the Syzygy probing code
scales better when the engine uses multiple cores.

For analysis, using both Syzygy and Gaviota tablebases at the same time is
recommended. This gives accurate mate scores and PVs when the search can reach a
5-men position (thanks to Gaviota tablebases), and game theoretically correct
results (also taking 50-move draws into account) when the search can reach a
6-men position (thanks to Syzygy tablebases).

Syzygy tablebases contain distance to zeroing move (DTZ) information instead of
distance to mate (DTM) information. DTZ values are converted internally to upper
or lower DTM bounds before being presented to the user. This means that there is
no separate score range for known tablebase wins, but it also means that the
shortest possible mate can be much shorter than the reported mate score
indicates.

For syzygy tablebases it is recommended to have the corresponding DTZ table for
each WDL table. Texel tries to handle missing tablebase files gracefully, but in
some situations missing DTZ tables may lead to trouble converting a won
tablebase position. This can happen even in relatively simple endgames that
Texel could have won without using tablebases at all.

Because of technicalities in the Syzygy probing code, 6-men tablebases are only
supported for the 64-bit versions of Texel.

The 6-men Syzygy tablebases will likely only increase the playing strength of
Texel if at least the WDL tables are stored on SSD.

Some further clarifications:

There is no harm in using both gaviota and syzygy tablebases at the same time for game play. The gaviota tables will not be used until the game theoretic value (win/loss/draw) of the root position has been determined, so their impact on elo should be zero.

Using 5-men syzygy tablebases should be effective even if they are stored on HDD, because they are small enough to be fully cached in RAM on typical computer hardware.

[mar]

Congratulations Peter!

For new Texel and for breaking 3k CCRL barrier on sigle core, 3100 on 4, for being original and hardworking while also helping others

I hope to see Texel in top 10 someday as we badly need something fresh in those waters.

[Sedat Canbaz]

Congratulations for all your hard work. I'm testing now against Critter 1.6a (tc=15+0.15 threads=7), and it seems to be holding its ground better than I expected (remember Critter is an absolute killer at these short tc).

It's rare to see honest engine developers break the 3100 elo barrier. I hope more engine testers were capable of understanding the difference of an engine like Texel and a cheap clone (Amitis, Elektro, Shark, Taipan, etc.)

Well, here is not topic to discuss origins...
But to not mentioned this I can't,
What about your testing engine: Critter 1.6a ???
According to my experience: it a cheap clone of Houdini 1.5a
For more details, check please the similarity and the strength of Critter and Houdini
Ok...Houidini can be be not 100 % original (based mainly on ideas of Rybka, Stockfish),
But at least I can say for Houdini's author: the great talent in computer chess !!

And very sad that,
Many testers (including programers) are so naive that can not see the reality and give credits to cloners )!
In this way...we will see a lot of more clone engines in 2015 !!


Btw,
Many Thanks to Peter, and very soon I will start testing Texel )


Best,
Sedat

[Joerg Oster]

Congratulations Peter!

For new Texel and for breaking 3k CCRL barrier on sigle core, 3100 on 4, for being original and hardworking while also helping others

I hope to see Texel in top 10 someday as we badly need something fresh in those waters.

+1!

[petero2]

Congratulations for all your hard work. I'm testing now against Critter 1.6a (tc=15+0.15 threads=7), and it seems to be holding its ground better than I expected (remember Critter is an absolute killer at these short tc).

Thanks. One advantage of not having enough computer resources to test at long time controls is that the engine tends to get optimized for short time controls.

Another advantage is that it also works quite well for typical android hardware, which is about 10 times slower than typical PC hardware.

[Bloodbane]

I think congratulations are in order. Texel is a fantastic combination of extreme strength and simple, clear and concise code from which I have learned a lot, from both a chess programming perspective and from a general software engineering perspective. I hope you will continue developing it with success.

[Frank Quisinsky]

Hi Sedat,

Little correction:
Houdini 1.0 based on Robbodini to 99% after all messages I am reading from other programmers (Thinker, Critter programmers, others). Enough proofs are public at this time and later ...

Code from Igorit are cleaning by Robbodini programmer Norman Schmidt. Norman added some own ideas at this time.

Houdini 1.0 is a clone from Robbodini!

Unclear is in my opinion from which one are the further Houdini improvements, maybe from Robert, maybe not.

The history of computer chess in questions of clones is allways the same. People using a work by others, do it with others sources too and in the most cases no own ideas added! Unclear in case of Houdini but after all what the Critter programmer wrote in Rybka forum and here I can't build an other opinion.

In my opinion you are right with Critter 1.6a. Most improvments from Critter 1.6a based on Houdini 1.5. Latest interesting Critter Version is in my opinion version 0.90!

Sorry, but I can't build an other opinion about it after all Information (a lot comes by mail from programmers which explain a bit).

Best
Frank

Please let us not start a clone discuss in the nice Texel thread. We can discuss by mail if you like or in the Orgin selection.

[Frank Quisinsky]

Hi Peter,

I am reading about your "Lego Project". It seems that you are full of ideas and I am sure you are looking not only on your Android project and Texel.

Droidfish is great!
I am using on my HTC M8 most time Droidfish.

My wish here, maybe possible?
Could you added with the next Droidfish version your actual Version of Texel too? Stockfish isn't the once flagship with open sources. You have an own and this should be added directly with Droidfish setup as second engine.

Best
Frank

[Frank Quisinsky]

Hi Mikko,

I am playing a bit with your engine on my Notebook and have your engine in my observation for FCT1. 50 Elo more "only" and Hakkapeliitta can start in my Frank's Continuous Tourney 1. Very nice newer development and a nice CEGT test I can see!

Good luck with your engine!

Best
Frank

Must have a chess engine from Finland in my tournament. I am a big fan of the ice hockey team of Finland (since I am 10 years old, looking with my father, my father was a big fan from the team too and the older German super star "Kühnhackl" have the most respect from this team too) and I am looking each game. In the near of my TV I have all the years a flag from Finland Not say it to the others ... to many American and Canadian people here ...

Go Hakkapeliitta, go ...