What's Vasik Rajlich doing today?

[vittyvirus]

Your rant has nothing to do with the thread title, but just one question for you.
How come Rybka 4.1 is still over 100 points stronger than Fabien's latest effort?

+100

Just a little correction,
Rybka 4.1 x64 1core is performed 485 Elo over Fruit 2.1 1core
http://www.sedatcanbaz.com/chess/?page_id=1339

Best,
Sedat

Well, he was talking about Senpai, I felt.
Anyway, 485 elo is A LOT, even if Rybka was based on Fruit.

[Frank Quisinsky]

I gave Graham my answere!

You made all the time the same miskakte. You have compare Version 1.0 with Fruit - bitboards and you can see ...

Again ...
Most improvements from the latest years are the early endgame in chess programming. The differents from Stockfish and Komodo to the most others here are around 300-400 Elo, not in the beginning of the game.

The improvements we can see today in this playing phase is chess knowledge comes from so many people after Fruit! I am sure from Vas and more sure from Larry too.

Read the Stockfish interview I made. Very nice what Marco wrote about programming moduls ... all ideas are good known, the secret is to bring all the ideas in a good code.

Read what GianCarlo (Sjeng) wrote about Stockfish 2. It's possible that with good tuning Stockfish will be 150 Elo stronger. Tuning is the next one. Most of the programmers thinking such things after looking in Stockfish if I remember on comments, also I got by mail.

But it's OK if the most looking on Elo only!

Best
Frank

I wrote it so many times to you Sedat!
Most important is the readme.txt ... I am using ideas from. I like the honest way!

[Highluder]

Well, there was a time when Rybka was the top engine standing above all other engines. This time has gone.

I think not only Vas alone made rybka so strong.

Even Larry Kaufman has contributed much to improving strength on the latest rybka versions (3 and 4). As far as i know, his part was mainly on getting correct evaluations. And rybkas evals where the best at this time. I remember that Larrys article on material imbalances was published in 1995, so when he worked for the rybka team he was the expert in material evaluation. It is a known secret that rybka has already material imbalance tables.

Vas could make rybka much stronger, no doubt. He could make the gap between rybka and Komodo/Stockfish/(Houdini) smaller, but getting number one again after Larry has moved on to Komodo is impossible.

I believe there will be no new version of rybka.

And I believe there will be no new version of houdini.

Vas, Robert: If you read this.
There are many computer chess fans hoping that you develop new versions of your engines. Even if not getting the number one again, your engines belong to the very best. Your engines contributed to engine development and any new version would do it too.

I hope to see new versions. We hope to see new versions.


Just my 2 cents,
Frank

[Sedat Canbaz]

Your rant has nothing to do with the thread title, but just one question for you.
How come Rybka 4.1 is still over 100 points stronger than Fabien's latest effort?

+100

Just a little correction,
Rybka 4.1 x64 1core is performed 485 Elo over Fruit 2.1 1core
http://www.sedatcanbaz.com/chess/?page_id=1339

Best,
Sedat

Well, he was talking about Senpai, I felt.
Anyway, 485 elo is A LOT, even if Rybka was based on Fruit.

Oh..I missed that,

But however,
You are programmer...you can understand for what I am talking...
To improve, to add Extra Elo points over World strongest engine is not so easy as it looks

Just my 2 cents more over this issue
Fruit 2.1 was one of the strongest engines during in the past
And I think that Vas is studied from Fruit's ideas...
But adding approx. 500 Elo...a such engine can not be Fruit clone


I am not a engine programmer, but I am a book maker,
So I can understand a little bit...
The Top (Best) things are oftenly blamed, attacked, criticized...

Sad, but true....

Best,
Sedat

[Frank Quisinsky]

Hi Sedat,

based on Rybka ideas?
That is the question Sedat?!
Possible, yes!

But the Igorit code is clearly weaker as Rybka in endgame. If I remember most problems with cleaning the Igorit code Norman have with endgame code. Later Houdini based on the work Norman do. Houdini 2.0 is clearly stronger in endame as Houdini 1.0, not in the early middlegame.

I am sure that the most own code Vas produced are knowledge for the early endgames / endgames. Later you can see that IPPs are stronger and stronger in endgames, the first playing phase was allways the same with so many IPP clones.

Critter 1.6a:
I believe Critter is an own engine but the programmer used a lot of good known code for improvements. I believe the programmer of Critter added enough own ideas but I believe too that Critter 1.6a and Houdini 1.5a are very similar. No sure what the Critter programmer do, recompling?

Elo improvement is not important alone. More the style of an engine. Critter have more knowledge to pawns if I am looking in my stats as Houdini 1.5.

But all in all ...
To similar to IPPs ... the reason I am testing Version 0.90. Here I can see nothing with stats and analyzes I do.

Best
Frank

[Sedat Canbaz]

But all in all ...
To similar to IPPs ... the reason I am testing Version 0.90. Here I can see nothing with stats and analyzes I do.

Best
Frank

Sorry Frank,
Again you make mistake: Critter 0.90 is not clean one, see and check more carefully the similarity of this Critter version

Best,
Sedat

[vittyvirus]

I gave Graham my answere!

You made all the time the same miskakte. You have compare Version 1.0 with Fruit - bitboards and you can see ...

Again ...
Most improvements from the latest years are the early endgame in chess programming. The differents from Stockfish and Komodo to the most others here are around 300-400 Elo, not in the beginning of the game.

The improvements we can see today in this playing phase is chess knowledge comes from so many people after Fruit! I am sure from Vas and more sure from Larry too.

Read the Stockfish interview I made. Very nice what Marco wrote about programming moduls ... all ideas are good known, the secret is to bring all the ideas in a good code.

Read what GianCarlo (Sjeng) wrote about Stockfish 2. It's possible that with good tuning Stockfish will be 150 Elo stronger. Tuning is the next one. Most of the programmers thinking such things after looking in Stockfish if I remember on comments, also I got by mail.

But it's OK if the most looking on Elo only!

Best
Frank

I wrote it so many times to you Sedat!
Most important is the readme.txt ... I am using ideas from. I like the honest way!

Certainly bitboards do make a (significant) difference. But then if we compare Strelka and Rybka...

[Frank Quisinsky]

Hi Sedat,

if I compare Critter 1.6a with Houdini 1.5 and Critter 0.90 with Houdini 1.0 or others. But your are right again ... Critter is simlar too in Version 0.90 but not to such a big Level 1.6a is to others. Again, in Critter are much own intersting ideas. I wrote about it in my News ... why I am testing Critter 0.90 and gave here some comments about it why I do that.

Best
Frank

[Evert]

Fabians Senpai 1.0 is without bitboards.

No it isn't.

Only clean code without a lot of chess knowledge. For that what Fabian do here is the same level as Hiarcs sensational!

It doesn't have "a lot of chess knowledge". It has a decent amount, but nothing sensational. It's not even tuned very much.

I believe if Fabian working on Senpai and add bitboards and chess knowledge Elo goes over 200 higher.

Wouldn't surprise me, but as I said earlier: I would have expected someone to try to do this just by tuning its evaluation...

[Sedat Canbaz]

Hi Sedat,

if I compare Critter 1.6a with Houdini 1.5 and Critter 0.90 with Houdini 1.0 or others. But your are right again ... Critter is simlar too in Version 0.90 but not to such a big Level 1.6a is to others. Again, in Critter are much own intersting ideas. I wrote about it in my News ... why I am testing Critter 0.90 and gave here some comments about it why I do that.

Best
Frank

It is ok my friend )

But why is disturbing me,
Because you are one of the people, who prefer testing one of the most clean engines
That's why...we should be more careful in case of picking the clean engines

In other words,
Critter 0.90 is just another clone...

Best,
Sedat