Stunning Komodo

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

Moderators: Harvey Williamson, bob, hgm

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Dann Corbit
Posts: 11622
Joined: Wed Mar 08, 2006 7:57 pm
Location: Redmond, WA USA
Contact:

Re: Stunning Komodo

Post by Dann Corbit » Mon Mar 14, 2011 11:25 pm

NATIONAL12 wrote:i just hope when Dan and Larry release MP version they will make it available for up to 16 cpu.i will def pay for it.in fact i dont mind doing a bit of private testing for them.i will still pay for it.
It's 'Don' as in 'Don Dailey'.

I am not sure where 'Dan' came from because it is definitely Don with an 'o' and not Dan with an 'a'.

benstoker
Posts: 342
Joined: Tue Jan 19, 2010 1:05 am

Re: Stunning Komodo

Post by benstoker » Tue Mar 15, 2011 2:46 am

Albert Silver wrote:
benstoker wrote:
Martin Thoresen wrote:You can follow the latest development version of Komodo in TCEC C which is live right now.

http://www.tcec-chess.org/tcec_c.php

Not sure that I can reveal Don's internal testing results of strength, but
just let's say it's gonna be a real treat when he releases it to the public.

Best,
Martin
Nice to see somebody take the Ippo* code to the next level.
That is really insulting to Don's hard work, and unwarranted.
:?:

yanquis1972
Posts: 1766
Joined: Tue Jun 02, 2009 10:14 pm

Re: Stunning Komodo

Post by yanquis1972 » Tue Mar 15, 2011 8:48 am

in other words, he didn't use ippo code. even if he did, i don't see how it's 'next level'. i personally wouldn't mind don using some kind of ippo/rybka search, but i'm quite sure he won't do that.

benstoker
Posts: 342
Joined: Tue Jan 19, 2010 1:05 am

Re: Stunning Komodo

Post by benstoker » Tue Mar 15, 2011 6:52 pm

fern wrote:Have you any proof that Dan has used Ippo code in the way you suggest? If not, better keep yourt mouth shut because otherwise you are commiting an insult.

Fern
Wow. It's all about "originality", right? And "originality" has to do with IDEAS, not CODE, right? Thus, copying one-input-one-output functions doesn't penalize "originality", right? The money is in those one-input-possible-many-outputs functions, such as Search() and Evaluate(), right?

Kaufman has here made numerous inquiries about the mechanisms of Ippo's Search(). And he's talked about how they are exploring Ippo's Search(), and it's all-nodes and cut-nodes, etc; and how they have tried out various aspects of Ippo's Eval(), etc. (as well as Stockfish's) in Komodo and how it works or doesn't work for them. Kaufman wants to understand what makes Ippo tick. What's wrong with that? Komodo apparently uses ALL of the same (and surely more) Eval() params Kaufman used in R3 and therefore Ippo*.

Thus, I hope they eventually do figure Ippo's IDEAS out and kick some Komodo ass. So, put a sock in it, cowboy.

Dann Corbit
Posts: 11622
Joined: Wed Mar 08, 2006 7:57 pm
Location: Redmond, WA USA
Contact:

Re: Stunning Komodo

Post by Dann Corbit » Tue Mar 15, 2011 7:35 pm

benstoker wrote:
fern wrote:Have you any proof that Dan has used Ippo code in the way you suggest? If not, better keep yourt mouth shut because otherwise you are commiting an insult.

Fern
Wow. It's all about "originality", right? And "originality" has to do with IDEAS, not CODE, right? Thus, copying one-input-one-output functions doesn't penalize "originality", right? The money is in those one-input-possible-many-outputs functions, such as Search() and Evaluate(), right?

Kaufman has here made numerous inquiries about the mechanisms of Ippo's Search(). And he's talked about how they are exploring Ippo's Search(), and it's all-nodes and cut-nodes, etc; and how they have tried out various aspects of Ippo's Eval(), etc. (as well as Stockfish's) in Komodo and how it works or doesn't work for them. Kaufman wants to understand what makes Ippo tick. What's wrong with that? Komodo apparently uses ALL of the same (and surely more) Eval() params Kaufman used in R3 and therefore Ippo*.

Thus, I hope they eventually do figure Ippo's IDEAS out and kick some Komodo ass. So, put a sock in it, cowboy.
I agree that there is nothing wrong with using ideas.
Further, Larry Kaufman worked with the Rybka team in designing the evaluation, so ideas like material imbalance:
http://home.comcast.net/~danheisman/Art ... alance.htm
are going to be found in both places, but Don will implement things in his own way.

The tone of your original post was simlar to the tone of your response "put a sock in it, cowboy".

There is a big difference from:
"Perhaps the Komodo team has learned something from the Ippo* design"
and
"Nice to see somebody take the Ippo* code to the next level."

Actually, I think that the impact of the Ippo* code so far is very small. People have not had enough time to learn from it. The impact of Fruit and Stockfish is far, far greater (so far).

benstoker
Posts: 342
Joined: Tue Jan 19, 2010 1:05 am

Re: Stunning Komodo

Post by benstoker » Tue Mar 15, 2011 8:02 pm

Dann Corbit wrote:
benstoker wrote:
fern wrote:Have you any proof that Dan has used Ippo code in the way you suggest? If not, better keep yourt mouth shut because otherwise you are commiting an insult.

Fern
Wow. It's all about "originality", right? And "originality" has to do with IDEAS, not CODE, right? Thus, copying one-input-one-output functions doesn't penalize "originality", right? The money is in those one-input-possible-many-outputs functions, such as Search() and Evaluate(), right?

Kaufman has here made numerous inquiries about the mechanisms of Ippo's Search(). And he's talked about how they are exploring Ippo's Search(), and it's all-nodes and cut-nodes, etc; and how they have tried out various aspects of Ippo's Eval(), etc. (as well as Stockfish's) in Komodo and how it works or doesn't work for them. Kaufman wants to understand what makes Ippo tick. What's wrong with that? Komodo apparently uses ALL of the same (and surely more) Eval() params Kaufman used in R3 and therefore Ippo*.

Thus, I hope they eventually do figure Ippo's IDEAS out and kick some Komodo ass. So, put a sock in it, cowboy.
I agree that there is nothing wrong with using ideas.
Further, Larry Kaufman worked with the Rybka team in designing the evaluation, so ideas like material imbalance:
http://home.comcast.net/~danheisman/Art ... alance.htm
are going to be found in both places, but Don will implement things in his own way.

The tone of your original post was simlar to the tone of your response "put a sock in it, cowboy".
"Nice to see someone take ippo to the next level" != ("Shut up" || "put a sock ..")
"shut up" == "put a sock ..."
There is a big difference from:
"Perhaps the Komodo team has learned something from the Ippo* design"
and
"Nice to see somebody take the Ippo* code to the next level."

Actually, I think that the impact of the Ippo* code so far is very small. People have not had enough time to learn from it. The impact of Fruit and Stockfish is far, far greater (so far).
Okay. Let's move on. I would imagine that the near chaos of the Ippo* code, and void of any commentary, explains why ppl haven't learned a lot from the ippo code. But, hats off to Komodo devs for trying.

User avatar
slobo
Posts: 2331
Joined: Mon Apr 09, 2007 3:36 pm

Re: Stunning Komodo

Post by slobo » Tue Mar 15, 2011 10:01 pm

Dann Corbit wrote:
benstoker wrote:
fern wrote:Have you any proof that Dan has used Ippo code in the way you suggest? If not, better keep yourt mouth shut because otherwise you are commiting an insult.

Fern
Wow. It's all about "originality", right? And "originality" has to do with IDEAS, not CODE, right? Thus, copying one-input-one-output functions doesn't penalize "originality", right? The money is in those one-input-possible-many-outputs functions, such as Search() and Evaluate(), right?

Kaufman has here made numerous inquiries about the mechanisms of Ippo's Search(). And he's talked about how they are exploring Ippo's Search(), and it's all-nodes and cut-nodes, etc; and how they have tried out various aspects of Ippo's Eval(), etc. (as well as Stockfish's) in Komodo and how it works or doesn't work for them. Kaufman wants to understand what makes Ippo tick. What's wrong with that? Komodo apparently uses ALL of the same (and surely more) Eval() params Kaufman used in R3 and therefore Ippo*.

Thus, I hope they eventually do figure Ippo's IDEAS out and kick some Komodo ass. So, put a sock in it, cowboy.
I agree that there is nothing wrong with using ideas.
Further, Larry Kaufman worked with the Rybka team in designing the evaluation, so ideas like material imbalance:
http://home.comcast.net/~danheisman/Art ... alance.htm
are going to be found in both places, but Don will implement things in his own way.

The tone of your original post was simlar to the tone of your response "put a sock in it, cowboy".

There is a big difference from:
"Perhaps the Komodo team has learned something from the Ippo* design"
and
"Nice to see somebody take the Ippo* code to the next level."

Actually, I think that the impact of the Ippo* code so far is very small. People have not had enough time to learn from it. The impact of Fruit and Stockfish is far, far greater (so far).
Of course, comrade Lenin.
"Well, I´m just a soul whose intentions are good,
Oh Lord, please don´t let me be misunderstood."

User avatar
fern
Posts: 8755
Joined: Sun Feb 26, 2006 3:07 pm

Re: Stunning Komodo

Post by fern » Tue Mar 15, 2011 10:32 pm

It depends how you define ideas, a very general term indeed. Besides, you does not have, also, any "idea" about which ideas Don used.
Specifically cloning or copying has only one referent, objetive referent: codes lines. If you go further you fall in the middle of a very fussy terrain.

Don was creating engines long before those guys in the Septembrist movement were born. If someone got "ideas" from another guy, they did with Don and the rest of the known an respected programers.

Fern

Post Reply