TCEC Season 7 - Main Thread

[M ANSARI]

[snip]
If Houdart wanted to provide a development/beta version of Houdini 5 he could have simply replied to one of my 4-5 emails sent to him the last few months. I have not had a single reply from him, which is sort of unprofessional. And besides, TCEC will not wait for engines to be released - there will always be new versions released.
[/snip]

Let me re-write that, "If Houdart wanted to provide a development/beta version of Houdini 5 he would have." Robert didn't want to make a public statement and didn't want to participate in this season's TCEC. Where is the confussion?

Martin, you're welcome to feel slighted he ignored your emails but do you need to slander him by calling the guy 'unprofessional'? Ignoring emails from an invitation-only tournament with no prize fund seems fair. TCEC has no contract with Houdart and Robert, has no professional obligation to TCEC.

As important as one may think the TCEC tournament is, it's far-fetched to think all engine authors must be at TCEC's beck-and-call.

That all said, I really do look forward to this year's tournament and appreciate your organizing it.

What you are missing is how enthusiastically Houdart was participating in the TCEC tournaments when Houdini was favored to win. Personally I think that Houdart has moved on as there is no longer an incentive to make a stronger Houdini as the effort to compete against SF and Komodo are too great with too little gain. This is also what happened with Vas and Rybka. My guess is that that they are both working on other projects and chances are that in the case of Houdart, it has nothing to do with chess.

[shrapnel]

+1.
My feeling also, Houdart has moved on to greener pastures, where his obvious talents will be better rewarded !

[Modern Times]

+1.
My feeling also, Houdart has moved on to greener pastures, where his obvious talents will be better rewarded !

+2
And same for Vas as well

[kranium]

Hi Martin,

hm ... but Houdini comes from the work the Fire programmer do.
Which one is the clone?

http://users.telenet.be/chesslogik/images/6xyk.jpg

I understand nothing, playing strength alone isn't all. Each programmer can find out improvements ... have a look in ICE ... in my opinion no reason to Support "Houdini".

But it's your tourney!

Allways the same ...
In two years nobody will speak about Houdini, Houdini and Rybka ... now we have two of this examples. A third will come and the Situation will be the same. Again, that is allways the same.

Such People, working with sources of others ... such people coming and going. So many examples we have in computer chess. Seems to be an event for so many computer chess fans.

Not very important ...

Enough nice development are included.
Good luck for all participant programmers and thanks for the work you do for us!

Best
Frank

Very interesting chart, did you make it or did you use a tool? Are there instructions on how to interpret the chart?

I did that, Ola Hansson added the labels, and then the chart ended in chesslogik website, which is what Frank links. See the original for comments:
http://www.talkchess.com/forum/viewtopi ... 52&start=0

The data was collected by T. Miller as the link provided

more details on how to interpret these plots here
http://www.talkchess.com/forum/viewtopi ... similarity

Miguel

Great work to everyone involved. I hope someone will update it with newer engines such as Stockfish 5 and Komodo 8. I'd also like to see several Rybka versions included.

After reading the instructions, I still don't understand what the numbers at the end of the lines represent. For example why does Onno have a 34 and Fruit 2.2.1 have a 44? From what I read Onno is similar to Fruit 2.2.1 but what do those numbers mean?

Thanks much for creating it Miquel...
I see it as a beautiful mandala of the chess engine circle of life.

Basically, the lower the number is, the more similar that individual of the 'branch' is to the origin (tip)...
Another way to look at is: the closer to the center the better.

Some observations on the chart:
Onno appears to be very closely related to Fruit, that's also well known from CCRL ponderhit stats I believe
Critter and Strelka are very closely related to Houdini (I think it's fair to say that both likely used RE to achieve this result)
Vitruvius is very closely related to Ivanhoe

I completely agree with Martin's decision to not include Fire, it's still too close to Ivanhoe and the rest...
although I have no issue with derivative engines, requiring some degree of originality is a good thing for all.


PS
In case anyone is interested about Fire development:
I promised to release 3.2 sometime this summer but backtracked because of instability and bugs...my apologies for that.

I have been busy re-writing, re-engineering it (8-12 hours per day, which I can afford to do as I'm retired) from the ground up since Fire 3.0 was released last Dec.,
in an effort to clean Ippo from it and create something really original.

24/7 development testing is running on 3 i7 3930K's
2 of the machines were purchased just this year in an effort to provide ample testing cores. I've been using 36 cores 24/7.

I've successfully implemented syzygy bases during the summer, which was a challenge, but now seem to be working well (no crashes for many weeks now).
Most recently I've completed a new and unique SMP implementation (far lighter and faster than the robust yet heavy 800 line Ivanhoe implementation).

I hope to release Fire 4.0 by end of year (or perhaps a bit later if need be)...at that point I'm quite confident it will be strong, and original at the source code level (and probably will deserve a new name).

Norm

[michiguel]

Hi Martin,

hm ... but Houdini comes from the work the Fire programmer do.
Which one is the clone?

http://users.telenet.be/chesslogik/images/6xyk.jpg

I understand nothing, playing strength alone isn't all. Each programmer can find out improvements ... have a look in ICE ... in my opinion no reason to Support "Houdini".

But it's your tourney!

Allways the same ...
In two years nobody will speak about Houdini, Houdini and Rybka ... now we have two of this examples. A third will come and the Situation will be the same. Again, that is allways the same.

Such People, working with sources of others ... such people coming and going. So many examples we have in computer chess. Seems to be an event for so many computer chess fans.

Not very important ...

Enough nice development are included.
Good luck for all participant programmers and thanks for the work you do for us!

Best
Frank

Very interesting chart, did you make it or did you use a tool? Are there instructions on how to interpret the chart?

I did that, Ola Hansson added the labels, and then the chart ended in chesslogik website, which is what Frank links. See the original for comments:
http://www.talkchess.com/forum/viewtopi ... 52&start=0

The data was collected by T. Miller as the link provided

more details on how to interpret these plots here
http://www.talkchess.com/forum/viewtopi ... similarity

Miguel

Great work to everyone involved. I hope someone will update it with newer engines such as Stockfish 5 and Komodo 8. I'd also like to see several Rybka versions included.

After reading the instructions, I still don't understand what the numbers at the end of the lines represent. For example why does Onno have a 34 and Fruit 2.2.1 have a 44? From what I read Onno is similar to Fruit 2.2.1 but what do those numbers mean?

BTW, Kai & Adam collected even more data, which has been posted here at one point. Adam's analysis has been more extensive, checking three digit numbers of engines.

Miguel

[Graham Banks]

...In case anyone is interested about Fire development:
I promised to release 3.2 sometime this summer but backtracked because of instability and bugs...my apologies for that.

I have been busy re-writing, re-engineering it (8-12 hours per day, which I can afford to do as I'm retired) from the ground up since Fire 3.0 was released last Dec.,
in an effort to clean Ippo from it and create something really original.

24/7 development testing is running on 3 i7 3930K's
2 of the machines were purchased just this year in an effort to provide ample testing cores. I've been using 36 cores 24/7.

I've successfully implemented syzygy bases during the summer, which was a challenge, but now seem to be working well (no crashes for many weeks now).
Most recently I've completed a new and unique SMP implementation (far lighter and faster than the robust yet heavy 800 line Ivanhoe implementation).

I hope to release Fire 4.0 by end of year (or perhaps a bit later if need be)...at that point I'm quite confident it will be strong, and original at the source code level (and probably will deserve a new name).

Norm

Looking forward to it.

[Frank Quisinsky]

Hi Norman,

I wrote it to you for around 4 years!

It made more sense if you develops your own without IPP code. After all what I read and saw from yourself it's more interesting if you use all the programming knowledge for your own ideas. In my opinion you are one of the TOP programmers today and your skills must be enormous.

You can be sure that I will test your material.
I am very interesting in your work.

It's more and more important that we collect good works because we life in an "Illusory World". Computer chess is a paragon for it.

Keep up your good work!
And thanks that you do it, one of the best computer chess messages this year for me.

Give me your address (private message) and you will get the obligatory coffee cup from myself. Send it three times in 37 years computer chess. Biggest compliment I can do.

With my coffee cup your new engine will be 50 Elo stronger, believe me ... more good ideas you can create.

Best
Frank

[kranium]

Give me your address (private message) and you will get the obligatory coffee cup from myself.
Send it three times in 37 years computer chess. Biggest compliment I can do.

With my coffee cup your new engine will be 50 Elo stronger, believe me ... more good ideas you can create.

Best
Frank

Thx!
I need a really big cup Frank...!

(and so do any others developers trying to compete)
The challenge for an individual programmer today is huge:

Robert H. has the support of an Arab sheikh (Ahmed M.), with remote access to many many cores in IBM's Dubai data center
Stockfish has Marco C. and hundreds of talented people contributing, and access to even more processing power
Komodo has ? well Larrk K. for one! (a computer Grandmaster wizard), and of course Mark L., and others coding

But in the end, there's one very important thing IMO that can (possibly) level the playing field: hard work

[Leto]

Thanks much for creating it Miquel...
I see it as a beautiful mandala of the chess engine circle of life.

Basically, the lower the number is, the more similar that individual of the 'branch' is to the origin (tip)...
Another way to look at is: the closer to the center the better.

Some observations on the chart:
Onno appears to be very closely related to Fruit, that's also well known from CCRL ponderhit stats I believe
Critter and Strelka are very closely related to Houdini (I think it's fair to say that both likely used RE to achieve this result)
Vitruvius is very closely related to Ivanhoe

I completely agree with Martin's decision to not include Fire, it's still too close to Ivanhoe and the rest...
although I have no issue with derivative engines, requiring some degree of originality is a good thing for all.


PS
In case anyone is interested about Fire development:
I promised to release 3.2 sometime this summer but backtracked because of instability and bugs...my apologies for that.

I have been busy re-writing, re-engineering it (8-12 hours per day, which I can afford to do as I'm retired) from the ground up since Fire 3.0 was released last Dec.,
in an effort to clean Ippo from it and create something really original.

24/7 development testing is running on 3 i7 3930K's
2 of the machines were purchased just this year in an effort to provide ample testing cores. I've been using 36 cores 24/7.

I've successfully implemented syzygy bases during the summer, which was a challenge, but now seem to be working well (no crashes for many weeks now).
Most recently I've completed a new and unique SMP implementation (far lighter and faster than the robust yet heavy 800 line Ivanhoe implementation).

I hope to release Fire 4.0 by end of year (or perhaps a bit later if need be)...at that point I'm quite confident it will be strong, and original at the source code level (and probably will deserve a new name).

Norm

Hopefully if you rename it and people decide that it's original enough perhaps CEGT will allow it to be tested.

[Frank Quisinsky]

Hi Norman,

you don't need a Arab sheikh, hundreds of talented people or a grandmaster. You can do it alone and finaly ... sheikh, grandmasters and all the others are jealous. You have to build on the abilities you has.

Thinking on Capablanca!
His skills no Arab sheikh can purchase with his luck to sit on an oel field.

Only one point I envy an Arab sheikh ...
Not say that to the others please ...

The harem a Arab sheikh have (not all the Computers or the boring money).
I will have such a harem too, sorry!

I am sure your result will be very good Norman!

Best
Frank