Later Tonight_1st Update- Houdini_Rainbow UNLimited

[geots]

3 or 4 games to be played and this match will have reached the 50 game mark- which means time for Update 1.

Again, in response to Uri- UNLimited is No.2. While other matches were running, I had more than enough time to run enough matches ag. Stockfish, Komodo, Rybka, Critter and Strelka to be sure of the statement I was making. As for the brilliant thought Uri came up with- that I did not have access to some private version or beta that Don has of Komodo, so it did not beat some version that "might or might not exist"- you can only test what you can load and play. Possibilities and promises don't contain executables. Someone explain that to Uri when you have some free time.

As for the other guy, sometimes it is better to be silent and just let some people think you are not the sharpest knife in the drawer- instead of opening your mouth and proving it to everyone.


At any rate- Houdini 2.0c x64 vs Rainbow UNLimited- 1st Update shortly.



george

[geots]

Houdini 2.0c x64 vs Rainbow UNLimited


Time- I have plenty of. Engines that might have a shot at Houdini- what you see is what I have left. I have thrown everything at him but the kitchen sink, and so far he has met every challenge. But we shall see- a lot can happen in 450 games.


Intel i5 w/4TCs
Fritz 13 gui
1CPU/64bit
128MB hash
Bases=NONE
Ponder_Learning=OFF
Perfect 12.32 book w/12-move limit
5'+5"
Match=500 games

Code: Select all

Houdini 2.0c x64     +35    +17/-12/=21   55.00%   27.5/50
Rainbow UNLimited    -35    +12/-17/=21   45.00%   22.5/50

Tonight-

george

[geots]

Houdini 2.0c x64 vs Rainbow UNLimited


I made the comment to someone after the 1st update that if Rainbow did not find a way to cut into Houdini's 5 game lead before he fell further behind- the writing was on the wall. Well, Rainbow in fact did not find a way. 390 games left and I am already calling it- doesn't make much sense- does it.



Intel i5 w/4TCs
Fritz 13 gui
1CPU/64bit
128MB hash
Bases=NONE
Ponder_Learning=OFF
Perfect 12.32 book w/12-move limit
5'+5"
Match=500 games

Code: Select all

Houdini 2.0c x64     +41    +35/-22/=53   55.90%   61.5/110
Rainbow UNLimited    -41    +22/-35/=53   44.10%   48.5/110

I have given this engine, along with its 2 betas, every opportunity I could to let it show me it can stay with Houdini, and the match is still running as I post this thread. Best I can say is "seeing is believing."




In search- somewhere- of a photo-finish,

george

[velmarin]

George,



Perhaps you are testing a "Houdini modified" with HoudiConfig or HEX.
A month ago someone who is dedicated to the "Houdini modified" signed up a forum in Spanish as "Rainbowfly".

Maybe just a coincidence.
Regards, George.

[geots]

George,



Perhaps you are testing a "Houdini modified" with HoudiConfig or HEX.
A month ago someone who is dedicated to the "Houdini modified" signed up a forum in Spanish as "Rainbowfly".

Maybe just a coincidence.
Regards, George.

No, my friend. I actually only have a vague idea of "HoudiConfig"- not really understanding the concept. I am just not into Houdini itself enough for that- except for how it might be beaten.

It would not surprise me tho, if you were right- but in the opposite from your thinking- that someone did sign up to some group and got the idea from me using the name here. That might be possible.

As for trying to bump up Houdini's elo- I doubt Robert needs any help from any of us there.



Best, (and good luck with your program)

george

[Uri Blass]

Sorry but something that is 41 elo weaker than houdini2c based on your result cannot be clearly number 2 and I see that you test 1 cpu 64 bits(maybe it is less than 41 elo weaker but you have not enough games to know).

From the CEGT rating list
http://www.husvankempen.de/nunn/40_40%2 ... liste.html

17 Houdini 1.5a x64 1CPU 3013 14 14
19 Houdini 2.0c x64 1CPU 3002 15 15
27 Critter 1.6 x64 1CPU 2989 15 15
30 Komodo 4.0 x64 1CPU 2981 13 13
35 Stockfish 2.2.2 x64 1CPU 2973 17 17
Houdini1.5a that seems to be the strongest houdini for 1 cpu and not the houdini2c is only 24 elo better than Critter and 32 elo better than Komodo and 40 elo better than Stockfish.

Houdini is only slightly better than Critter Komodo and Stockfish(excpet very fast time control that is not interesting) and you need at least 1000 games at 40/20 to be convinced that houdini is stronger than Critter when we talk about houdini1.5a and for houdini2c it seems that 1000 games are not enough to be convinced that it is stronger than Critter(there is a possible error of 15 elo when the difference between houdini2 and Critter1.6 is only 13 elo).

If you want to be sure that something is stronger than Critter and weaker than houdini at 40/20 CEGT time control then you probably need many thousands of games.

[geots]

Sorry but something that is 41 elo weaker than houdini2c based on your result cannot be clearly number 2 and I see that you test 1 cpu 64 bits(maybe it is less than 41 elo weaker but you have not enough games to know).

From the CEGT rating list
http://www.husvankempen.de/nunn/40_40%2 ... liste.html

17 Houdini 1.5a x64 1CPU 3013 14 14
19 Houdini 2.0c x64 1CPU 3002 15 15
27 Critter 1.6 x64 1CPU 2989 15 15
30 Komodo 4.0 x64 1CPU 2981 13 13
35 Stockfish 2.2.2 x64 1CPU 2973 17 17
Houdini1.5a that seems to be the strongest houdini for 1 cpu and not the houdini2c is only 24 elo better than Critter and 32 elo better than Komodo and 40 elo better than Stockfish.

Houdini is only slightly better than Critter Komodo and Stockfish(excpet very fast time control that is not interesting) and you need at least 1000 games at 40/20 to be convinced that houdini is stronger than Critter when we talk about houdini1.5a and for houdini2c it seems that 1000 games are not enough to be convinced that it is stronger than Critter(there is a possible error of 15 elo when the difference between houdini2 and Critter1.6 is only 13 elo).

If you want to be sure that something is stronger than Critter and weaker than houdini at 40/20 CEGT time control then you probably need many thousands of games.

Uri, you are supposed to ask before you come into my sandbox and play with my toys. I might tell on you.

gts

[geots]

Sorry but something that is 41 elo weaker than houdini2c based on your result cannot be clearly number 2 and I see that you test 1 cpu 64 bits(maybe it is less than 41 elo weaker but you have not enough games to know).

From the CEGT rating list
http://www.husvankempen.de/nunn/40_40%2 ... liste.html

17 Houdini 1.5a x64 1CPU 3013 14 14
19 Houdini 2.0c x64 1CPU 3002 15 15
27 Critter 1.6 x64 1CPU 2989 15 15
30 Komodo 4.0 x64 1CPU 2981 13 13
35 Stockfish 2.2.2 x64 1CPU 2973 17 17
Houdini1.5a that seems to be the strongest houdini for 1 cpu and not the houdini2c is only 24 elo better than Critter and 32 elo better than Komodo and 40 elo better than Stockfish.

Houdini is only slightly better than Critter Komodo and Stockfish(excpet very fast time control that is not interesting) and you need at least 1000 games at 40/20 to be convinced that houdini is stronger than Critter when we talk about houdini1.5a and for houdini2c it seems that 1000 games are not enough to be convinced that it is stronger than Critter(there is a possible error of 15 elo when the difference between houdini2 and Critter1.6 is only 13 elo).

If you want to be sure that something is stronger than Critter and weaker than houdini at 40/20 CEGT time control then you probably need many thousands of games.

Uri, you are supposed to ask before you come into my sandbox and play with my toys. I might tell on you.

gts

Seriously- see, you start off with a screwy assumption. So that means everything after your first sentence is garbage as far as it concerns me. What is 41 elo weaker than Houdini 2.0c based on my tests?
I have run exactly 110 games with this version against Houdini 2.0c. So how from these 110 games do you come up with the assumption that this version is 41 elo weaker? What do you base that statement on? You said "based on my results". Exactly which results do you refer to?

This is what headaches are made of. Can't you find someone else to worry the shit out of?


gts

[geots]

Uri, obviously you "don't know nothing about nothing." Based on that- would you please stay out of my business and no longer respond to any of my threads. I am doing nothing short of completely wasting my time with you.


george