My Problem Right Now With The Ivanhoes!!??

[geots]

After the 3 matches that run while I sleep- and post tomorrow- I will have concluded the tests with Komodo, Stockfish and Rybka each playing all 15 Ivanhoe versions in 50 game matches. The PGN sets for the Komodo and Stockfish complete match sets are still available thru the links I have posted- and will be for another week along with the Rybka set I will post upon completion of those matches. I thought it best to just leave all 3 side by side for 7 days.

As most know, this is an effort to rate the Top 10 or Top 12 Ivanhoe versions. To even get a more accurate result, I have been tossing around the idea of a 4th opponent for the Ivanhoes. Actually that was my initial intent- as that would give each Ivanhoe version 200 games.

But the 4th version has always been the sticking point. These 3- Komodo, Stockfish and Rybka were ideal. Win some-lose some. On the bottom end, I cannot come up with a version that can give them much competition. For instance- one version struggling to make it to the Top 10 or 12 has beaten Fritz 13 by the score of 27 - 2. On the other end, they will have little or no success with Houdini and Strelka- and not likely Critter to an extent. And RobboLito is certainly a no-no. That's family- results worthless there.

I don't know- it's a damn shame. I really needed one more engine.


gts

[ZirconiumX]

And you haven't looked at Shredder?

Matthew:out

[rodolfoleoni]

What about using a weakened personality of Houdini? Maybe, with the utility by Mathias Gemuh...

[geots]

And you haven't looked at Shredder?

Matthew:out

Well, I am not quite sure Shredder 12 could win a match, unless maybe I used Deep Shredder 12 and ran him on 2CPUs. That is a thought. I don't guess the world would blow up if I borrowed Deep Shredder from someone just for this one test. But no, that would not work- because I need to run 3 matches at once. I just don't know. Thanks, tho.

gts

[geots]

What about using a weakened personality of Houdini? Maybe, with the utility by Mathias Gemuh...

You are probably on to something- but there is something about handicapping an engine I don't like. I can't explain it.

[ZirconiumX]

Run a 100 match tournament between shredder and IvanHoe.

If IvanHoe dominates, then drop Shredder.
If Shredder dominates, then drop Shredder.
If the score is about even +-10 then keep Shredder.

I have just looked at the CCRL ratings. IH is about even versus Critter, though the LOS chart begs to differ.

Matthew:out

[geots]

Run a 100 match tournament between shredder and IvanHoe.

If IvanHoe dominates, then drop Shredder.
If Shredder dominates, then drop Shredder.
If the score is about even +-10 then keep Shredder.

I have just looked at the CCRL ratings. IH is about even versus Critter, though the LOS chart begs to differ.

Matthew:out

Look, I am running 50game matches with 15 different Ivanhoe versions ag. common opponents. How it does ag. the common opponent is not relevant- I am doing this to get a half-way decent rating lists for the Ivanhoes 64bit at 1CPU. I have used Komodo 4, Stockfish 2.2.2 and Rybka 4.1. I need a 4th, and tried looking at CCRL's list. But it confuses me. Granted I am running 40/3 repeating benchmarked for 40/4 repeating. And I was looking at 40/40 lists. But I don't care if it was 40 moves in 40 days- something is wrong.

I will tell you that if I run each of these 15 ag. Critter 1.4, there is better than an even chance that Critter would win all 15 matches. It is capable of it. Ag. Komodo 4- Komodo 4 won 5 matches out of 15 with Ivanhoe versions, drew 3 and lost 7. Now this is where common sense comes into play. Do you honestly think if I ran Komodo 3- not Komodo4- Komodo3- ag. all 15 Ivanhoes that it would do better than Komodo4. Not a chance in hell. So you explain to me how Komodo 3 is rated higher than Critter 1.4.

And look at the huge gap between Strelka 5.1 64bit and Critter 1.4 64bit. You run a 100 game match between Strelka 5.1 and Critter 1.4, anything from 40/4 to 40/40- I got good money that says Critter will kick the shit out of Strelka 5.1

So I am getting no help there. According to that, Komodo 3 would be a stronger opponent than Critter 1.4- at least =. And Komodo 3 would do much better than Stockfish 2.2.2? Bullshit.

I gotta think about this a bit- and I guess I better go ahead and post the final 3 matches ag. Rybka and show final results in totality and give a link to the PGNs.

gts