Milenium Genius News???

scandien · Post by **scandien** » Tue Sep 15, 2015 6:12 pm

The CENTURION is based on FRITZ 1 when MASTER CHESS and ATLANTA are based on Fritz 3.10 ( more solid in strategy). These are not the same program.

I think the MASTER CHESS and ATLANTA are running the same program, but i could check the effect of hash table.

At the beginning i was planning to used 30 minutes per player , but perhaps for this group ( the strongest) i will use 1 hour (it depend on my availability)

On other group i am using the 30' per player time control ( except when i am playing ( i am using at least one hour or one hour and an half)

BR

Nicolas

Post by **Fernando** » Tue Sep 15, 2015 6:20 pm

scandien wrote:The CENTURION is based on FRITZ 1 when MASTER CHESS and ATLANTA are based on Fritz 3.10 ( more solid in strategy). These are not the same program.

I think the MASTER CHESS and ATLANTA are running the same program, but i could check the effect of hash table.

At the beginning i was planning to used 30 minutes per player , but perhaps for this group ( the strongest) i will use 1 hour (it depend on my availability)

On other group i am using the 30' per player time control ( except when i am playing ( i am using at least one hour or one hour and an half)

BR

Nicolas

A< propos of Fritz, the first pure windows program by CB with Fritz was called Chessica. It came in a Cd with as chess tutor and the full package was called "Task Chess CD".
Chessica is strong and interesting.

Fern

mclane · Post by **mclane** » Tue Sep 15, 2015 6:29 pm

I will put the machine into my oldie tournament.

I guess there have been played 600 tournament games so far.

paulwise3 · Post by **paulwise3** » Tue Sep 15, 2015 7:26 pm

Fernando wrote:
paulwise3 wrote:When, oh when comes the postman to bring me this new toy...?
I will embrace it, cuddle it, and then...

My Mephisto Chess Challenger 32(or even 40?)MHz will jump on it...

Regards, Paul
Why you talk of the postman?
The product is no9t yet on sell
Or is it'?????

Terrified regards
Fern

Well ok, I ordered it directly via de Millennium website, and on my question then, they suggested it could be medio september. So, just...

Hoping for a planning to come true regards,
Paul

Post by **Fernando** » Tue Sep 15, 2015 7:28 pm

paulwise3 wrote:
Fernando wrote:
paulwise3 wrote:When, oh when comes the postman to bring me this new toy...?
I will embrace it, cuddle it, and then...

My Mephisto Chess Challenger 32(or even 40?)MHz will jump on it...

Regards, Paul
Why you talk of the postman?
The product is no9t yet on sell
Or is it'?????

Terrified regards
Fern
Well ok, I ordered it directly via de Millennium website, and on my question then, they suggested it could be medio september. So, just...

Hoping for a planning to come true regards,
Paul

JESUS!
I was expecting an official announcement of those guys.....So...

I go at once at the sire regards
Fern

Monsieur Plastique · Post by **Monsieur Plastique** » Tue Sep 15, 2015 11:27 pm

scandien wrote:The CENTURION is based on FRITZ 1 when MASTER CHESS and ATLANTA are based on Fritz 3.10 ( more solid in strategy). These are not the same program.

I think the MASTER CHESS and ATLANTA are running the same program, but i could check the effect of hash table.BR

Nicolas

Yes, I'm aware of the lineage of the machines, but in your post you mentioned you wanted to obtain a rating for Genius from the games you intend to play. You also intimate time is limited (obviously for anyone it is). The most accurate way to obtain a rating in your particular case would be to spread the number of games across the stronger machines so in the case of the Morsch programs - purely for rating purposes, it is better to just stick to the Atlanta. That way you have more games against opponents closer to the strength of the machine you are trying to rate which will result in a more accurate rating. 5 games "saved" against a Centurian and Master Chess means 5 games you can play against a Diamond II or the Fidelity instead. And taking off Centurion and Master Chess might also allow you to play the hour per side games rather than the 30 minutes per side ones.

The three Morsch machines you are using are all improvements upon one another and the extra "strategical" capabilities of the Master Chess (so more nodes to churn through each move) and Atlanta are facilitated by the higher CPU speed, meaning their tactical ability is also greater than Centurion. Which in turn means there is nothing a Centurion can do, tactically or otherwise, that a Master Chess can't, let alone an Atlanta.

I agree they are all good opponents to play out of interest, but not for strict rating purposes when it comes what will be a small number of games in the end from a statistical perspective. Sure, if you could play hundreds of games at slow time controls, all of them would be OK for rating purposes, but sometimes when time is limited - as it is with one person - and you want to obtain a reasonable ballpark rating as soon as possible with a limited number of games, the choice of opponents becomes more critical.

Monsieur Plastique · Post by **Monsieur Plastique** » Wed Sep 16, 2015 12:27 am

Sorry I was rushing a bit when I typed that reply and now cannot edit it. I meant of course that for a given program, all other things being equal adding strategical knowledge and heuristics reduces the nodes per second at a given amount of computational power. So although the Master Chess and Atlanta have added knowledge over the Centurion, they are running faster CPUs (and more RAM - in the case of Altanta obviously far more RAM) than the Centurion so that the more tactical based considerations and ply depth are not compromised as a result (if anything further improved, especially in the late endgame with Atlanta).

I ran across this exact same "Morsch" rating issue when I played a 100 game 40 moves in 2 hours rating matches with the Nintendo DS Fritz. The Nintendo scored 17.5 to 4.5 against Mephisto Miami (Morsch 16K at 10 Mhz), 9.5 to 10.5 against Mephisto Expert Travel (32K at 10 Mhz) and 2.5 to 3.5 against Master Chess (64K at 20 Mhz). Therefore the results were consistent with the Morsch programs being improvements upon one another with the scores being 79.55%, 47.50% and 41.67% against the three machines respectively. Therefore the best opponents to pick in that particular fight were Expert Travel and Master Chess. Miami would have distorted the rating favourably had I not played a very large number of games - had it just been 20, 40 or even 60 games, it would have had a significant impact on the final rating of the Nintendo. If I had been forced to play a limited number of games (I wasn't in this case and could manage the full 100) then I would have dropped the Miami results altogether.

scandien · Post by **scandien** » Wed Sep 16, 2015 5:52 am

I am agree with you.. To rate a machine we need many games versus a variety of opponent . I'll just introduce the CHESS GENIUS in my tournament (there is upper stage in it) without withdrawing any other machine and i'll see the result.

I am quite confident that, soon or later, i'll get relevant information about Millenium Chess Genius level. If this new Genius, get a good result in such a tournament it will probably be strong, it it is crushed ( why not) it will probably belong to another class.

Its level will be slowly adjusted with time ( and results from other fans of course

)

BR

Nicolas

Monsieur Plastique · Post by **Monsieur Plastique** » Wed Sep 16, 2015 9:26 am

I'm particularly interested to see how it goes against the London (and even if there are common moves between the two in similar positions) and also the Diamond II. With the hash table Novags (Diamond II, Star Diamond, etc) achieving high prices on eBay (and insane asking prices in some cases), it will be really interesting to see how the new machine scores against these Novags. One thing is for sure. The new machine is a bargain at any rate, but considering the hardware and the hash tables, even more so. Nearly 30 years ago I would have walked into the local chess shop and paid several thousand for a 32 bit Lang machine (albeit in wood). To think anyone can have a dedicated machine that plays even stronger than these old late 80s Mephistos at a tiny fraction of the price is even more amazing.

My feeling is that by the time the machine gains a true rating (I am sure the SSDF will officially rate it), it will probably be almost bang on the marketing estimate of 2000 (per the current SSDF scale).