Critter stronger than Naum?

[mhalstern]

It is too late for me to edit the post. I intended to write, I an NOT disputing the ipon list.

[Dann Corbit]

Lots of people live in the dark ages, which is to say, they have a 32 bit CPU and operating system and only one core.

That's all that a lot of people need, so I wouldn't castigate them for it.

It was intended to be humorous. I suppose I should have included a smiley.

[Graham Banks]

Lots of people live in the dark ages, which is to say, they have a 32 bit CPU and operating system and only one core.

That's all that a lot of people need, so I wouldn't castigate them for it.

It was intended to be humorous. I suppose I should have included a smiley.

[gerold]

http://www.inwoba.de/
Look at the rating list.

It's the kind of thing that could make a person go slack-jawed, buggy-eyed.

Indeed! My early results show it's close behind, if not equal to Rybka, but still a tad weaker than Houdini. However, Critter pulls off some amazing attacks. An engine to watch for sure. Do we know if Critter is a original program ?

Indeed a tad weaker than Houdini.
On Dual core. 5/3 TC.
Houdini vs. Critter. 29.5/45. 19-5-21. 66%. +139.

[Jouni]

I am very impressed by Critter 0.90 It's better than 0.80 in ALL testsuites so far (tactics, positional and endgame) even if 0.80 already
is very good!

Jouni

[De Vos W]

Re: Critter stronger than Naum?
Look at the rating list.
http://www.inwoba.de/

It's the kind of thing that could make a person go slack-jawed, buggy-eyed.

Why?

That's my way of saying:
I am very impressed.

Oh.
I am too. Without implying anything indirectly, it's a fact these days that new programs are hiking to top places every once in a while.

In the "older" days, this wasn't happening by any means and we had a clear distinction in strength terms to professionals and amateurs where the pros were much stronger and couldn't be reached, e.g Fritz, Shredder, Junior, Hiarcs, Rebel, Tiger all had a league of their own and the others just followed. And situation went this way for years and years.

Right now every single while a new engine hikes and takes a place in the top positions.
I think everything started with Fruit's 2.1 open code , didn't they?

Fruit was clearly a revolution.
I think also that Stockfish has great ideas in it.
Also, testing procedures have improved greatly.
Still, it is amazing when someone who fiddles around for a hobby can club professionals who do it for a living (I saw another list where Critter tops Fritz 12 also).

When Ippo-Families came out the source code was immediately exposed. From that time all Chess-programmers had a brilliant open source to work from.

[Dr.Wael Deeb]

Re: Critter stronger than Naum?
Look at the rating list.
http://www.inwoba.de/

It's the kind of thing that could make a person go slack-jawed, buggy-eyed.

Why?

That's my way of saying:
I am very impressed.

Oh.
I am too. Without implying anything indirectly, it's a fact these days that new programs are hiking to top places every once in a while.

In the "older" days, this wasn't happening by any means and we had a clear distinction in strength terms to professionals and amateurs where the pros were much stronger and couldn't be reached, e.g Fritz, Shredder, Junior, Hiarcs, Rebel, Tiger all had a league of their own and the others just followed. And situation went this way for years and years.

Right now every single while a new engine hikes and takes a place in the top positions.
I think everything started with Fruit's 2.1 open code , didn't they?

Fruit was clearly a revolution.
I think also that Stockfish has great ideas in it.
Also, testing procedures have improved greatly.
Still, it is amazing when someone who fiddles around for a hobby can club professionals who do it for a living (I saw another list where Critter tops Fritz 12 also).

When Ippo-Families came out the source code was immediately exposed. From that time all Chess-programmers had a brilliant open source to work from.

A fact,yes....Even Vasik can get some benefits,as usual,without rising so much noise....
Dr.D

[Robert Flesher]

I am very impressed by Critter 0.90 It's better than 0.80 in ALL testsuites so far (tactics, positional and endgame) even if 0.80 already
is very good!

Jouni

Perhaps then Vas can read these sources and give us a bug fix. But, also a Rybka 4.0 version that understands basic chess endings would be nice.

[IWB]

Hi Marc,

I was looking at the rules/conditions of the Ipon rating lists, and it states that all engines run on 1 core. Is this an accurate test? Is it possible the Naum either scales better than Critter, or for other reasons, Naum with 4 cores > Critter with 4 cores?

I am disputing the Ipon rating list, just questioning it. Has anybody taken the time to compare the results of engine matches, using 1 core, against matches with the exact conditions, except the engines use 4 cores?

Why do the ipon people test with 1 core - can they get in 4x the amount of games in the same amount of time this way?

The IPON started years ago as a beta test list. The intension was and still is to measure strength increases between different versions and of course ponder on (as pondering is the normal way of playing chess ...).
Now I did some testing with 2 cores in the past and found out that, even if the engines are scaling a bit different, none of the engines is getting into a new "ball park" (is that the right expression). It might be that one engine going from 1 to 4 cores is only getting 80 Elo, while another gains 120. The difference (40) is, because of the limited number of games, most of the time in the error bar anyhow. Is that worth testing?
The second thing is, that the Top 10-15 Engines in my list will beat any human at any time control on one core as well (when played a reasonable number of games). The result of a 4 core test is totaly artifical as WE hujmans can not really judge about it. Is that worth testing?
And last, if a normal MP implementation is already there, the time invested into a code to sqeeze out another 10 ELo in MP is better invested in the enigne itself. So again: Is it worth testing?

That are three reasons why MP testing is less important to me ("Me" are the 'people of IPON' you are refering to ) This might be different if no one would test 4 cores and yes, I compare all lists to mine on a regular basis. None of the 4 core results really is totaly bogus (or vice versa) compared to my list. There is no big difference (the 40 ELo from abouve are reached nowhere!). Of course the fact, that I can do a resonable number of games in the same time I would do an MP test helps to make the IPON more accuate than others as well ...

The current result in my list sees Critter ~70 Elo in front of Naum 4.2 I doubt, that in a propper match vs many opponents Naum can catch Critter 0.9 on 4 cores even if the MP implementation into naum is one of the better ones (scroll down my news section to 2010.08.22, you will find my latest 2 core results right before I removed the games from the IPON). Have a look in my individual results. Critter 0.9 beats Stockfish 1.9.1, nonetheless it could not pass stockfish at the end ... Wait until there are 1000 4 core games (at best vs the same opponents as 4 core Naum 4.2) for Critter 0.9 and then have a look again.

Regards
Ingo

PS; I personaly like it if someone is questioning conditions, that is the reason why I made and publish my own list ... I was and I am not happy with the old established lists ... but I am pretty sure that I will go that way too .. sooner or later

[Sylwy]

Perhaps then Vas can read these sources and give us a bug fix......

Oh my God ! I can't belive ! Piracy ?

S