Author: Christophe Theron
Date: 12:16:23 07/17/98
Go up one level in this thread
On July 17, 1998 at 13:03:10, Bruce Moreland wrote:
>On July 16, 1998 at 13:06:46, Christophe Theron wrote:
>
>>Game/30 with fisher 30s is fine. Could we all agree on this? I guess some will
>>have problems because they didn't implement fisher clock (it's my case!), but we
>>have plenty of time to do it before the tournament.
>
>30 30 is a fine time control, but I would like to mention a few things.
>
>Fischer time control is designed to allow good play in endings and prevent the
>necessity of adjournments. This has some drawbacks, in that a game can go on
>for a tremendous length of time. In computer chess you will sometimes see a 200
>move game with a decisive result. That's potentially over four hours.
>
>And if you are going to have a break for adjudication, you might as well set it
>up beforehand, and use a normal tournament time control that gets you there.
>
>The big advantage of normal tournament time controls is that the games can be
>stopped in finite time.
>
>If a game is still contentious at that point, it can be continued while everyone
>else is asleep. If it's not contentions, the tournament director can strongly
>suggest a resignation or draw offer, and if that doesn't work, he can simply end
>the game as he sees fit.
We could vote on that. CCC offers the opportunity to vote, and many chess
programmers could give their opinion. Personnaly I don't mind in fact. The most
important is that we have an event with enough rounds to get a reasonnably
reliable result.
>>Let's limit to 32 entries. Why not doing "qualification rounds" before the
>>event? Maybe we could ask several testers to run these rounds for amateur
>>programs. I'm thinking about Torsten Schoop for example, but there are others.
>>Maybe they would be glad to help.
>
>More players is more fun, but 32 is probably fun enough.
>
>I don't think anyone should have to send their program in to someone they don't
>know, in order to qualify.
>
>If this is held in the US, you'll probably lose some of the Germans, and that
>should keep the numbers down a bit.
OK, so maybe we don't have to decide who has to be excluded. It's even better.
>>Most of the programmers that have a "big iron" program have also a microcomputer
>>program. It is your case, Bob. I suppose Don also has a micro version of
>>CilkChess, or can easily do a one processor version.
>>
>>I don't know about you, but I'm personnaly mainly interested in knowing how my
>>PROGRAM compares to other programs, and not how my HARDWARE compares to others.
>
>In at least some of the tournaments, it should be absolutely anything goes. If
>this is one of those, no problem, I'll probably just show up with my Alpha.
I agree, there should still be an "anything goes" tournament. But without a
sponsor it is difficult.
So we are talking about a "nearly uniform platform" event, WMCPC, which could be
organized without any sponsor, as we already have a place to do it.
Would you like to enter WMCPC with Ferret?
>>That's why I would like to see something close to a uniform platform event. I
>>know this cannot be the case, because we already have at least 2 classes: PC
>>programs and Alpha programs.
>>
>>But maybe we could set a limit in the clock speed. Say 400MHz for x86 computers
>>and 600MHz for Alphas, so the championship does not turn into a race for the
>>fastest hardware. The numbers can be discussed.
>
>I will probably take issue with any concrete proposal to limit microcomputer
>hardware, because there will be problems with any way you can think of.
>
>I'd be happy to compete in a uniform hardware event, but we all know what the
>platform will be. It will be an Intel-compatible machine, probably slightly off
>the leading edge.
>
>Boy, everyone will flock to see that.
So let's do it.
It is impossible to organize a stricly uniform platform event, but it is
possible to get near if we accept, say, a 25% tolerance. If we are wise enough
to agree on the rules, it would be a very interesting event.
>>If we rule that the computers will be one processor running up to 400/600MHz, I
>>think we will have a better idea of who are the best programmers (hence the name
>>"World Microcomputer Chess Programmers Championship").
>
>I just thought of another problem with putting a mhz cap on things. If you put
>a mhz cap that is higher than that of the supplied machines, but lower than that
>of cutting edge machines, you may nail someone who has delayed a machine
>purchase.
>
>Let's say we put a 300 mhz cap on this event. That would suit me fine, I have a
>300 mhz machine that I bought last November, right after the prices started
>coming down. This machine was the best buy at that time.
>
>It would thoroughly screw you though. You have a 100 mhz machine. You need to
>upgrade. If you had a choice from all the machines you can buy, would you buy a
>300 mhz machine at this point? I sure wouldn't.
>
>So for you, a 300 mhz machine would be an extravagance. If you got one at all
>it would probably be through a sponsor, so we're back where we started.
A MHz limit does not solve every problem, but helps getting closer to a uniform
platform event.
We have a huge difference between:
1) Finding a sponsor to provide a 700MHz multiprocessor computer (open event)
2) Finding a sponsor to provide a 300MHz PC (WMCPC)
Number 2 seems to be reasonnably easy. And if you already have a 266MHz PC, you
can consider coming with this one. Maybe it is even possible to find a friend
who has a 300MHz PC, to buy one, or to find another solution.
Maybe it is possible to find an arrangement if someone comes with a 333MHz or
350MHz PC, by giving him a time handicap...
A 300MHz limit for Intel family seems reasonnably possible so nobody gets
excluded by lack of means, and not ridiculous in term of state-of-the-art
technology.
>>In Paris, we had AMD providing K6-200 computers. Can't we get in touch with
>>them, or with Intel, to see if they could provide computers?
>>
>>If we manage to get computers, couldn't we do a close-to-uniform platform event?
>>I don't care if my opponent has a 50% faster Alpha CPU, but what if he has a 5x
>>faster computer because I had to bring my old K5-100 on my back?
>
>If you show up with a K5/100, prepare to get smoked, no matter what others might
>bring, because what they bring will be faster.
I hope I won't come with K5/100. But if there is no hardware limit, I won't be
able to buy something good enough to compete anyway. I'm sure I'm not the only
one to have this problem.
>I think that people think that Alphas are really exotic because of the high mhz
>number and because they are seen as special-purpose because they are not made by
>Intel. For purposes of comparison, I believe that my program would run the same
>speed on a 533 mhz Alpha as it would on a 400 mhz P2. A 600 might compare with
>a P2/450.
That's a valuable information.
So the limit could be 300MHz for x86 and 400MHz for Alphas.
Time handicap could be given for >300MHz x86 and >400MHz Alphas.
The handicap should be strong enough that you have no interest in coming with a
too fast computer.
What do you think?
Christophe
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.