Author: Christophe Theron
Date: 20:38:55 12/14/99
Go up one level in this thread
On December 14, 1999 at 17:11:39, Jari Huikari wrote:
>On December 14, 1999 at 12:52:00, Christophe Theron wrote:
>
>(I answered this, and changed to other subject below.)
>
>This subject:
>-------------
>
>>You don't need a fast computer to create a great chess program. I would even say
>>that working hard on slow hardware is profitable. With fast hardware you tend to
>>skip a lot of details, but these make the difference in the end.
>>
>>I realize I'm preaching in the opposite direction of the current trend, but it's
>>not a backward attitude. Getting the best out of what you have is the point.
>
>That's how I'm going to continue. Using my current machine (486) and my
>current programming language (Pascal). The finished program can later be
>changed to other language (and to other platform), if needed.
>
>On the other hand I can imagine that programming with faster hardware
>also has it's good sides. If program A beats program B when both are
>using slow hardware, B may still be better than A, when both are run
>in faster machine? I guess that usually program A would play better with
>any machine, but perhaps it's not that trivial. And one who has developed
>his program with fast hardware may have more experience, what works best
>with faster machines.
>
>New subject:
>------------
>
>I have tried some programs against each other with long and short
>thinking times, and got the feeling that there are less differences in
>their strenght when I give them a lot of time. Has anyone got similar
>results?
Both subjects are related in my opinion. Here is what I think. It is not the
word of god of course, but I have this in mind when I work and so far it works
rather well for me:
* I think that a good balanced program will be good at any time control
* I think that playing with long time controls gives results much closer to 50%
than playing with short time controls
These points mean that (amongst other things):
* You must also pay attention to the results your program gets at fast time
controls. I have heard many people saying that blitz games mean nothing, but if
a program does much better at long time controls than in blitz, in my opinion it
is not a balanced program, and it has a structural weakness somewhere
* At long time controls, "outside" effects will have a bigger impact on the
results and must be worked out carefully: book and learning come to my mind.
I have nothing to back up these points, except that so far it has been my
philosophy and at least it is not worse than other points of view. At least it
stands the test of real life...
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.