Author: Russell Reagan
Date: 09:51:38 12/24/02
Go up one level in this thread
On December 24, 2002 at 08:36:01, Arshad Syed wrote: >If the program was twice as fast in assembly wouldn't that translate to an >additional 100K (approx.) nodes evaluated per sec, given that avg. NPS is ~100K? Of course. The problem is that you don't even gain one extra ply, unless your program has a branching factor of 2.0. I know of no programs that have that kind of branching factor consistently. Consider this. Look at the computer chess world championships. The person with the fastest hardware hasn't won in over a decade. That is all writing in assembly would amount to (having faster hardware). For some reason people are obsessed with speed in computer chess (and many other areas in computing), but there is little to suggest that speed is what makes a good chess program, or what makes a good 3D video game, unless (as I already mentioned) the difference is significant. A power of two in terms of speed is not "huge". All other things being equal, the faster engine should win most of the games, but the fact is that "all other things" are not equal. I've heard of programs that are twice as fast as Fritz, but some of those same programs are barely above master level, while Fritz is capable of challenging the best humans in the world. This phenomenon is even more clear when you look at a program like Hiarcs. I'm sure there are a great number of programs that are twice as fast as Hiarcs, and all but a handful (Fritz, Tiger, etc.) would stand a chance against it. Don't get me wrong. Increasing the speed of an engine is a good thing. It opens the doors for you to do other things, such as create more complex search heuristics, more complex evaluation factors, and so on. The problem is that most people are going to have a difficult time implementing new ideas in a program that is written in 100% assembly (at least compared to an engine written in C or C++, or some other higher level language than assembly). Some, like Frans Morsch, can obviously make that approach work for them. However, I think it's wrong to say, "that is how Michael Jordon does it, so that is how I will do it," because let's face it, there is only one Michael Jordan, and you're not him. This is directed towards the masses, not the exceptions. Like I said before, if you have to ask, you're not an exception, so stick to the rule.
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.