Author: KarinsDad
Date: 10:17:48 02/12/00
Go up one level in this thread
On February 11, 2000 at 11:44:19, Tom Kerrigan wrote: >On February 11, 2000 at 08:19:17, Robert Hyatt wrote: > >>You can obviously forget about the 680x0's. :) However, I'd expect most >>new RISC designs to do a pair of instructions at a time. Of course, I don't >>really follow the 'blender-type processors' very closely. :) I'm more >>interested in the other end of the spectrum. > >Yes, I like the high-end too, but a quick look at the top programs shows that >bitmaps are not necessary for a strong chess program. I think the issue may be one of degree. If a program got modified from non-bitmap to bitmap (which would probably be a lot of work since so many elements of a chess program are dependent on each other), it may increase in speed by 10% to 20% (maybe). And, this may correspond to a 7 to 18 ELO improvement (depending on whether a 2x speed increase is a 50 ELO or a 70 ELO improvement; I've heard both of these figures). So, given this, it is not just the bitmaps that are important. There are a lot of other important elements of a chess program that give it it's strength (or lack thereof). However, I think that if you had two programs that were basically identical with the exception that one used bitmaps to handle certain portions and the other did not, that the bitmap version would be slightly faster. But this speed increase would not be enough to notice more than a slight ELO difference between the two programs. Hence, this is a potential reason behind your observation on the top programs. The reason I think a lot of people are thinking and talking about bitmaps is not because it is a new idea. But rather, the hardware is becoming sophisticated enough that the idea now has merit. KarinsDad :)
This page took 0.01 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.