Author: Jürgen Hartmann
Date: 05:20:43 03/21/99
Go up one level in this thread
>> >>- 11 years x 0.75 gives a Pentium at 4000 Mhz (8 times faster) > >hmm... isn't it rather (1.75)^10, which is about 270 times faster? >[snip] Empirically it is rather 50% per year, that makes 1.5 ^ 10 == 57. e.g.: Autumn 96: 200MMX Autumn 97: PII/300 Autumn 98: PII/450 And 10 years ago we were at 500 / 57 = 8-10 MHz, which seems about right. So in 2010 we will have 500 * 57 = 28GHz. But the motherboard will contain 256 - 512 processors. Optimistically assume that yields a factor of 100 (??!) for chess programs. So the overall factor could be around 5000. An average chess program searches 200,000 nps. It will then be 200,000 * 5,000 = 1000,000,000 nodes per second. Say, that translates to six extra ply giving 20 ply in a tournament game for a standard program. Even with nowadays knowledge, we will see some pretty deep planning or can you imagine a standard plan longer than ten full moves? Kings attacks suddenly materializing out of the blue... I am looking forward to a Blitz game against Rebel20 (running under Linux 2010). This will be weird, weird chess. Jürgen
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.