Author: walter irvin
Date: 17:49:13 02/06/02
Go up one level in this thread
On February 06, 2002 at 20:32:15, Dann Corbit wrote: >On February 06, 2002 at 20:30:11, walter irvin wrote: >>i was thinking of the best way to use 8 cpu's >> >>move >>1 cpu #1 >>2 >>3 cpu #2 >>4 >>5 cpu #3 >>6 >>7 cpu #4 >>8 >>9 cpu #5 >>10 >>11 cpu #6 >>12 >>13 cpu #7 >>14 >>15 cpu #8 >>16 >>ect >>ect >> >>with cpu #2 cutting off cpu #1 pv if bad , cpu #3 cutting off cpu #2 pv if bad >>ect ect >> >>because i dont believe that an 8 cpu pc would see 8 ply deeper , this way i >>believe it would ,to me this is better than splitting the problem at the root . > >for sure, an 8 CPY machine won't see 8 plies deeper than a 1 CPU machine given >the same time. If you have a branching factor of 3, then you will probably see >1.5 plies deeper, given some SMP loss. i was just thinking that if the cpu's followed the pv up the line then at least the score it gave that pv would be based on a much deeper search .i mean in the end it is going to make the best move based on depth x . when i play chess my pv is searched as far as i can until i lose concentration and can no longer visualize the resulting position anymore.
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.