Author: Robert Hyatt
Date: 08:30:02 07/20/04
Go up one level in this thread
On July 20, 2004 at 04:29:32, Peter Berger wrote: >On July 19, 2004 at 23:13:39, Robert Hyatt wrote: > >>On July 19, 2004 at 08:57:39, Tom Brake wrote: >> >>>r1b2R1R/pp1k4/8/P3p3/2n5/6P1/4KP2/8 w - - 0 31 >>> >>>31. g4 wins. >> >>Crafty finds the right move quickly, a winning score takes a little longer... >> >> 12 2.16 -1.80 1. Rf6 Kc7 2. Rh7+ Bd7 3. a6 bxa6 4. >> Rxa6 Nb6 5. Ke3 Kc6 6. Ke4 Kd6 7. f4 >> Bc6+ 8. Kf5 (s=8) >> 13-> 7.64 -1.44 1. g4 Kc7 2. g5 Bg4+ 3. Kd3 Rxf8 4. >> Rxf8 Nd6 5. g6 Bf5+ 6. Rxf5 Nxf5 7. >> Ke4 Ne7 8. g7 Kd6 9. f4 (s=2) > >What does s=8 or s=2 mean ? S=2 means that the first two moves at the root should be searched one at a time, by all processors (serial search = 2 is the meaning). This is done when it appears that based on the nodes searched for each root move, the second root move has a node count that is very high compared to moves 3-n, and it appears that it might become the best move with a deeper search. In that case I want to use all processors to search that single move and get the score, before I split the search at the root and use one processor per move. If I did that, the new best move would take a long time to search using a single processor, adding to search overhead and perhaps making it impossible to get the fail high before running out of time. S=8 means 8 moves are close to each other in terms of the size of the tree they produce, and each will be searched, serially, using all processors on each.
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.