Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Question for the MTD(f) experts

Author: Richard Pijl

Date: 01:43:29 04/14/04

Go up one level in this thread



>
>I pulled the pm from the hash table, so the pv move is right (when it is there)
>but the search still blows chunks of green goo:
>
>pvs:
>post
>sd 99
>st 99
>go
>Depth   Eval    Time    Nodes   PV
>1       48      0       21      d2d4
>2       0       0       84      d2d4 d7d5
>3       35      40      797     d2d4 d7d5 b1c3
>4       5       80      4313    e2e4 d7d5 f1b5 c8d7 b5d3
>5       35      241     19619   e2e4 e7e5 d2d4 d7d5 g1f3
>6       13      872     88472   e2e4 e7e5 d2d4 e5d4 d1d4 g8f6
>7       30      4517    570969  e2e4 d7d5 e4d5 d8d5 d2d4 d5e4 g1e2 e7e5
>8       18      30655   3279256 e2e4 b8c6 d2d4 e7e6 b1c3 d7d5 g1f3 g8f6
>
>mtdf:
>post
>sd 99
>st 99
>go
>Depth   Eval    Time    Nodes   PV
>1       48      0       172
>2       0       30      2224    d2d4
>3       35      60      4602
>4       5       451     41509   e2e4
>5       35      1362    103001
>6       13      7741    799578  e2e4
>7       30      15553   1623933
>move e2e4

Have you tried fail soft? That greatly improves the usability of the hashtable.
Do you also use the bestmove from the hashtable for move ordering?
TSCP uses the pv array to search the pv first, and an empty pv array breaks
moveordering somewhat. (This should improve the non-MTD search as well, as it
normally doesn't have a move to search first at non-pv nodes.)

Richard.



This page took 0.02 seconds to execute

Last modified: Thu, 07 Jul 11 08:48:38 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.