Author: Robert Hyatt
Date: 18:12:30 06/19/01
Go up one level in this thread
On June 19, 2001 at 13:24:16, Vincent Diepeveen wrote: >On June 19, 2001 at 12:46:36, Robert Hyatt wrote: > >>On June 19, 2001 at 12:20:36, Vincent Diepeveen wrote: >> >>>On June 19, 2001 at 12:08:36, Gian-Carlo Pascutto wrote: >>> >>>>On June 19, 2001 at 12:05:33, Bas Hamstra wrote: >>>> >>>>>The Deep Blue clone FriarTuck >>>> >>>>I feel like the last few days of refuting Vincent's post here >>>>have been void. >>>> >>>>FriarTuck is crafty with singular extensions. Not Deep Blue. >>>>It runs 500Knps, not 200Mnps >>> >>>but my modifications have made crafty nearly as inefficient as deep blue >>>was searching. I'm not saying crafty is tactical as strong at the >>>same depth as DB was. What i'm saying is that the overhead it >>>requires to get near a depth n is more near the overhead of what DB >>>needed to get to a depth n, because nowadays very few who post >>>here can still imagine how bad those days were when one >>>didn't have nullmove and no shared hashtables. >>> >>>Best regards, >>>Vincent >>> >>>>-- >>>>GCP >> >> >>Vincent... give it up. I used null-move to win the 1983 WCCC tournament. It >>has been around for a _long_ time. Just because they chose to not use it >>doesn't mean they are stupid. They _chose_ to not use it. Just like I >>_choose_ to not use the double-null-move stuff because it isn't very efficient >>to do so... > >First of all i never said they were stupid to not use it. It would >not have beenmy choice. In 1994 if i would have had the same choice >as them i might have chosen for the same solution. > >However with this reaction you imply already how successful nullmove >with hashtables at todays hardware is. So successfull that you imply >that not using it is stupid! Nope. Every program is _different_. With different design decisions about how the search (and everything related) functions together. Cray Blitz used non-recursive null-move, R=1. It did quite well. Crafty is using R=2~3 and it is doing ok as well. A few are using R=0 (no null-move at all) and they are doing fine too... I don't see anything that says someone _must_ use null-move to be competitive today... > >Amazing how opinions change within 3 years of time! Not mine on this topic... > >>There is a difference between "choosing to not do something" and "not knowing >>enough to do something". A _big_ difference. >> >>This experiment is folly from front to back, and is pointless... > >But it is showing the reality as it is in 2001. It is not saying anything >about choices in 1994 of course! It isn't showing _anything_ about any program other than a fizzled-up crafty. Not related to deep blue or deep six or anything else.
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.