Author: A. Steen
Date: 09:17:57 11/24/05
Go up one level in this thread
On November 24, 2005 at 09:40:32, Uri Blass wrote: >On November 24, 2005 at 08:54:54, A. Steen wrote: > ><snipped> >>[D]r2qk2r/ppp1ppbp/n5p1/2PQ4/1P2pP2/P1N2P2/7P/R1B1KB1R w KQkq - 0 15 >> >>I hope you can find my move here. However much time Fruit seems to be given >>in post-mortem analysis, it does not find it for white (it oscillates from 15. >>Q-d2 to 15. Qxd8+). > > >Here is some analysis of fruit on relatively slow hardware(A1000 with 64 mbytes >hash) > >It likes Bb5+ but later changes it's mind to Qd2 and I suspect that both moves >are winning Thanks, Uri. Of course you know such analyses very much depend on what is in the program's hash at the time it starts the evaluation. In post-mortem, where I just stepped from move to move with the computer's permanent brain on, pausing at ones I thought interesting, Fruit at 15. would not at all see the bishop check. In M.V. mode (I have repeated it) it would not climb higher than #3. But when I pasted the FEN directly into a freshly started F9 GUI, Fruit 2.2.1 found B-b5+, but finally preferred QxQ+. Hash contents i.e. recent machine history influence much testing, and many wrong conclusions are drawn because of it. Best, A.S.
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.