Author: Andrew Williams
Date: 11:12:03 07/20/99
Go up one level in this thread
On July 20, 1999 at 12:12:14, Dann Corbit wrote: >I think it would be interesting to benchmark chess algorithms: >0. Move generators -- all types >1. Alpha-Beta vs MTD(f) >2. Bitboards vs 0x88 >3. etc. > >Prepare a large crosstable and do a large number of runs with as many >implementations as possible and under as many different conditions as possible. > >Change the search time from very short searches (10 sec or less) up to half an >hour to find the bit O(f(n)) properties of the algorithms. > >A systematic study might eliminate a lot of guesswork or even tell us *where* >certain algorithms work better than others. For instance, we might use one >algorithm at a certain time control and a different algorithm at a longer time >control and yet another at correspondence chess time controls. This is certainly an interesting proposition. I think (having read some of the discussion below) that the best way to compare two approaches is in one program. That minimizes the number of variables. But even with this there are difficulties. eg Having started the MTDF/PVS discussion below (with a wholly innocent, and as yet still unanswered question), I thought it might be useful to hack together a PVS implementation of my program. I did this last night, and it works, after a fashion. But it's not going to match my mtdf implementation unless I spend a *lot* of time on it. At the moment my mtdf is much better than my pvs - but is that because I've got a bug? Or could it be that I've not fully understood some of the subtleties of pvs? Or is it that my carefully designed hash table, which suits my mtdf version fine, is hindering my pvs version somehow? All I can say is that two years working on mtdf gives me a better result than two hours on pvs - which isn't a very revealing conclusion :) This isn't to say that it's not a worthwhile effort, so long as we all preface our results with the phrase, "in *my* program..." Otherwise, we'll all just end up arguing again. I've just had a very interesting conversation with Vincent on ICC, where he said that calculating his attack-tables at the tips is much faster for him than doing them incrementally. Whereas I've just spend several evenings "proving" that in my program calculating them incrementally is much faster than doing them at the tips :-) regards Andrew
This page took 0.01 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.