Author: Bas Hamstra
Date: 15:57:12 06/19/01
Go up one level in this thread
On June 19, 2001 at 13:30:20, Robert Hyatt wrote: >On June 18, 2001 at 16:00:56, Bas Hamstra wrote: > >>On June 18, 2001 at 13:09:24, Robert Hyatt wrote: >> >>>On June 18, 2001 at 11:45:25, Bas Hamstra wrote: >>> >>>>On June 18, 2001 at 11:00:14, Ulrich Tuerke wrote: >>>> >>>>>On June 18, 2001 at 10:51:12, Bas Hamstra wrote: >>>>> >>>>>>On June 18, 2001 at 08:33:21, Ulrich Tuerke wrote: >>>>>> >>>>>>>On June 18, 2001 at 08:28:08, Bas Hamstra wrote: >>>>>>> >>>>>>>>On June 17, 2001 at 01:09:50, Robert Hyatt wrote: >>>>>>>> >>>>>>>>>On June 16, 2001 at 22:59:06, Vincent Diepeveen wrote: >>>>>>>>> >>>>>>>>>>Hello, >>>>>>>>>> >>>>>>>>>>From Gian-Carlo i received tonight a cool version of crafty 18.10, >>>>>>>>>>namely a modified version of crafty. The modification was that it >>>>>>>>>>is using a small sense of Singular extensions, using a 'moreland' >>>>>>>>>>implementation. >>>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>>>Instead of modifying Crafty to simulate Deep Blue, why didn't you >>>>>>>>>modify Netscape? Or anything else? I don't see _any_ point in >>>>>>>>>taking a very fishy version of crafty and trying to conclude _anything_ >>>>>>>>>about deep blue from it... >>>>>>>>> >>>>>>>>>Unless you are into counting chickens to forecast weather, or something >>>>>>>>>else... >>>>>>>> >>>>>>>>I don't agree here. It is fun. Maybe not extremely accurate, but it says >>>>>>>>*something* about the efficiency of their search, which I believe is horrible. I >>>>>>>>think using SE and not nullmove is *inefficient* as compared to nullmove. We >>>>>>>>don't need 100.0000% accurate data when it's obviously an order of magnitude >>>>>>>>more inefficient. >>>>>>> >>>>>>>May be you are right, if the program is running on a PC. However if you can >>>>>>>reach a huge depth anyway because of hardware, may be you can afford to use >>>>>>>this, because it doesn't matter too much wasting one ply depth ? >>>>>> >>>>>>I don't see why inefficiency becomes less of a problem at higher depths. >>>>>>Nullmove pruning reduces your effective branching factor to 2,5 where brute >>>>>>force gets 4,5. So you could suspect at higher depths the difference in search >>>>>>depths grows, starting with 2 ply, up till how much, 5 ply? >>>>>> >>>>>>Of course nullsearch has holes, but they are certainly not big enough to offset >>>>>>a couple of plies, or none would use nullmove! In practice a n ply nullmove >>>>>>search sees more than a n-2 ply BF search. >>>>>> >>>>>>Keeping that in mind, give Crafty 1000x faster hardware. It would search at >>>>>>least 20 ply (normally 13 average according to Bob plus at least 7). I can tell >>>>>>you DB does not search 18 ply BF. Therefore Crafty would in principle see more, >>>>>>given the same eval. The SE thing only makes it worse. >>>>>> >>>>>>>I rather doubt that you can really learn something about Deep Blue this way. >>>>>> >>>>>>I don't see why not. He simply shows how inefficient their search is. Where does >>>>>>Vincent's "emulated" search fundamentally differ from DB's, in your opinion? >>>>> >>>>>Except for the authors, nobody knows. That's the problem. >>>>>We can't even be sure if they had some kinds of pruning. >>>> >>>>As far as I know they only pruning they did was futility in the qsearch. At >>>>least they seemed to have told Bob Hyatt FP was a win, therefore the probably >>>>used it. >>> >>>Obviously so. But what else did they use that we don't know about? IE how did >>>they get their effective branching factor under 4.0? With so many unanswered >>>questions, posing such a basically flawed experiment is really a waste of time. >> >>>>>If I got it right, their "engine" was a combination of software and hardware >>>>>implemented stuff. So, you cannot just scale the crafty results by some factor >>>>>and compare then with DB results. DB executed on a platform which is very >>>>>different from todays PCs. >>>> >>>>But we can compare search model A and B and talk about it. >>> >>>But you don't know much about "B". Which means you have no idea how >>>close "A" and "B" are. So you can "talk about it" yes. But you can't >>>learn anything usefule from it. >> >>I don't know that exactly Vincent tries to prove, suppose the modified Crafty >>does worse on all suites as compared to the normal Crafty and it loses all games >>against it. In my case that raises the suspicion that, though DB was good, it >>probably could have been even better. Hsu is not God, he doesn't know everything >>and nullmove wasn't as popular back then as it is now. And Hsu had no >>competition with comparable nps, else he would have learned it pretty fast. >> >> >>Best regards, >>Bas. > > >First you do know that null-move was used in the early 1980's? I used it in >the 1983 world championship tournament. It was suggested to me by someone that >had already experimented with it after reading something by Don Beal. > >It was around. Hsu knew of it. Which means that if he didn't use it, he had >other reasons than "not knowing about it". Hsu did have competition in NPS. >In 1987 he was doing not quite 2M nodes per second. We were doing 1/2M >ourselves. in 1989 he was still doing roughly 2M, we were doing about 1M on >the C90 at the WCCC in Canada that year. > >IE he didn't have a huge NPS advantage against us. In 1987 we probably should >have won, but didn't. From then on they were pretty convincing when we played >them. Ok, at the time you had a nps that was not too far behind, did you try recursive nullmove against them? If you didn't, he would have learned it if you did! If you did, I have some more questions :-) Bas.
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.