Author: Robert Hyatt
Date: 09:18:04 05/19/04
Go up one level in this thread
On May 19, 2004 at 10:29:28, Vincent Diepeveen wrote: >On May 18, 2004 at 14:07:37, Robert Hyatt wrote: > >>On May 18, 2004 at 13:52:29, Vincent Diepeveen wrote: >> >>>On May 18, 2004 at 13:25:18, Robert Hyatt wrote: >>> >>>>On May 18, 2004 at 12:34:31, Vincent Diepeveen wrote: >>>> >>>>>On May 18, 2004 at 11:44:27, Robert Hyatt wrote: >>>>> >>>>>Yes, you can't afford to leave USA 1 day, but you can afford $15k+ machines >>>>>always. >>>> >>>>I don't own a single 15K machine, period. I own one sony laptop, one gateway PC >>>>in my home. >>>> >>>>And you talk about "selective math". In your case it is "non-math" as every >>>>number you puke up is utter nonsense. >>> >>>So you deny that you wrote speedup = 8.81 in your thesis >>>and that you wrote in your DTS article speedup = 11.1 >> >>Please quote where I denied that. I didn't deny _either_ result... > >8.81 != 11.1 > >and your 11.1 results are based upon data which can be proven as a big fraud. First, 8.81 came from BK at 5 plies. 11.1 came from a set of game positions at 10 plies. 8.81 carried nothing from position to position. 11.1 carried _everything_ from position to position. The experiments were _different_. I posted recent opteron BK positions and even _they_ produced a worse speedup than the DTS positions. I could test to 1/2 the depth to see if it gets even worse, but I'm pretty certain it would. You don't understand the basic idea that the _same_ algorithm can produce _different_ results when important control conditions such as search depth or positions used changes. There is little I can say to remove your ignorance there... Second, there was no "big fraud". The speedup numbers were computed correctly. The times were derived from the speedup numbers when someone asked for the times. The nodes were _always_ computed from the speedup numbers because I did not display node counts in the middle of an iteration for the same reason I can't do it in Crafty. But don't let that small fact/detail deter you from you basic mission... > >As done by so many in august 2002. See CCC. > >None of your data supports therefore 11.1 speedup. Only 8.81 is the claim in >your thesis. OK. Use that. You still could not even beat that. So start taking that one apart too. IE if you can't beat it, then fabricate reasons to discredit it. Oh yes, you tried that with my 1.7 and 3.1 numbers too, but that seems to have blown up in your face after _several_ have produced similar 2cpu numbers even though you claim it is impossible. Credibility == 0. > >Also you claim that for a depth-first search (see thesis title), so trivially >not for anything you ran with at tournaments, because all software using >iterative deepening is a depth limited search. You are the most ignorant person involved in computer chess I have ever seen. You _do_ know that minimax is "depth-first search"? If not, please go to any good book on AI and look it up. "depth-first" is one alternative. "best-first" is another. You _do_ know that alpha/beta minimax is "depth-first" search, correct? Do you know what depth-first means? Hint: search first move at ply=1, then first move at ply=2, ..., first move at ply=n, then back up the score. It can have extensions, reductions, in fact it looks just like what I do in Crafty. In fact, "depth-first" _is_ exactly what I do in Crafty. And, to reduce your ignorance, it is _exactly_ what you are doing in Diep. You _really_ look like a idiot now. Keep going. At least you can get no lower... > >This all is very suspicious. What? Your incredible ignorance about precise computer-science terms such as "depth-first" vs "best-first" and so forth??? Your "motives" are suspicious. You can't do anything decent yourself it seems. So try to make something that _does_ work as advertised look worse. Much better to take on Crafty of today and get back to the 1.7/3.1 speedup numbers because there _everyone_ can prove you to be an idiot since the numbers are easy to reproduce and several have done it. Anything to say there besides "I don't care, it doesn't work for me..." > >>> >>>And you deny that when asked why you frauded the numbers in icca journal 1, 1997 >>>edition (volume 20), that they were proven in your thesis. >>> >>>You deny this? >> >> >>I have no idea what you are talking about. The DTS test and my dissertation >>tests were wildly different. One had 24 unrelated positions, searched to fixed >>depth, on very slow hardware. One had 24 consecutive positions from the same >>game, searched to variable depth, on very fast hardware... >> >>And I didn't "fraud" any numbers whatsover. That is your word. But you are the >>fraud expert of course... being the biggest fraud here by orders of magnitude. >> >> >> >> >>> >>>I personally see a substantial difference between 11.1 in a frauded article and >>>8.81 in a thesis. >> >> >>So do I. Different test. Run a different way. With pre-loaded hash tables >>carried across moves. Etc... BTW the difference is 20%. Of course that is >>"substantial" in your small world. Even knowing the two tests were not intended >>to be compatible. My dissertation did the same test everyone previous to that >>had done, the BK positions. The DTS article tried to answer a _different_ >>question, but apparently that subtle difference is lost to your "great" >>intellect... >> >> >> >> >> >> >>> >>>If everything crashed at your harddisk , how can you still post here at ICC >>>results from matches crafty - cray blitz? >> >> >>Do you remember anything I have written about this? About the fact that old CB >>executables exist in old Cray users group libraries? There used to be plenty of >>executables to be found. There may still be executables around if you look hard >>enough. But no source. >> >> >> >> >> >>> >>>"cray blitz annihilated crafty". >>> >>>I remember that posting. >> >> >>So do I although again you blew the wording... What you claim I wrote I _never_ >>wrote. But that is nothing new is it??? But as far as that match goes, it was >>interesting. It was made possible by a friend (who shall remain nameless) >>calling and saying "we are shutting our C90 down today. But it won't be moved >>until the weekend. Want to try a demo or something as I have an old copy of CB >>running on it that I use to test against my program..." >> >>So I ran for a while, until it became obvious that it was not going to be very >>interesting. I'd bet the quad opteron would have come out ahead. But not the >>quad xeon I had when I played the match... >> >>But what does this have to do with anything???
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.