Author: Scott Gasch
Date: 13:06:04 04/03/02
Go up one level in this thread
On April 03, 2002 at 15:10:24, Miguel A. Ballicora wrote: >On April 03, 2002 at 14:49:26, Russell Reagan wrote: > >>On April 03, 2002 at 14:26:47, Nolan Denson wrote: >> >>>Just wondering .. when writing a chess program why doesn't everyone program them >>>to work in dual processing. Is it that much more difficult ... Also for some of >>>the ones that are single engine now ... can you just simply take some common >>>code from a program such as crafty and If so .. what parts of crafty does the >>>dual processing ( lock.h etc etc)? It seems that many see writing there >>>programs for dual processing is a lot of work. >> >>I also would love to know the method of upgrading a program from single >>processor to include multiple processor support. I have asked various people >>this question in the past, and when I ask, "How do I make my program take >>advantage of multiple processors?" I get the vague answer, "multithreading". I >>know how to write multithreaded applications, but I have no idea how to use >>multithreading in this situation. Does this mean I need to create each search in >>a different thread for it to be slpit across multiple processors, or do I just >>need more than 1 thread in my program? I have no idea, and the one word answer >>that I've always recieved does nothing to get me from single processor support >>to multi processor support. >> >>So if anyone would care to explain how this is done, or direct me to a website >>or old topic on CCC, I would appreciate it. I did a search on CCC a while back >>for "multi processor" I think and I couldn't find anything relating to this >>topic. Maybe I missed it or searched for the wrong thing. In any case, I would >>also like information on how to make a program handle multiple processors. > >try this >http://www.it.ro/ccc_search/ccc.php?art_id=20326 > >(I did a search with YBW = Young brother wait, which is the name of an >algorithm) > >In the same search you will find the first post where B.Hyatt was testing >his parallel search on crafty. > >Vincent Deepeven also explained a simple one (not very efficient though) >developed by a french person. I could not find it now. > >Regards, >Miguel > > There are a TON of different ways people have come up with for searching a..b game trees in parallel. Youngest Brother Wait (YBW) is one. The one you could not think of above is called ABDADA. There are others like APHID. If you search on the net for these terms you'll come up with a bunch of papers that compare various parallel search methods. I seem to remember Marsland wrote a good one. Also Hyatt is an expert, his PhD thesis was on the topic. One fine day I will take the plunge and convert my engine to parallel... but it is a non-trivial exercise. The thing holding me back is that I want to be sure that there are no lurking bugs so that as I do the conversion I can be sure that any problems that come up are the result of the parallelization. Scott
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.