Author: Robert Hyatt
Date: 15:55:46 11/12/03
Go up one level in this thread
On November 11, 2003 at 14:06:48, Dann Corbit wrote: >On November 11, 2003 at 09:49:45, Aaron Gordon wrote: > >>On November 11, 2003 at 09:39:20, Jorge Pichard wrote: >> >>>On November 11, 2003 at 08:50:53, Aaron Gordon wrote: >>> >>>>It would be better had they used a quad or 8-way Opteron running 2GHz or more. >>>>From some testing I've done in the past you can figure a single Opteron 2GHz == >>>>a P4-3.6GHz in Fritz 8 (32bit mode). So, a Quad Opteron 2.0 == Quad P4-3.6. >>>>Almost 30% faster, plus the memory bandwidth available would probably push it a >>>>bit over that with large hash table sizes. 8-way Opteron 2.0 would of course be >>>>like 8 p4-3.6's (however with some 40gb/s+ memory bandwidth available depending >>>>on bus speed). >>>> >>>>Why not use the best hardware? Seems like if you'd want to promote your new >>>>'awesome' chess program you'd want to give it the best chance of winning. >>> >>> >>>If you were in charge of advertising the AMD company, you probably would have >>>offered the Fritz team as well as Mr. Kasparov a good deal of money to select a >>>Quad 4-way Opteron running at 2GHz or more, but it seems that Intel made a >>>better proposal knowing how much revenue it can bring their company. >>> >>>PS: Remember how good IBM stocks did after the 1997 match against Mr. Kasparov, >>>it is NOT just a match, it is a well promoted Intel advertisement Show, shown >>>all over the World by ESPN. >>> >>> >>>Jorge >> >>I'd prefer a stronger machine and more potential money rather than a slower >>machine and more money now.. but thats just me. :) When stuff like this happens >>there is always a voice in the back of my head going, "If it was run on THIS >>instead, what would have happened?". 8-way Opteron could potentially pull 70 or >>so elo (if you figure 50 per doubling) over the quad xeon. More if you consider >>it could be 64bit. This at this point is probably not realistic as I'm sure >>Franz and whomever else would need a good bit of time to produce a nice 64bit >>engine. >> >>Better to have focused on getting the engine tuned right now for this particular >>match.. which of course they did. Now, hopefully, they can turn their attention >>to an Opteron(64bit) version of Fritz.. unless of course Intel has stuffed >>entirely too much money up their you know what. :) >> >>Again this is where you wonder, "could that extra 70 elo have helped Fritz draw >>this particular game, avoiding losing to Kasparov?".. or.. whichever way you >>want to do it. Just doesn't seem right not using the best available.. not as >>interesting IMHO. > >The 8-way boxes are pretty much all beta stage (except for HP Alpha machines and >stuff like that). While 64 bit Linux has been released, the commercial chess >programs run on Win32 systems. Win64 is beta right now. Probably, it will take >some work to port 32 bit systems to 64 bits to really get the full benefit, >anyway. I imagine that the spectacular crafty numbers Eugene posted were using >the compiler in 64 bit mode and running on 64 bit windows (both in beta stage >right now) > >All beta systems carry considerable risk of failure. I think their choice of >hardware is exactly what I would have chosen, if I could pick anything I wanted >to run a 32 bit Windows program on. They will have had plenty of time to test >and prepare on a system such as that, and they would know it is rock-solid >reliable. Trying to eak out a few more ELO on a shaky system would not make >good business sense. This is a tough call. IE I can think back to the Cray Blitz days. We almost _always_ ran on experimental hardware (so that we could use the fastest thing Cray had to offer) using experimental software (the O/S and compiler guys were just as busy trying to test on the new hardware). We had good results. We had a couple of disasters. But even on reliable hardware you can see things go to hell in a handbasket when something simple like a DRAM chip goes south and things don't crash, they just don't run right. > >Two years from now, all contests of this nature will be on 64 bit hardware, >because it will be fully stable by then. Maybe, but there is always "the next generation" lurking in the background. IE at the Paris WMCCC event, three people used the Kryotech alphas that were pushing the leading edge by 25%+. And most everyone competing considers the WMCCC to be a "money-making" affair for the "winner".
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.