Author: Robert Hyatt
Date: 15:28:39 01/14/06
Go up one level in this thread
On January 14, 2006 at 04:14:31, Gian-Carlo Pascutto wrote: >On January 14, 2006 at 00:23:51, Robert Hyatt wrote: > >>On January 13, 2006 at 17:07:23, Uri Blass wrote: >> >>>On January 13, 2006 at 15:11:43, Robert Hyatt wrote: >>> >>>>On January 13, 2006 at 13:34:52, Gian-Carlo Pascutto wrote: >>>> >>>>>On January 13, 2006 at 11:46:19, Robert Hyatt wrote: >>>>> >>>>>>On January 13, 2006 at 03:52:34, Gian-Carlo Pascutto wrote: >>>>>> >>>>>>>On January 13, 2006 at 01:07:52, Aloisio Ponti Lopes wrote: >>>>>>> >>>>>>>>Did someone test this? >>>>>>>> >>>>>>>>A. Ponti >>>>>>> >>>>>>>I saw CEGT is now testing dual programs and had Shredder 9.12x64 SMP on top of >>>>>>>the ratingslist (but with few games). >>>>>>> >>>>>>>I hope they can more as it will be interesting to see the actual ELO gain from >>>>>>>SMP. >>>>>>> >>>>>>>-- >>>>>>>GCP >>>>>> >>>>>> >>>>>>I don't see why it would be any different at all, than just comparing to a CPU >>>>>>that is X times faster, since that is all SMP does. I'd expect my dual 2.8 to >>>>>>be about as fast as a single 5ghz processor, assuming everything else (memory, >>>>>>cache, processor internals, etc) are identical... >>>>> >>>>>Bugs. >>>>> >>>>>-- >>>>>GCP >>>> >>>> >>>>I ignore that angle. As bugs could come in any flavor, including a non-SMP >>>>program... >>> >>>Of course bugs can also be in non smp programs but there are bugs that are only >>>in smp version so the only way to know how much smp version is stronger is to >>>test the smp version and not to test faster processor that you believe to be the >>>same. >>> >>>I can add that different programs get different speed improvements from 2 >>>processors. >>> >>>Uri >> >>That's certainly true enough. But if you allow the existance of bugs, then no >>comparison is possible. > >Are you saying Crafty is bug-free? Ahaha. Nope. read carefully. I said if you concern comparisons with bugs, then no comparisons are possible. Bugs exist. To an extent. My parallel search is bug-free. Verified with nearly a million games played now. But proving a program has no bugs is impossible, and therefore a moot point... > >>I've played enough games with my program to know that >>it doesn't do strange things. If someone shows up with an untested program, >who cares how it does. >>I've seen non-SMP programs do bizarre things due to a lack >>of testing as well. But allowing that into the discussion negates everything. > >Realistically, the SMP mode of most programs is a lot less tested than the >normal mode, because almost noone used to have a dual machine before the X2 >arrived. I don't buy that at all. I've been teaching parallel programming for many years. Students had dual-cpu machines 5 years ago. They were not particularly pricey, other than factoring the extra CPU. I've been running parallel on ICC since 1996 or so when the first dual-cpu machines started showing up, and my old quad pentium-pro 200 box was delivered. I think the commercial guys have quad boxes available for testing all the time... > >So the chance of some serious issue being left in there are always bigger. > >Deep Fritz lost a game in the 2004 WCCC because of an SMP bug. You think Fritz >wasn't tested in advance? > >How can you "leave bugs out of the discussion" when 80% of our programs strength >is probably solely determined by bugs that we do or do not have? If you want to say "you can't compare things when there are bugs" then you might as well say "you can't compare things." In my case, I don't feel uncomfortable at all saying that if a dual processor runs 1.7 times faster than a single processor, then that program is approximately equal in strength to running crafty on a box 1.7x faster with only one CPU. Can't speak for others. But the statement (whatever the speedup being claimed is) makes sense, with or without the potential for bugs. There is nothing that says that an SMP algorithm won't cover up a bug that a serial algorithm has, so that the SMP algorithm actually plays better than the serial algorithm. Worrying about such won't help much... > >-- >GCP
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.