Author: Aaron Gordon
Date: 02:04:04 02/08/02
Go up one level in this thread
If I am correct Dan you cannot directly compare a computers speed with mips vs mflops as MIPS (I believe) have to do with the integer unit. From what I've seen/heard chess programs are mainly integer related so all the floating point operations in the world won't make much of a difference. A 2GHz AthlonXP would be about 5500 MIPS and 2755Mflops (8000Mflops theoretical with 3DNow!) according to the Sisoft Sandra (http://www.sisoftware.demon.co.uk/sandra) CPU benchmark. This is assuming my scores at 1.81Ghz scale linearly (which they do). My scores were 4977 Mips and 2473Mflops (straight FPU, non-3dnow). On February 06, 2002 at 19:51:28, Dann Corbit wrote: >On February 06, 2002 at 19:32:39, ALI MIRAFZALI wrote: > >>On February 06, 2002 at 19:28:21, Frank Nelson wrote: >> >>>On February 06, 2002 at 17:48:52, walter irvin wrote: >>> >>>>i wonder if a 2ghz pc with the latest crafty beat the cray blitz that won the >>>>world champoinship ?????????? >>>> >>>>my vote is with crafty >>> >>>You are forgetting that cray blitz was runned on a supercomputer at many >>>m.i.p.s. than that which a home pc can acheive. So my bets are with cray blitz! >>Actually your last Statementis not True.A 2 gig Home PC has more m.i.p.s then >>the old Supercomputers which ran Cray Blitz. > >Cray Blitz ran on a C94, which was a 1000 MIP machine. Not too shabby, even by >todays standards. >A 2GHz AMD will be a little over 2000 MFlops (which is what the MIPS above are >measuring). >So it seems that the PC should be twice as fast. > >But when you look at the memory speed for things like the hash table the Cray is >enormously faster. And the disk access for tablebase files would benefit >incredibly. However, the old Cray Blitz did not use tablebase files, so that is >neither here nor there. Other plus factors for the Cray include a vector unit >that can do lots of operations in parallel. > >In any case, it would be a fair fight, and I'm not sure which program would win. > >I suspect the increased sophistication of the PC program would win out, because >so much more development and debugging has been done on that platform. > >My extrapolation (read:guess) is that the old cray would search wider and the >new crafty would search deeper. So if the new crafty stayed out of trouble it >would probably have the better of it. But if hidden dangers were overlooked it >would be a benefit for the cray. > >It would be nice if we could find some cray time and try it. >;-)
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.