Author: Robert Hyatt
Date: 12:53:47 09/03/02
Go up one level in this thread
On September 03, 2002 at 15:19:57, Uri Blass wrote: >On September 03, 2002 at 14:57:58, Dana Turnmire wrote: > >>"We talk about mass fraud here." >> >>Would you explain in laymen's terms what you are talking about and what the >>purpose would be for Mr. Hyatt to be lying? I'm sure many here aren't that >>knowledgable about programming. > > >Here are the numbers copied from vincent reply that concinced me that the data >is illogical >2830 is time for one processor(position 1) >1415 is time for 2 processors(position 1) > >procs: 1 , 2 , 4 , 8 , 16 >pos >1 2830 1415 832 435 311 >2 2849 1424 791 438 274 >3 3274 1637 884 467 239 >4 2308 1154 591 349 208 >5 1584 792 440 243 178 >6 4294 2147 1160 670 452 >7 1888 993 524 273 187 >8 7275 3637 1966 1039 680 >9 3940 1790 1094 635 398 >10 2431 1215 639 333 187 >11 3062 1531 827 425 247 > >You can see that in almost every position the time of 2 processors is exactly >half of the time of 1 processors. > >It does not make sense because it is known that the speed improvement from more >than one processor is not constant. Did you look at each entry? Some got 2.0 (which might be anything over 1.95) Others got 1.7... But if you get a speedup of 2.0, wouldn't you _expect_ for the 2 cpu time to be 1/2 the time of the one cpu test? I do... Should I show you a couple of 2.0 speedups from Crafty, to see what I mean? It won't produce 2.0 in every test, neither did Cray Blitz. > >Uri
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.