Author: Ricardo Gibert
Date: 20:27:35 02/08/02
Go up one level in this thread
On February 08, 2002 at 18:03:43, martin fierz wrote: >On February 08, 2002 at 18:00:10, martin fierz wrote: > >>On February 08, 2002 at 17:29:39, Matthias Gemuh wrote: >> >>>On February 08, 2002 at 14:39:46, martin fierz wrote: >>> >>>>On February 08, 2002 at 03:51:14, Matthias Gemuh wrote: >>>> >>>>> >>>>>Timing is a complex issue. >>>>>Believe the profiler. >>>>> >>>>>Regards, >>>>>Matthias >>>> >>>>why? what is wrong with my simple experiment? it's so simple, i don't understand >>>>what can be wrong about it. and it is real-world, whereas when it's running >>>>under the profiler everything is messed up with the profiling code. personally, >>>>i never believed the profiler too much, but after this, i don't believe it at >>>>all... >>>> >>>>aloha >>>> martin >>> >>> >>> >>>Bob Hyatt explained a lot (convincingly) on this topic. >>>Search the CCC archive for his postings. >>> >>>Matthias. >> >>i did - can't find anything under hyatt & profiler or hyatt & profilint. but do >>you really believe that a program doing >> >>for i=0 to bignumber >wait, i wanted to go a line down, not post the message :) >so: > >t = clock(); >for i=0 to BIGNUMBER > test1(i); >t1 = clock()-t; > >t=clock(); >for i=0 to BIGNUMBER > test2(i); >t2 = clock()-t; > >will not produce accurate results for how fast test1 and test2 are running? if "for i=0 to BIGNUMBER" is part of your program, then the timing will be accurate, otherwise the timing is likely to be misleading. If test2() makes its appearance, here and there in your program among a lot of other instructions, it is liable to not be operating completely out of cache, both data and code. If test2() is by itself in a tight loop, it will probably work entirely within the cache, both data and code. Actually this depends how big test2() is, but you should get the idea. > >aloha > martin
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.