On Fri, Jan 28, 2005 at 10:06:25AM -0800, Fred Townsend wrote:
> I suspect the answer to your question is like the answer to the joke as 
> to why they use lawyers in place of rats for some laboratory tests.
> 
> The answer is 'There are certain things even rats won't do.' 
> 
> I suspect there are certain things that even software can't do. Does a 
> sprinter carry a stop watch so they can measure their own time in the 
> 100 yard dash? To be accurate you need a reference point and that can be 
> very difficult for the UUT to do.
> 
> Simply stated I question the accuracy of the software, particularly 
> where different architectures are used.

Well, usually the reference is the good old XT-system timer clock of the
mainboard. Should be accurate for at least 200ppm(!?). I think the reference
is not the question. The software itself tries to eliminate influce of
CPU speed and other things e.g. by measuring the loops with and without RAM.
(and I think it is running itself from free video memory to be able to test
the whole ram).
So I would think all values in the range of +-10% can be treated as the 
"same speed" because of this uncertainities. Actually my experiences with VIA
chipsets shows that these chipsets are usually slower - because of some bugs
and quirks of the hardware (e.g. the PCI-burst modes are terrible with VIA,
a bt848 frame grabber works nicely only in boards with intel chipsets).

In a windows system most speed brake is the harddisc (e.g. at startup),
so users do not really feel the speed of the RAM accesses ;-).

M.
-- 
Author: Matthias Weingart
  INET: [EMAIL PROTECTED]

Fat City Hosting, San Diego, California -- http://www.fatcity.com
---------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB CHIPDIR-L
(or the name of mailing list you want to be removed from).  You may
also send the HELP command for other information (like subscribing).

Reply via email to