[plug] Error in Time() command
billk at iinet.net.au
Mon May 24 09:07:49 WST 2010
The reason why its hard to find is its a "how long is a piece of string"
check "man time" and "man times" for some of the reasons. Basicly there
are so many outside variables - even what shell you are running under -
that it will only ever be approximate.
To get a reasonably accurate figure I think you would need to start with
the right hardware, a real time OS and be able to characterize all
software in the chain and be able to control I/O, interrupts and any
other influences - time does try and separate some of these influences,
but even that seems vague.
For what you are trying to do, you will need to measure and build your
own metric on the system you are using. This will allow you to quote a
ballpark figure if it is repeatable. I would suggest that the metric be
a comparison - software A takes 1 hour on average over 100 runs while
software B takes an average of 30 minutes using the same parameters.
The standard deviation in both cases was X - other relevant statistics
are ... That way any error or variables in measurement are included in
the figure and because its a comparison, they can be ignored.
If its a small enough piece of software, you might be better counting
execution cycles rather than measuring actual run times.
Benchmarking software is a very fraught process with a huge number of
You could also ask experts like Nvidia and ATI on how to get the figures
you want ... :)
On Sun, 2010-05-23 at 16:12 +0800, tenzero at iinet.net.au wrote:
> Hi everyone,
> I'm seeking a preferably citeable reference to the amount of error in the returned result from a Time() command. I want to be
> able to quote the level of error in timing the execution speed of my project.
> While I have been using this command partly on Ubuntu for my project, I would particularly like to find out the level of this error
> on Mac OS X. Google and Google scholar have been disapointing.
> I concede that this is a little off topic and will move it OT if people are unhappy.
> :: "Never ascribe to malice that which may adequately be explained by incompetence! -- Napoleon Bonaparte"
> PLUG discussion list: plug at plug.org.au
> Committee e-mail: committee at plug.linux.org.au
More information about the plug