Anand,
I sorry if I misrepresented my results, but they were only
approximate figures, taken as averages over a certain interval.
Also, the significance wasn't necessarily in the values themselves
but the difference between programs.
For your information, the way I determine my figures is as follows:
- Start the process that I am interested in.
- From another terminal, run these commands:
cat /proc/stat | grep cpu0; sleep 10; cat /proc/stat | grep
cpu0
- This will give you counts of how much time has been spent in user
mode, kernel mode and in idle, at the start and the end of the 10sec
period.
- Get the difference in each of the three columns, then add the
first two numbers (user and kernel time) and take as a percentage of
the third (idle time) to get a rough average of utilisation over the
period.
My usual background utilisation is usually low enough to not be
significant against what I am monitoring, so I usually don't worry
about that. Also, I know the values I get are reasonably valid
because of the way I wrote the examples. They write each second of
sampled data to file as it goes, so that gets included in the above
profiling also.
As for your third point, see Jim's message, but basically it is easy
for myself as I wrote the driver and the usermode code, so I know
pretty much what everything is doing.
Hope this helps,
Cheers
Phil
--- In anand bhavnani <> wrote:
>
> Hi All,
>
> Jim u wrote about 2 differnt programmes having different cpu
utilisation..
> how can one come to a conclusion what exact cpu utilisation is a
programme doing ...
> and on factors does the cpu utilisation for a particular
programme depend..
>
> good bye,
>
> yours,
> anand.
>
> Jim Jackson <> wrote:
>
> Hi Phil,
>
> On Sun, 12 Feb 2006, Phil wrote:
>
> ...snip.....
> > Firstly I tried a quick and dirty program 'example2.c' (see
below)
> > to sample one channel at 22.05kHz, it was able to do it however
the
> > cpu utilisation was at ~70% and there were glitches in the audio.
> >
> > So then I changed the way the prog received the data 'example3.c'
> > (see below) to do the same sampling. Much better cpu utilisation
of
> > just 3% and absolutely perfect audio (for 22.05kHz mono). No
> > noticable freq shift from an inaccurate sampling clock, no
glitches.
> > Also, I had the program writing the output over the network via
NFS.
> > I did try at 44.1kHz but the driver had issues, I guess that's
> > getting a bit high but still I will look into it.
>
> You beat me to it. I read your code last night, and was going to
ask what
> samplerate (interrupt) rate you thought could be sustained. It
appears the
> answer is - it depends :-) I'm surprized that 22.05KHz was
sustainable,
> and that's really really impressive. Opens up a whole new range of
apps.
>
> Many thanks for the code and for sharing your efforts. Now to
build the
> use the module.......
>
> cheers
> Jim
>
>
> SPONSORED LINKS
> Computer internet security Linux os Computer internet
business Computer internet access Computer internet privacy
securities Computer internet help
>
> ---------------------------------
> YAHOO! GROUPS LINKS
>
>
> Visit your group "ts-7000" on the web.
>
> To unsubscribe from this group, send an email to:
>
>
> Your use of Yahoo! Groups is subject to the Yahoo! Terms of
Service.
>
>
> ---------------------------------
>
>
>
>
>
> " MINDSET: Its Everything " Anonymous
>
> ---------------------------------
> Yahoo! Mail
> Use Photomail to share photos without annoying attachments.
>
Yahoo! Groups Links
<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/ts-7000/
<*> To unsubscribe from this group, send an email to:
<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/
|