--- In Jim Jackson <> wrote:
>
> On Sat, 30 May 2009, charliem_1216 wrote:
>
> > It was just what I needed. Here:
> > http://tech.groups.yahoo.com/group/ts-7000/message/2085
> > Phil reported easily taking data at 22 kS/sec, with 3% CPU. Note the
> > difference between his example2 and example3 programs: it makes a
> > difference what you do with the data.
>
> I've always been suspicious of cpu usage figures when the load is in the
> kernel - dunno exactly why, but I get a twinge in me joints (might be
> artheritis :-) 3% CPU for 22kS/sec, and issues in the driver at 44sS/sec!
You're right, I should have been more precise. The time is accounted for in
"%sys" field in top, and that was very low as well.
Here is a link to a lkml thread I saved, where kernel gurus explain how cpu
usage is measured in the kernel:
http://lkml.org/lkml/2007/2/12/6
>
> As the sampling is done in a timer ISR, I've always been a bit uneasy about
> jitter on the samples. I'm no ARM processor expert, but from experience
> with other processors just because an interrupt is raised doesn't always
> mean the processor with immediately jump to the ISR. The processor could be
> already be processing an ISR with interrupts disabled, and the interrupt
> will be actioned when interrupts are enabled at the end of the ISR. This
> means that time interval between samples could be subject to some jitter,
> which is likely to get worse the higher the sampling rate.
>
> I was going to see if I could do something to measure the jitter, but I
> never had time and I didn't really need very accurate high sample rates, so
> never persued it. I'd be interested to know if anyone has looked at this.
I haven't looked at jitter, but you are trying to find the highest workable A/D
speed, you definitely need some more sophisticated tests. It wasn't needed for
what I had to do.
regards ....... Charlie
>
> > Also, the examples won't work until the frequency IOCTL that I broke is
> > fixed.
> >>
> >> Any idea what IO scheduler would be best to achieve fastest sampling?
> >
> > I don't think that would have an effect.
>
> ditto.
>
> Jim
>
------------------------------------
Yahoo! Groups Links
<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/ts-7000/
<*> Your email settings:
Individual Email | Traditional
<*> To change settings online go to:
http://groups.yahoo.com/group/ts-7000/join
(Yahoo! ID required)
<*> To change settings via email:
<*> To unsubscribe from this group, send an email to:
<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/
|