Hi Steve,

And the LocalTime<TPrecision> return ticks, correct?

The goal is to estimate a practical worst-case execution time of the 
conversion, and after that try and model the firmware using a formal notation 
and verify the schedulability of it etc.

--Anders
On Friday, May 13, 2011 at 10:38 PM, steve ayer wrote: 
> hi anders,
> 
> at 8mhz, 150 uS is 1200 clock ticks. you can also try using a tmicro 
> timer if you really need details; what are you trying to accomplish by 
> measuring the conversion time?
> 
> -steve
> 
> On 05/13/2011 11:27 AM, Anders Sørensen wrote:
> > > that said, if you want to use software for timestamps, then use localtime.
> > > 
> > > also bear in mind that there's a latency penalty introduced by atomic
> > > code blocks (iirc, about 22 ticks for just the setup and teardown of
> > > atomicity, plus code).
> > Thanks for the pointers. I tried using the LocalTimer<T32khz> already 
> > instantiated with the BoilerPlate firmware. However, reading from that 
> > gives suspect values.
> > Measuring the average execution time (again, from before 
> > 'triggerConversion' to when the DMA is ready) using this LocalTimer with 
> > the precision of 32kHz is 5. That is, the difference between the two 
> > recorded timestamps.
> > 
> > Am I right to assume that the value returned from LocalTime.get() is in 
> > ticks? If so, then the value 5 translates into roughly 5/32768 s<=> 
> > 0,00015s<=> 0,15ms which isn't much. Is this realistic?
> > 
> > --Anders
> 
_______________________________________________
Shimmer-users mailing list
[email protected]
https://lists.eecs.harvard.edu/mailman/listinfo/shimmer-users

Reply via email to