[Discuss-gnuradio] Re: plasma-probe am demodulator project

[Matt Ettus]

> thanks for the file!  I appreciate the help.  I've been following the
> mailing-list; with regard to the "block"- my understanding is it prevents
> total utilization of the cpu- is this correct?

Yes.  Normall if you run a gnuradio app which uses a soundcard, USRP, or
any other ADC or DAC, the sample rate is controlled by the device.  When
you are using only files, your system will provide data as fast as it
can, and so you will get 100% utilization.

> If so would this "block" then address the following difficulty I've had
> with the example file wfm_rcv_gui.py (and others that display multiple
> plots simultaneously): namely that I cannot close the gui using the
> File-exit menu because that button becomes unresponsive.  I've tested
> reducing the number of plots displayed (by default there are 3)  by
> commenting out the sinks and found that for say 1 or maybe 2 plots, with
> less than 100% CPU utilization, the File Exit menu is usable.  I'm using a
> 1.6 GHz Pentium 4 laptop (running Mandrake 10.2 RC2).

No.  wfm_rcv_gui uses the USRP, so it is already paced properly.  The
fft graphs use a lot of CPU, so turning them off can speed things up.
However, my 1.7 GHz Pentium M is plenty fast to run 3 graphs.  What you
might be seeing is that your processor may be slowed down to save power.
 When you start wfm_rcv_gui, it takes a while for the cpu frequency
daemon to turn up the speed, and during that time the process falls
behind and builds up some huge buffers.  The way around this is to make
sure your machine is running at full speed before you start the program.

> 
> When I experienced the above problem, I used "CNTL-C" to kill the process,
> only to discover that the A/D chips remained hot, as discussed on the
> mailing list last month.  Speaking of which is there any improvement on
> the power utilization in the latest CVS?  (Both my A/D chips become hot.)

No, it has not been done yet.  The chips get hot, but it does not affect
performance or reliability.  It's a little hotter than I'd like, but
it's not a problem.

> Lastely, I was interested to read about the assembly coding for the FIR
> filters which use the SSE/3Dnow registers.  Are those improvements also in
> CVS?

Yes