discuss-gnuradio
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Discuss-gnuradio] Differences between Linux and Mac gnu radio?


From: Glen I Langston
Subject: [Discuss-gnuradio] Differences between Linux and Mac gnu radio?
Date: Tue, 23 Jan 2018 12:51:07 -0500

Hello

This email concerns questions about differences in performance of gnuradio
with ubuntu linux and MacBook air implementations.

I’ve run gnu radio on a windows dual core PC booted with ubuntu 14.??.?? on a thumb drive
and also installed on an Odroid XU4 eight core ubuntu 16.??.??. 
The SDR is an AIRSPY R2 (10 MHz) or AIRSPY mini (6MHz).

My implantation of radio astronomy software runs on these computers above and also on
a macbook air 2011 vintage and MacBook pro 2013 vintage.  The code appears to
run reasonably well, not dropping many (any) data.

However there is a problem with the Ubuntu captured data.   The current version
of this code uses the WX GUI and has modifications to capture all the data
in the messages passed.   Note that the WX FFT GUI for spectral line plotting
only plots the first of N messages captured.   This code as a “notes” window
with parameters describing the observations and a “plot” window showing the
spectra and also averaging all incoming spectra.

The two differences between ubuntu and mac executions are:
1) The mac versions seem to run at high speed, gathering data, when the non-plotting
window is displayed, while the UBUNTU version halts averaging of data when the
notes window is up.  (The mac behavior is desired).
2) The UBUNTU spectra are mixed up, with some of the data having the correct peak
at 1420.4 MHz and some at the flipped end of the spectra.  In the attached figure,
labeled Odroid X4, the center frequency is 1419 MHz, the wrong data show up as a
peak at 1417.6 MHz.  The mix up does happen in both ubuntu computers and
the spectrum aliases do not happen in either Mac implementation.
The mac figure shows only a single astronomical signal at 1420.4 MHz.

The questions are:
1) does anyone else observe these behaviors with AIRSPY and Ubuntu.
My hope was to use the cheaper, 8 core Odroid UX4, device for world wide
application.   Unfortunately I can not figure out what to do next.
I’m guessing there is some sort of data loss or FFT configuration problem with
UBUNTU, when running at high speed.
2) Is there a way to force all threads to run at full speed in ubuntu gnu radio,
even when the plot is not displayed.

The ODOID plot shows the CPU drop to zero at the interval when I was looking at
the observing parameters.

Sorry for the long involved email.   Any suggestions on modifications to the
ubuntu implementation?
 
Best regards

Glen
Ignore the extra bumps in the Mac spectra at the left side of the plot on the
green line, they are due to some other tests I was running.   The correct plots
shows a strong peak only at 1420.4 MHz


reply via email to

[Prev in Thread] Current Thread [Next in Thread]