[USRP-users] Extending 'tx_sample_from_file ' to send multiple files over multiple TX

Ali Dormiani sdormian at eng.ucsd.edu
Thu Feb 7 13:34:13 EST 2019


1. 1.25 Ms/s would be nice but lower is ok. We are trying to sound some
channels as a test so bandwidth is not a big deal.


2. If gnuradio can run on the arm cpu then thats great! We already have
python files (made by gr-companion) so all we would need to do is make a
custom N310 filesystem that has gnuradio on it?

But why would the ARM cpu be too weak to stream via UHD yet still be good
enough to run gnuradio enabled python files? I'm not a computer scientist
but python is a lot slower than a low level c++ api right?

3. Yea my University does not have a Vivado license. I'm still trying to
convince the engineering department to get a campus wide one.

On Thu, Feb 7, 2019 at 6:43 AM Rob Kossler <rkossler at nd.edu> wrote:

> > I just realized that even one instance of TX_sample_from_file from the
> internal Linux is a no go.
> >
> > In fact I think I was in a thread about this very subject a few months
> ago. Someone explicitly said the ARM cpu would be too weak to stream
> samples.
> >
> > Even streaming one file at the lowest possible sampling rate got 100%
> underflow.
> >
> What streaming rates are you hoping to achieve?
>
> > I'm trying to avoid gnuradio for this situation because that would
> require borrowing a bunch of laptops and finding some Ethernet to USB
> adapters.
> >
> I'm suggesting running gnuradio on the N310 CPU so you wouldn't need
> extra computers.  There would be no GUI, but you don't need one.
>
> > I guess that is the best option as my FPGA knowledge is too limited to
> go down the custom RFNoC route.
> >
> If you did decide to go the custom RFNoC route, it might not be too
> bad assuming that
> 1) you can tolerate random relative timing between the transmit
> channels so that you can use the existing 'replay' block
> 2) you can build an FPGA image (requiring Vivado license).  Note: I'm
> not talking about doing any development, just a build.
> 3) you can modify the existing replay example c++ program which
> presently only works with 1 Tx channel
>
> Rob
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ettus.com/pipermail/usrp-users_lists.ettus.com/attachments/20190207/d5c622ea/attachment.html>


More information about the USRP-users mailing list