Re: fastest way to interpolate

[Federico 'Larroca' La Rocca]

Hi Martin,

Thanks for the reply. I thought the same, and tested the system by using a corrected (by hand) sampling rate on the USRP (and disabling the resampling block): I get about 40-50 MSps with virtually no dropped samples. That's why I thought about correcting the sampling rate through messages, but would like to check whether there's a faster way to resample the signal on software.

best

Federico

El jue., 28 may. 2020 a las 17:39, Martin Spears (<mspears@icmt.ca>) escribió:

Frederico

I am not certain about this but could it be the USB bus. I know with my Fairwaves XTRX which connects through 2x miniPCI-e the transfer rate is close to 8Gbps, and the SDR can do 120Msps is SISO mode

Martin Spears

Chief Technical Officer

Ingenious Countermeasures Technologies Inc.

mspears@icmt.ca

705-885-1990

---

From: Discuss-gnuradio <discuss-gnuradio-bounces+mspears=icmt.ca@gnu.org> on behalf of Federico 'Larroca' La Rocca <flarroca@gmail.com>
Sent: Thursday, May 28, 2020 4:11:01 PM
To: GNURadio Discussion List <discuss-gnuradio@gnu.org>
Subject: fastest way to interpolate

 

Hi everyone, 

I've finally been testing gr-tempest with some SDR hardware and the following problem arised: sampling correction takes way too much CPU. In my (relatively new) laptop I get dropped samples at a sampling rate above 15 MSps. I'm currently using the Fractional Resampler block, and I would need at least 30MSps to get a decent image.

The question is then: is there any other faster possibility to interpolate and correct the incoming's signal sampling rate? Right now, I'm thinking about sending messages to the USRP Source block, but it would not work with other hardware. Note that recording the signal and passing it through the provided flowgraph still works, but it would be much more interesting to have a real-time system.

best

Federico