Re: [Discuss-gnuradio] Getting "0"s (overflow) in the output when probing a null sink

[Marcus Müller]

Hi Cristian,

agreeing with the other Marcus :) Also, 100 Mb/s might sound like much – but it's in fact relatively little. Each sample coming from your USRP has 16 bit real and 16 bit imaginary part, so that's a 32 bit per sample, or 320 Mb/s for a simple 10 MS/s stream.

Try with a USB2.0 port (not blue) and a rate of 8 MS/s. That should work without any "O". 

Since you don't seem to run into CPU shortage, I recommend trying a different USB3 controller. Some simply don't work well with the high, continued rate transfers that SDR devices need.

Best regards,

Marcus

On 01.05.2017 01:39, Marcus D. Leech wrote:

On 04/30/2017 07:32 PM, Cristian Rodríguez wrote:

Hi all,

I'm doing a simple flowgraph so i can check if i am able to get enough samples to work with gr-ieee_802.11.

It is the flowgraph i set.

Sample rate=50MHz → Several 0s, i think, a 0 each 5 seconds.

Sample rate=20MHz → a few of 0s, i think, a 0 each 30 seconds.

Sample rate=10MHz → a few of 0s, i think, a 0 each 60 seconds.

For 10MHz was strange the fact, that Gnuradio seems ask for 40MHz in the clock rate.

The cpu usage was around 5%, and 60Mb-200Mb ram, the name of the process Python2.

I checked the USB 3.0 speed by copying a file. The speed was around 100Mb/s

I have the USRP B210 over USB 3.0. My pc is an I7 4710HQ (up to 3.5 Ghz 8 threads), 12 Gb RAM.

It is supposed i can get until 62 Msamples/s, then i don't know what it is going wrong.

What do you think i should set so i can get 20Mhz of sample rate without "0"s in the console?

Thanks in advance.

Best Regards,

Cristian

Try setting

num_recv_frames=128

In the device arguments, and see if this helps.

Also, the B210 has a flexible master clock--the UHD drivers will pick a master clock rate that provides good filtering performance for your chosen
  sample rate, at 10Msps, a 40MHz master clock is an appropriate choice made by the driver.

Also, keep in mind that even when you aren't "doing anything" with the samples (as in your Gnu Radio example), a *tremendous* amount of
  "stuff" has to happen to those samples on the way up from the hardware in to your application.

_______________________________________________
Discuss-gnuradio mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio