Thank you everyone for your help! I’m still not sure exactly the ratio between the above parameters and timed command queue depth, but I empirically determined that I now have a queue depth of 64. The other key observation was to set rf_freq_policy to NONE, which increased the effective depth of my command queue from 5 to 64. I suspect that I was still issuing commands to the analog components even though I ensured they never needed to retune, and changing the policy to NONE fixed that.
If any other users stumble across this thread in the future, I have been able to achieve sub 200 us hopping consistently as long as I pipeline \~1 ms in advance. I use separate threads for transmit, receive, and issuing timed retune commands. My application is half-duplex, so I only hop tx or rx at time.
_______________________________________________ USRP-users mailing list -- usrp-users@lists.ettus.com To unsubscribe send an email to usrp-users-le...@lists.ettus.com