Hi,
I ran experiments using benchmark_rate using the latest versions of the
3.9.7 (UHD-3.9.LTS), 3.11.1 (maint), and 3.12.0 (master) branches.

benchmark_rate --args="addr=192.168.50.2" --rx_rate=100e6 --tx_rate=100e6
--channels="0,1"

Results

   - With 3.9.7, the command completes with no errors, consistently
   - With 3.11.1 and 3.12.0, the command completes with around 500
   underruns that occur right near the beginning
   - If I change the args to add ",second_addr=192.168.150.2", I get many
   more underruns and sequence errors (for 3.11.1 and 3.12.1 only; N/A for
   3.9.7)

Note that these data rates are only half the max rate of 200 MS/s per
channel.  (If I change rates to 200e6 (with dual 10Gb), the errors
dramatically increase).

I am wondering if the problem could be with my workstation HW or config or
if the problem is with the various UHD versions.  It seems that it is the
latter given that these experiments were run in succession using the exact
same workstation & USRP.  If indeed the problem is with later UHD versions,
I am wondering if there is a time frame for when such errors will be
fixed.  I would really like to move to the later versions for a number of
reasons, including support of N3xx.

Thank you.
Rob


Details

   - Ubuntu 16.04
   - Intel XL710 NIC with 1:4 fanout DAC
   - X310/UBX
_______________________________________________
USRP-users mailing list
USRP-users@lists.ettus.com
http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com

Reply via email to