[USRP-users] Installing PCIE drivers for x310

James Humphries james.humphries at ettus.com
Wed Oct 28 21:38:05 EDT 2015


Hi Jared,

Just want to let you know I am trying to replicate the results you are
seeing. For reference, I am using a core-i5 processor @ 3.5GHz and 8GB RAM
(I can get specific details if you are interested). I am also using the
PCIe kit. With this setup, I am able to run the benchmark_rate utility all
the way to 200MS/s without error. Also tried a simple GNU Radio flowgraph
with 100MS/s and 200MS/s that is a USRP Source --> FFT GUI Sink. Again, no
indication of dropped samples.

Here is my test process if you would like to try and emulate, maybe we can
compare results (these results are for 100MS/s):


   - Boot X300
   - Boot Host Computer
   - Start niusrprio service
   - Check interrupts for niusrpriok ($ cat /proc/interrupts); Should show
   zero at this point since nothing has been sent.
   - Check dstat interrupts; Again, should be showing zero
   - Run benchmark_rate utility ($ ./benchmark_rate --rx_rate "100e6"
   --duration "60")
   - Check dstat interrupts while running (Each reading shows about 30k
   interrupts)
   - Check /proc/interrupts again after test is finished (mine shows core
   0: 1550998; core 1: 360899; core 2: 9214; core 3: 0)
   - From benchmark results, I see 5999859534 samples received with no
   errors.

It appears to me that this is some bottleneck on your machine that is
limiting your total throughput. I noticed that dstat showed my number of
interrupts is much higher, but if you provide me with the exact benchmark
you ran (or run with the settings here), we can compare numbers.

-Trip


On Tue, Oct 27, 2015 at 8:40 PM, Jared Dulmage via USRP-users <
usrp-users at lists.ettus.com> wrote:

> Trip,
>
> Thanks for all the advice.  I did
>
> > cpupower frequency-set -d 2.5GHz
>
> such that
> > cpupower -c 0-11 frequency-info  -f
> analyzing CPU 0:
> 2501000
> analyzing CPU 1:
> 2501000
> analyzing CPU 2:
> 2501000
> analyzing CPU 3:
> 2501000
> analyzing CPU 4:
> 2501000
> analyzing CPU 5:
> 2501000
> analyzing CPU 6:
> 2501000
> analyzing CPU 7:
> 2501000
> analyzing CPU 8:
> 2501000
> analyzing CPU 9:
> 2501000
> analyzing CPU 10:
> 2501000
> analyzing CPU 11:
> 2501000
>
> benchmark_rate still gives overruns at 100 Msps.
>
> Jared.
> ------------------------------------------------------
> Jared Dulmage
> Engineering Specialist
> Digital Comm. and Implementation Dept.
> Aerospace Corporation
> 310-336-3140
>
> _______________________________________________
> USRP-users mailing list
> USRP-users at lists.ettus.com
> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ettus.com/pipermail/usrp-users_lists.ettus.com/attachments/20151028/47939b7e/attachment-0002.html>


More information about the USRP-users mailing list