On 08/28/2020 12:23 PM, Nando Pellegrini wrote:
Hi Marcus ,
I am back with more or less same problem, trying to restore the former
situation.
I did not add any device argument simply because i really never needed
to change the default.
I do not pretend to have the problem solved at once on a release
candidate , i was simply reporting them.
In addition but not sure how could help , and now i am talking about
the low level of performance, my testing application is a simple and
short code looping on a receive function (with high priority) but
unable to pull out all the expected data. The CPU load is very low and
it seems like the lower levels of software are not able to engage the
cpu. I normally set the float32 format as output but tried to avoid
the conversion setting 16bit integer. Nothing changes.
Of course i do not know how lower starting from the UHD library end
ending with the OS process scheduling , and , strange, but the same in
two different CPUS where the fastest has the worse behavior.
I made also an attempt starting with the short latency version of
kernel , exactly the same!
That's the status.
nando
If you use the 'benchmark_rate' tool that comes with UHD, what are your
results with both this new release candidate and 3.15?
_______________________________________________
USRP-users mailing list
USRP-users@lists.ettus.com
http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
