I sent this email out a little prematurely. The stock software works fine with the non-RFNoC flowgraph I have. Some difference are: Stock gnuradio-config-info -v = 3.7.13.4 uhd_usrp_probe = [INFO] [UHD] linux; GNU C++ version 7.3.0; Boost_106400; UHD_3.11.0.1-0-unknown My cross-compile gnuradio-config-info -v = 3.7.12.0 [INFO] [UHD] linux; GNU C++ version 7.3.0; Boost_106400; UHD_4.0.0.rfnoc-devel-702-geec24d7b So unfortunately, there are numerous variables at play here: 1) my UHD being newer, 2) their GR being newer, 3) and the fact that I have a rfnoc-devel UHD and they are running stock. I am hoping maybe these comments will strike a chord in someone's mind and there might be something easy to look into that might be causing the issue! TIA --------- Original Message --------- Subject: UHD not getting full bandwidth on E310 From: "Jason Matusiak" <[email protected]> Date: 9/20/18 10:17 am To: "Ettus Mail List" <[email protected]>
I have a cross-compile setup for my E312 that is working fine. If I run a benchmark_rate using my compiled UHD and bitfile, it works as intended. I see about 10MHz throughput from the FPGA to the ARM before I start dropping samples, that is about right based on my readings. Now, if I use all of that, and try to do a similar flowgraph in GRC, I run the python I basically have to get below 1MHz sample rate to not lose samples. What's the difference???? GR is obviously playing a part in this, but it UHD is certainly playing a part in the change. I ultimately want to use RFNoC, but I see the issues with both RFNoC radio source and UHD USRP source.
_______________________________________________ USRP-users mailing list [email protected] http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
