Hi, I am using UHD with custom RFNoC images from a custom C++ app. I am having some trouble understanding the implications of "packet sizes". I looked over some previous posts related to this topic (such as "difference between vlen and pkt_size in XML") but still don't quite understand. (Note: I do understand the math relation between samps_per_packet and bytes_per_packet, so no explanation needed there). A few comments:
- When constructing an RFNoC graph, the graph->Connect() function includes a "pkt_size" parameter, which defaults to zero, but produces a warning message if the default value is used - When getting an rx or tx streamer, the streamer args include the argument "spp" that the user can provide - Some of the XML files contain an "spp" argument Now, a few questions: - I believe that in the FPGA, the actual packet sizes are determined by the "tlast" wire. Perhaps in the Radio block (as a source), the "spp" argument will actually dictate this "tlast" assertion such that the xbar packet size will match "spp"?? - What is affected by "pkt_size" in the graph->connect() function? I just tried to create a typical RFNoC graph with Radios, DDCs, DUCs, & DmaFIFO and I purposely used unique, arbitrary pkt_sizes for each link in the graph and the graph creation occurred with no warnings and streaming occurred with no warnings. But, this was just a simple test case. - How about with the rx or tx streamers: does the "spp" arg need to match anything (such as the actual packet length determined from "tlast" on the stream coming from say the DDC)? Thanks for any help. Rob
_______________________________________________ USRP-users mailing list USRP-users@lists.ettus.com http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
