As a follow up to this: I've just done a wireshark capture of a stream, using the benchmark rate @ 200 MSps and the CHDR dissector included in the repositories, and I see every data CHDR packet has a size of 1F40 (8000 bytes): is this to avoid Ethernet fragmentation from the get-go? Should I do this myself?
Thank you again! On Wed, Apr 25, 2018 at 1:58 PM Leandro Echevarría <leoechevar...@gmail.com> wrote: > Hey everybody, > > I'm planning on building the CHDR packets myself for some blocks of my own > (or at least proviing my own header when using the AXI Wrapper), and I've > got two simple questions about this: > > 1. If I specify a CHDR length of FFFF, is there a determined timeframe > within which I HAVE to complete the packet? Are there timeouts that could > prevent me from waiting until I've got enough samples to fill the 65536 > bytes? > 2. CHDR packet length is specificied as 16 bits. That would give us a > 65536 bytes of max length (and a payload of 65528 or 65520, whether we use > a timestamp or not). But UDP payload maximum size would be 65,507 when you > take into account the headers from both UDP and IPv4. Could it be useful / > more efficient to use a maximum CHDR packet size of only FFE3 or less? Or > should I not worry about this at all? > > Thanks a lot! > > Leo >
_______________________________________________ USRP-users mailing list USRP-users@lists.ettus.com http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
