when I receive a block with size N < samps_per_buf, the next call to rx_stream->recv 

Hi usrp folks, Since may last short question has no answer, I expose my problem in more details (perhaps it was not clear enough): I am interesting to know how many samples are lost during a streaming process. I receive samples with a "uhd::rx_streamer" object of size "samps_per_buff". My experience about overflow with usrpN210 is : when I receive a block with size N < samps_per_buf, the next call to rx_stream->recv will generate an overflow error code (and zero size returned). Does this means that the recv method should always returns samps_per_buf samples in normal conditions (excepting overflow) ? How can I estimate the right number of losted samples ? - is it (samps_per_buff - N) + samps_per_buff - should I use the meta_data::time_spec to compute the number of losted samples ? does this time_spec already significant even for overflow conditions ? - another method ? Any help is appreciated. thanks. Olivier

  when I receive a block with size N < samps_per_buf, the next call

On 12/18/2012 03:13 PM, raboteauo wrote: > Hi usrp folks, > > Since may last short question has no answer, I expose my problem in more > details Some of us have to get the coffee 1st. > (perhaps it was not clear enough): > > I am interesting to know how many samples are lost during a streaming > process. I receive > samples with a "uhd::rx_streamer" object of size "samps_per_buff". My > experience about > overflow with usrpN210 is : > > when I receive a block with size N < samps_per_buf, the next call > to rx_stream->recv will > generate an overflow error code (and zero size returned). Does this > means that the recv method should always > returns samps_per_buf samples in normal conditions (excepting overflow) ? When continuous streaming, yes. The buffers will always be maximally filled. > > How can I estimate the right number of losted samples ? > - is it (samps_per_buff - N) + samps_per_buff > - should I use the meta_data::time_spec to compute the number of losted > samples ? does this time_spec > already significant even for overflow conditions ? > - another method ? Basically. lost_samples = time_delta*sample_rate I hope benchmark_rate.cpp will be helpful -josh > > Any help is appreciated. > thanks. > > > Olivier > > _______________________________________________ > USRP-users mailing list > USRP-users@lists.ettus.com > http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com