Dear all, I've got an application that uses timed commands to change Tx frequency and gain and sent at to schedule the transmission of my data, however, I noticed that the time it takes for my app to generate and transfer the data to the USRP increased after the introduction of the timed configuration. My app works like this
1) Generate data at upper layer with timestamp t0+2ms (i.e., send at t0+2ms) 2) Execute timed configuration (Tx freq. and/or gain) at t0+2ms-300us (i.e., the 300 us before the actual transmission time) 3) Encode the data and transfer to USRP. Before the introduction of timed configuration the step 3) would take around 300/400 us but after I started using timed configuration it went to more than 1 ms. Based on what I have explained, my questions are: 1) Are timed-configuration blocking calls? 2) Is there a non-blocking way of doing this? 3) Is there a way of scheduling data transmission at t0+2ms and timed config at t0+2ms-300us in the same timed configuration? I'm looking for a way to avoid this blockage. I have tried to create a thread that would only apply the timed config and another one that would encode data and transmit at the specified time, however, the UHD seems not to like being played with in separate threads (I have mutexes in my calls, but is doesn't help). My app crashes with a message saying UHD was waiting for message 1234 but received acknowledgement for message 2000. I use jumbo frames (MTU 9000) with 10 Gb ethernet card and probably the crash is related to aggregation and or reordering. BTW, I would like to change the MTU size right now. I would like to find a better solution to the blocking call, if possible. Many thanks in advance, Felipe _______________________________________________ USRP-users mailing list USRP-users@lists.ettus.com http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
