Douglas Geiger wrote:
Question about the USRP2 - I just received my order earlier this week, and got them up and running with no problems. However, I have a question about the transfer rate out of them. When I run usrp2_fft.py, or usrp2_rx_cfile.py I am able to get 25MS/s out of them (using -d 4) with few or no S's. However, now that I'm playing around with looking at 802.11b packets, and trying to use the USRP2 with the (modified for hier_block2) BBN code, I cannot run it at that rate without filling the screen with S's. I can imagine the BBN codeis more cpu intensive than writing to the file, but does it make sense that usrp2_fft.py has no problem? Any suggestions on where to start looking for possible bottlenecks? I am able to get some results with -d 8 (i.e. 12.5MS/s), however all the packets I'm receiving have CRC errors (the transmitter is only ~3 ft away). In case it helps, I'm running on a Core2 Duo @ 2Ghz.
25 MS/s is a very high data rate to keep up with. If you were to just run a 40 tap FIR filter you would easily be eating up half of a processor. I haven't looked at it, but the 802.11b code is bound to use more processing than the fft display does.
Does the bbn code accept other decimation rates? you could try 5 or 6. Matt _______________________________________________ Discuss-gnuradio mailing list [email protected] http://lists.gnu.org/mailman/listinfo/discuss-gnuradio
