I'm trying to do the same thing with a Pi 5 and a bladeRF. I've seen TX underruns on an amd64 laptops and arm64 devices. I don't know if it is 100% a Pi problem. I did some digging on the blade forums/github and there were some suggestions of rolling back the FPGA firmware to v0.12.0. On a PC I saw an improvement, with almost no TX underruns at 36MSPS. However, when I tried the same thing on the Pi 5 the amount of TX underruns got worse as compared to v0.14.0.
During my testing I noticed that there was a tight coupling between the number of transfer, number of buffers and buffer size.
Steve - I'm curious what firmware you are running and what settings you are using for the transfers.
During my testing I noticed that there was a tight coupling between the number of transfer, number of buffers and buffer size.
Steve - I'm curious what firmware you are running and what settings you are using for the transfers.
Statistics: Posted by trust_me — Sun Nov 24, 2024 2:54 pm