Been playing with my radio a bit, and came to some conclusions regarding controlling it with a Raspberry Pi, which was something I’d wanted to do. Frankly, the Pi‘s not up to being the main computer for that task.
I’ve got a Pi4 8Gb, with 64-bit OS, and running that off an SSD. In other words, about as speedy as I can get with one of those. I like the digital modes in HF, and use wsjt-x with Grid Tracker for an absorbing hour with the rig. But I’ve been noticing a distinct lack of decodes, despite a waterfall that’s lit up to shit on the set.
So I hooked it up to my M1 iMac, and erm, wow. What a difference. It really surprised me, because both the Pi and the Mac are arm64 architecture with 8Gb RAM. Granted the clock speed and number of cores on the Mac are double those of the Pi, but even so, the difference is shocking.
One example is the time taken to compile wsjt-x. On the Pi, there’s really no choice than to compile from source for a recent version, so I compared it with the same operation on the Mac. Literally, a tenth of the time on the M1 vs the Pi to compile it.
Wsjt-x is giving me many more decodes per period, the Mac will cope with “Deep” whereas the Pi really has to be in “Fast” mode, which has predictable results in terms of decodes. I’m wondering what I’ve been missing in the past now, using the Pi to control the rig.
So it looks like the Pi(s) are relegated to other tasks. Time to rebuild my shack around the M1.
I’ll post something on the Pi thread with more info on the Pi, but on this thread, I am a happy bunny!