This post is stating what must be obvious to everyone but me, but wow, a "real" computer (mini pc) is a lot faster than a Raspberry Pi 4 w/SSD!
Over the past couple of years I've gone from a Raspberry Pi 3b with 1Gb memory to a RPi4b w/4Gb, to a RPi4 w/8Gb + SSD. I run all of KStars/Ekos/Indi on my RPi, just using my laptop as a remote-desktop monitor. The system has been very good to me, allowing low cost compute for astro-imaging right on the telescope, and for the most part the hardware has been reliable (especially once I ditched SD Cards). Starting with the RPi4, I've been able to compile KStars/Ekos/Indi on the machine, though it can take over an hour for many compiles.
However, I've been getting the itch for a faster telescope computer for some time, mostly because of the slow compiling. In the past year it's been hard to find electronics. Recently I happened to see a NUC10 (
this one the i7
). I picked it up along with
16Gb of DDR4 ram
and a
500Gb M.2 NVMe SSD
and installed Ubuntu 20.04LTE--and that's all that's needed to build the computer. Yes, it's ~$800 vs ~$100 for the RPi4, for sure an unfair comparison.
I put together all the software on my desk and configured Ekos, then moved it outside, and quite literally just unplugging a few wires from the RPi and plugged them into the NUC. I did have to switch to the NUC's power-brick, and am not sure yet how/if I'd power the NUC if I took it traveling (it takes 19v). The NUC's form factor is a little bigger than the RPi, but not bad, it's about 4"x4"x1" (though its power brick is at least that same size as well, but can be placed out-of-the-way). The NUC uses a bit more power, but is not too bad while imaging.
Cutting to the chase, the speedup in UI over remote desktop using NoMachine is astonishing. I used to think I had network issues which caused remote desktop sluggishness. I guess it turns out that the RPi4 was the reason my remote-desktop networking was slow and sometimes flakey. The speedup in compiling KStars is even more astonishing. It went from over an hour down to 5-10 minutes. All that is compared with the fastest RPi running with an SSD, which itself is significantly faster than SDCard-based RPis.
Honestly, I really enjoyed using the RPi4, and it really is "the little computer that could", so, for sure I'll keep around a few RPi4bs and image with them on a secondary rig, e.g. for testing, but from a personal imaging perspective, it would be very hard to go back.