At least on non-Apple laptops i can’t remember having any real wifi issues wrt. Linux in a very long time. Either the onboard open-source drivers worked, or proprietary drivers were - usually automatically - installed and worked. Proprietary is never ideal, but in the case of wifi at least it doesn’t interfere with PTB’s need.
Ofc. highly proprietary things like tablets might differ, although on the MS Surface Pro that wasn’t an issue either. The Surface on Linux project i linked to provides custom build kernels which support most of the hardware at least on the older Pro models well. As of the MS Surface Pro 6, i think only the cameras are non-functional todo items, although they seemed to make some progress in the last couple of weeks.
My worst experience so far was my Apple MBP 2017, but after many hours of googling i found a solution that works. More modern Apple machines than 2017 however are quite the dumpster fires when it comes to compatibility, as Apple just can’t play along nicely with others and stick to industry standards :(.
One problem with typical tablets, for other than their default operating system, is that they are usually ARM SoC’s and standardization of components or how these are discovered and configured is non-existent or in its infancy, whereas Intel compatible PC’s had standards for that since forever.
Also extreme variety of components, and at least in the past almost only proprietary drivers for their gpu’s, so the OSS community needs to do a lot of reverse engineering for each new gpu model. The stance there seems to be slowly changing with at least some vendors, when they recognize their advantage in promoting instead of fighting open-source drivers.
Additionally, Linux open-source graphics/display system (Mesa OpenGL/Vulkan drivers + Linux DRM/KMS low level display drivers) has certain features which make it especially suitable for vision science / neuroscience applications, in some part due to my active involvement in adding and improving those features. That’s great and one of the many reasons i strongly recommend Linux for neuroscience research.
But given that most driver writers are not familiar with the needs of vision science and graphics cards manufacturers are generally not in the business of neuroscience, it still needs advocacy and active work by myself to bring specific gpu drivers up to the needed standards, and keep them there. This is way less work and way more effective than dealing with Microsofts OS or Apples trainwrecks, but it is still work.
That’s why i can only recommend stuff that i’ve actually tested or gotten trustworthy feedback on, or improved myself as needed, when it comes to embedded/mobile/tablet hardware with SoC’s.
So in general Windows tablets with Intel onboard graphics should work well, because the Linux Intel gpu drivers are excellent. You could run into other problems though like exotic sound chips, unusually broken touchscreen hw/firmware, hostile bootloaders etc. Always good to google for a while.
Wrt. RaspberryPi based hw, that’s the one SoC that i have two models of now, and at least the RPi 2B (and likely the Rpi3, as the hardware is very similar but untested by myself) work already pretty well for its niche. That was after i spent the better part of a months improving the Linux vc4 DRM/KMS display driver wrt. display timing. I’m aware of about two labs that use it happily, but i’m still surprised about the relatively low uptake in general. With its price, size, reasonable capabilities for lower end visual stimulation paradigms, and especially all its hardware i/o abilities like easy digital i/o, TTL triggering, etc., i thought more labs would be excited about this.
-mario
