I'm looking for tips on how to approach this.
I have an AMD server with some beefy hardware and a slower laptop with not so beefy hardware (Intel skylake). I'd like to use the AMD server to produce binary packages for my slower laptop with all the trimmings.
Reading over the binary package guide, it looks like I have 4 options.
The first, which is my least preferred, is I could weaken the optimizations down to a common target compatible for everyone. This is definitely the easiest thing to do.
The second, is a chroot environment. The main issue here (as I see it) is I'd be targeting skylake in this env, but I'd be running on an AMD box. that MAYBE fine as I believe my AMD CPU (zen3) has all the same instructions as skylake, but it also might not be if there's some oddball intel only instructions which get inserted into the output.
The 3rd option would be to (somehow) leverage crossdev to create binary packages specifically for my laptop. The issue here is that my laptop has ultimately the same architecture as my host system and cross dev doesn't appear to allow me to split the two up. (Maybe I'm missing something here). This would be my most preferred route, but IDK how to make it work.
The 4th option would be to use qemu (somehow) and emulate my laptop. This is probably doable but it's nasty. Ideally I'd want the host system to be able to build these packages on a schedule and a virtual machine just makes that annoying to deal with. I'd rather not do this.
Any suggestions here would be great. I'd also like to potentially extend this for other computers I own which run into similar problems. It'd be nice to have these packages compiled by my server for those computers while I'm sleeping.


