Hi! Been checking out Fly’s YouTube channel alongside the docs and in this one he says if you do a Docker build on a non-Intel Mac (M1-3, I have a 2 and a 3) it will build its arm64 architecture which would subsequently not run properly on Fly.
Does Fly still only support Intel Mac machines for building images? Is there maybe something I’m missing, like with containerization it’ll still ship if it doesn’t work on my own machine? (Macs are my only machines so far, but not for long…) Not sure if that goes logically, since I wouldn’t be running locally to deploy, but want to bring up the arm64 thing since I only have that version.
My other wonder is, what happens when I add GPUs for my chatbot app? (I do have several specific docs bookmarked so I can look at the GPUs section again too)
Your source code is uploaded to the builder by flyctl as an initial step (glossing over some details here). That way, you both compile and run on the same platform: Linux x86-64.
In contrast, compiling on one architecture but then running on another (i.e., cross-compiling), while often possible, can prove challenging even for experienced users, .
It definitely does! That scenario is literally what I was imaging it might be, like eliminating the problem of me having Silicon macs by their system doing the builds remotely.
Also great as I plan to do a custom PC build for GPU dev work/ML and then I have Linux x86-64.
.