@idontcare_
I have the right solution for you!
I built a native container system for Mac called MacDocker, where you can run Mach-O binaries straight on macOS. No Linux VM, no emulation and Sandbox/seatbelt isolated.
Have you tried to use your M4's GPU with a Docker container? Or build and sign your Swift code in Xcode it? Short answer: you can't.
Docker Desktop and Apple Container runs a Linux VM, and it sees no Metal, no MPS, no MLX and no XCode.
What Docker Desktop can't do, MacDocker can!
🔥 Real GPU access: PyTorch (MPS) and MLX using Apple Silicon's GPU and unified memory.
🛠️ Xcode in a container: upload your project .zip, get back a .dmg or .ipa signed with your certificate.
⚡ Containers up in ~130 ms: every job starts from an APFS clone (copy-on-write) — disk is used only for what changes.
🔒 Isolated by default: per-job macOS sandbox, dedicated service user, network blocked while running, and caps on memory (GPU included), processes and time.
We are looking for beta testers.
https://macdocker.com:8000