AI 工作站由 Windows 遷移至 Mac：遷移過程揭示的依賴與設定問題
Moving an AI Workstation From Windows to a Mac: What the Move Exposed
作者將約 80 個儲存庫由 Windows 遷移至新的 Apple Silicon Mac，並用掃描器根據儲存庫 manifest 盤點安裝需求。掃描器在 27 個儲存庫找到 Python、17 個找到 Node、6 個找到 Docker；作者認為，實際儲存庫清單比通用工作站清單更適合作為安裝依據。
An operating-system move is also a dependency audit. I moved about 80 repositories from Windows to a new Apple Silicon Mac. Cloning them was quick. Finding out which tools they needed took longer.
The short field note explains how I replaced guesswork with a scanner that reads repository manifests. It found Python in 27 repos, Node in 17, and Docker in 6. The point is not the counts. It is that the repository set gave me a better install list than a generic workstation checklist.
The note also records three details that changed the setup:
- The repositories did not declare every tool they depended on, so the scanner surfaced needs before installing.
- Some steps needed a human boundary or a guard: password prompts stay with the person at the terminal, and config writes check before replacing an existing file.
- GPU assumptions were machine-specific. A small voice model ran on Apple’s GPU with an MPS option, while the repository originally hard-coded CUDA.
The measurements are from one migration and one Mac, not a benchmark. If you are moving a development setup, the useful first step is to inventory the repositories you actually plan to use.
來源：DEV Community · dev.to