A notebook holds a GPU for eight hours and uses it for twenty minutes, which is what sharing is for and what makes it hard. The sharing mode to pick and the one to refuse across users, the idle-reclaim loop that keeps sessions alive without holding hardware, the quota arithmetic, and how many GPUs 300 users need.
Updated Sep 2026 · Grounded in real AI infrastructure interview loops and written to a senior-engineer editorial bar, with every number worked and every diagram hand-built.
A notebook holds a GPU for eight hours and uses it for twenty minutes, which is what sharing is for and what makes it hard. The sharing mode to pick and the one to refuse across users, the idle-reclaim loop that keeps sessions alive without holding hardware, the quota arithmetic, and how many GPUs 300 users need.
Ranked by how closely each one overlaps this question's topic, so the first card is the thing to read if the answer above moved too fast.
Scored on choosing MIG for cross-user sharing and rejecting MPS across users, on idle reclaim with state preserved as the central mechanism, and on deriving the GPU count from concurrency and duty cycle rather than from headcount.
No comments yet — be the first to share your approach.
