The facility asks you to cap GPU power by 30 percent for the summer. What does that cost a training run, and how would you do it?

Power rises faster than clock speed does, so giving up thirty percent of the power costs closer to eleven percent of the throughput. The relationship behind that, the uniformity requirement that matters more than the level, and why an uncoordinated cap is worse than a deeper coordinated one.

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.

20 answers per topic instead of 10, plus saved progress and bookmarks · no cardor unlock all 283 remaining answers · ₹2,000 / $25

The concepts behind this question

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.

Advanced

🩺 Fleet Reliability & Observability🔒 Premium

Stragglers and HangsSynchronous training runs at the speed of its slowest rank, so one GPU that is 30% slow makes a thousand GPUs 30% slow, and one rank that never arrives at a collective makes the other 1,023 wait in silence until a watchdog fires ten minutes later. Finding the slow rank and the stuck rank is the most common on-call task on a training fleet, and the tooling for it (per-rank timing, the NCCL flight recorder, stack dumps across ranks) is specific and learnable. This page derives the straggler tax from first principles, lists the causes in the order they actually occur, and gives the procedure for a hang.

Foundational

🔌 Networking & Storage

Debugging a Slow All-ReduceA training job reports its all-reduce at a third of what the fabric should deliver, every node passed its health check, and nothing is logged. This page is the isolation order that finds the cause in an hour instead of a day: measure the collective in isolation, split the job until the slow pair or rank appears, then check the specific things that make a link, a node or a placement slow. Most cases end at one NIC, one topology mismatch, or GPUDirect silently off.

Advanced

🧮 Napkin Math & Capacity🔒 Premium

Power and Datacenter ConstraintsThe binding constraint on new GPU capacity in 2026 is not chips or capital but megawatts: an H100 node draws about 10 kW, a GB200 NVL72 rack about 120 kW, and a 100,000-GPU cluster needs on the order of 150 MW with cooling. This page converts GPU counts to power, power to cooling and facility requirements, and both to cost, so a candidate can size a training hall from a power budget and explain why liquid cooling, PUE and the local grid decide where the next cluster goes.

Foundational

🧮 Napkin Math & Capacity

KV Cache SizingThe KV cache is the memory that decides how many users a serving replica can hold and how long their context can be. Its size per token comes from four numbers in the model's config file (layers, KV heads, head dimension, bytes per element) and one formula; multiplied by context and concurrency it is the number every capacity plan is built on. This page derives it, works it for four models including an MLA one, and shows the two places candidates get it wrong by a factor of eight.

Explore the full curriculum →

UP NEXT ON YOUR JOURNEY

Next in this trackA GPU computes the wrong answer and reports no error. How would you detect that, and what does it look like in a training run?Next in this trackDesign the alerting rules for a GPU platform so that a page is always worth waking for. What are the mechanics?Next in this trackCoolant flow to a rack stops. What happens in the next sixty seconds, and what has to be automatic because a human cannot act in time?

FEDITOR'S NOTE

Scored on the superlinear power-frequency relationship making capping efficient, on uniformity across a job mattering more than the cap level, and on coordinating the cap with checkpoint boundaries.

DISCUSSION · 0

No comments yet — be the first to share your approach.