AI Infra Interviews logo
← GPU & Accelerator Architecture / 33
hardNewGoogle

Your TPU matmul width is a multiple of 256. Why does wider sharding introduce padding?

Trace a 6,144-wide projection through two shard counts, calculate local tile capacity, and distinguish a layout optimization from a change that requires training a different model.

Updated Nov 2026 · Learn how AI infrastructure works through explanations, worked examples and diagrams, then practise applying it to interview questions.

Trace a 6,144-wide projection through two shard counts, calculate local tile capacity, and distinguish a layout optimization from a change that requires training a different model.

The full answer is part of Premium. A free account includes more answers in each topic, but does not unlock this one.

20 answers per topic instead of 10, plus saved progress and bookmarks · no cardor unlock all 327 remaining answers · ₹2,000 / $25

Work through the numbers

Change the assumptions in a free calculator, then explain what the result establishes and what still needs measurement.

UP NEXT ON YOUR JOURNEY
FEDITOR'S NOTE

This is an original practice scenario, not a reported Google interview. Score the local shape calculation, the distinction between axis fill and measured utilization, and the candidate's authority to change the model.

DISCUSSION · 0

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