Your training collectives are slow. How do you debug the NCCL/interconnect path and find where bandwidth is lost?
When all-reduce is your bottleneck, the culprit is nearly always a misconfigured path: traffic on the wrong link, a dead NIC, or a topology NCCL failed to discover. Here is the methodical way to track down the missing bandwidth.
Updated Sep 2026 · Grounded in real GenAI, LLM, and AI/ML engineering interview loops and written to a senior-engineer editorial bar.
When all-reduce is your bottleneck, the culprit is nearly always a misconfigured path: traffic on the wrong link, a dead NIC, or a topology NCCL failed to discover. Here is the methodical way to track down the missing bandwidth.
Lead with where the obvious approach breaks, because that is the judgment they are screening for — most candidates jump straight to the happy path and lose the room.
Then walk the failure back through the pipeline in order, naming the one metric the customer's exec sponsor actually cares about before you propose the fix.
