AI Infra Interviews logo
CUDA, Triton & Kernel Engineering / 12
hardNewNVIDIAFireworks

Here is an Nsight Compute report for a slow kernel. Read it, name the bottleneck, and tell me what you would change.

Four sections, read in a fixed order, and one number that usually names the bottleneck outright. The report of a kernel at 12 percent of DRAM bandwidth while its memory pipeline reads 82 percent busy, what that gap means, the fix it implies, and the numbers the fixed kernel reports back.

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.

Four sections, read in a fixed order, and one number that usually names the bottleneck outright. The report of a kernel at 12 percent of DRAM bandwidth while its memory pipeline reads 82 percent busy, what that gap means, the fix it implies, and the numbers the fixed kernel reports back.

more free answers with an account · no card

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
Kernels & Compilers🔒 Premium
Profiling with NsightNsight Systems answers where wall-clock time goes across CPU, kernels and copies; Nsight Compute answers why one kernel is slow, from hardware counters. The skill interviewers test is the order: timeline first, then the Speed of Light section, then the two or three metrics that name the bottleneck, so that a memory-bound kernel is recognized from its profile in under a minute and the fix is bytes, not occupancy.
Advanced
Kernels & Compilers🔒 Premium
Occupancy and Register PressureOccupancy is the fraction of an SM's 64 warp slots that are resident, and it is capped by the 65,536 registers and 228 KB of shared memory each block consumes. It decides how much memory latency the hardware can hide for free, but the fastest kernels on a GPU routinely run at 25 percent, so the interview skill is knowing when to raise it and when to stop.
Core
🕸️ Distributed TrainingSign in
MFU and HFUModel FLOPs utilization is the fraction of a GPU's peak that goes into the model's own forward and backward math, computed from 6ND and the step time; hardware FLOPs utilization also counts recomputation. Production LLM training lands at 35 to 45% MFU, and knowing where the other 55% goes is the job.
Foundational
🧩 GPU & Accelerator Architecture
Roofline ModelThe roofline plots a kernel's attainable throughput against its arithmetic intensity, FLOPs per byte moved from memory. Below the ridge point (peak FLOPS divided by memory bandwidth, about 295 on an H100 in bf16) a kernel is memory-bound and no amount of clever code reaches the peak; above it, compute is the limit. One picture explains why decode runs at under 1% of peak and why fusion and batching are the two levers that move it.
UP NEXT ON YOUR JOURNEY
FEDITOR'S NOTE

Scored on the reading order (Speed of Light, then Memory Workload, then Occupancy and Warp State), on spotting the memory-pipeline versus DRAM-throughput gap, and on predicting the post-fix numbers rather than just naming a fix.

DISCUSSION · 0

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