Scenario: Graphics/Compute Tradeoff

Scenario

A graphics-heavy scene shares GPU resources with compute post-processing. Fragment queues and compute queues interfere under thermal throttling conditions.

OBSERVED METRIC
frame-time spikes appear after enabling async compute workload

45-MINUTE INTERVIEW FLOW
0-5: scope workload and KPI
5-15: map SIMT/memory/scheduler mechanism
15-25: isolate first failing evidence
25-35: propose bounded fix with owner
35-45: state validation matrix + rollback

Common pitfalls

• Tune only graphics stages without queue arbitration analysis

• Ignore thermal/DVFS state transitions during spikes

• Treat driver scheduling and hardware QoS as independent problems

Scenario debrief

Score candidate response on workload framing, bottleneck proof, mitigation choice, and regression discipline.

kernel intent -> hardware execution -> measured outcome

throughput trend

Debrief prompts

1. Which resource bottleneck appears first in evidence?

2. What hardware or software lever changes the bottleneck?

3. What reproducible benchmark gate proves closure?

Key takeaways

• Always pair micro-kernel metrics with end-to-end workload impact.

• Lock toolchain, driver, and launch metadata before comparing performance results.

Common pitfalls

• Optimizing occupancy without checking memory-system saturation.

• Comparing profiler captures from different driver or compiler builds.

• Declaring wins without reproducible accuracy and performance gates.