Skip to main content
T3 Chips and hardware NVIDIA

NVIDIA AI Infra

NVIDIA AI Infra interview questions covering operator optimization

Operator optimization

Project experience

  1. Introduce project experience in GPU or parallel computing.
  2. When optimizing a CUDA Kernel, what angles do you usually start from?
  3. Describe locating and fixing a performance bottleneck in a resource-constrained embedded setting.

Coding

  1. Given a time string (e.g. “3:45”), compute the minimum angle between hour and minute hands.
    • Follow-up: if millisecond precision is required, how should the algorithm change?
  2. Design a multithreaded program so three threads print increasing numbers in order in a loop (A prints 1, B prints 2, C prints 3, A prints 4, and so on).
    • Follow-up: if one thread exits abnormally, how do you keep the system from deadlocking?