Your question is GPU Architecture Basics. Take a moment with it on the right.
Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).
Explain GPU architecture and how GPU memory is organized; basics of Triton.
Asked in the Prescreen stage. Asked because the candidate's research area was Efficient ML.
Cover the execution model, memory hierarchy, and practical implications for ML inference. Explain how Triton improves model serving and what you would measure when deploying a model.