Peak GPU Memory Usage

Practising this easy GPU memory profiling problem in Python

This is a straightforward easy coding problem that tests your ability to extract and track memory metrics from a time-series log. It is representative of the kind of practical instrumentation work that GPU-accelerated teams at firms like Nvidia encounter when building profiling and diagnostics tools.

The problem requires you to scan a sequence of memory readings and identify both the maximum value and the first time step at which it occurs. Clean solutions are simple and efficient: they typically iterate once through the input, maintain a running max, and return both the peak and its earliest position. The main nuance is handling the empty-list edge case correctly.

• Single-pass iteration and state tracking
• Index management in Python sequences
• Handling boundary conditions