This brief Intel® VTune™ Profiler video tutorial uses a Mandelbrot set that generates the Python* script mandelbrot.py as the workload to analyze. A hot spot analysis is configured with user mode sampling and the managed code profiling mode is set to auto before starting the data collection. After data collection completes, the Hot Spots Summary tab shows the elapsed time, CPU time, total thread count, and the greatest hot spots. The source view shows the lines of Python code that consume the greatest CPU runtime.

You can apply this methodology to your own Python workloads, and identify thread contention and other performance bottlenecks. Metrics for the interpreted Python application include elapsed time, CPU time, thread count, and hot spots. In-depth performance profiling for Python is available with the ease and level of insight you are used to for C and C++ workloads, and you can use the same UI to get it.