Making it work: logging and profiling

Profilers

Shell

• time
• hyperfine
• perf-stat
• perf-record

Python

• timeit
• yappi
• CProfile
• line_profiler
• py-spy

Other

• (most platforms - Memory profiling) valgrind
• (gcc toolchain) gprof
• (C, C++, Fortran - Threaded and Parallel Code) DDT

There is a longer list of profiling tools available on wikipedia list of performance analysis tools and on the hpc2n debugger and profiler tools list.

Code-examples

Logging

• log_example.py
• log_example_mp.py
• log_example_mp_v2.py
• log_example_mp_v3.py

Profilers

• cprofile_example.py
• timeit_example.py
• yappi_example.py
• py_spy_example.py
• custom_profile_example.py
• kernprof_example.py

Homework assignments

Add logging

This homework assignment concerns implementing logging in a project. My recommendation is to do this for the chat server that you are currently working on in the larger practical. Since logging will be project wide, everyone will need to start using it in their implementations.

Profile a program

For this task you should profile a program that actually takes time to execute and where optimization might be possible. My recommendation is to profile the differential equation integrator from the first practical task and the calculator from one of the homework assignments (it could also be one of your current work projects). Use the profiling results to highlight where the program is bottlenecked.

How we can use these results to optimize the code will be covered in later lectures.