Celia – Multi-threaded Performance Tool
A C++ and OpenMP tool that measures and visualizes multi-threaded CPU performance using parallel matrix math.
Celia measures how much faster a program gets when it uses more CPU cores. It multiplies large matrices with different thread counts, times every run, and charts the results.
Capabilities
- Parallel matrix multiplication, with the outer loop split across threads by OpenMP
- Sweeps matrix sizes and thread counts
- Reports time and GFLOPS, keeping the fastest of several runs
- Checks the parallel answer against the serial one before timing anything
- Saves results to CSV and plots them with Python and Matplotlib
Built with: C++17, OpenMP, CMake, Python, Matplotlib