parallel-mandelbrot

Parallel implementation of mandelbrot. Created as part of the final assignment for the class "High Performance Computing" at UniGe.

Dependencies

To be able to build all binaries you need a working intel oneAPI basekit installation version 2025.0.4, a working CUDA installation and working AMD ROCm installation. If you only want to build some binaries, you can comment out the includes in the CMakeLists.txt file in the root directory.

How to run

source /opt/intel/oneapi/setvars.sh --include-intel-llvm
mkdir build && cd build
cmake ..
make clean
make run

Other scripts

visualize.py

Script that reads the output of a binary and creates a picture using matplotlib.

visualize_opengl.py Script that reads the output of a binary and creates a picture using vispy with opengl acceleration for larger interactive rendering.

benchmarks/ Scripts to build and run all binaries for CPUs and scripts to generate plots from the gathered data that show speedup/efficiency etc.

gpu_benchmarks/ Benchmark script for gpu binaries and scripts to generate plots from that data.

advisor/ Script to execute advisor runs with roofline analysis and create snapshot for each binary and csv report for later analysis.

result_compare/ Write output files for all binaries and a script to analyze the output to find out which outputs are equal, to check correctness.

Final report

This repository contains a final report in the file final_report.pdf that explains the implementations and discusses the results. Data included in the report, such as benchmarking and Intel Advisor reports can be found on the report-data branch.