This workflow was tested successfully on linux.
- Compile code for your machine, e.g.
g++ -O3 -ffast-math -march=native main.cpp dense/dense.c sparse/bcsr.c papi/my_papi.c sparse/tcsc.c -lpapi
- Run the benchmark, e.g.
./benchmark.sh run
- Convert output stored in
out.txt
to csv format, e.g../parse-out2csv.sh > out.csv
- Plot the performance, e.g.
python3 performance.py