poc-stencil

Investigating performances of Stencil Matrix Operations

In order to run these tests, you need

• compile the core file (stencil_vx.py) with pyccel
• run the test_stencil.py file

Compiling the core file with pyccel

run

pyccel stencil_v0.py --flags="-O3 -march=native -mtune=native -mavx -ffast-math" 
pyccel stencil_v1.py --flags="-O3 -march=native -mtune=native -mavx -ffast-math" 
pyccel stencil_v2.py --flags="-O3 -march=native -mtune=native -mavx -ffast-math" 

Compiling the core file with pyccel using OpenMP

run

pyccel stencil_v0.py --flags="-O3 -march=native -mtune=native -mavx -ffast-math"  --openmp
pyccel stencil_v1.py --flags="-O3 -march=native -mtune=native -mavx -ffast-math"  --openmp
pyccel stencil_v2.py --flags="-O3 -march=native -mtune=native -mavx -ffast-math"  --openmp

Adding new versions

Whenever you add a new version (vx), put it in a new file stencil_vx.py then following the same steps as before. The test_stencil.py file will be unique and should provide an output as a table, for the different versions, flags etc