can we add support for build systems

[TomMelt]

@isazi we spoke at the RSECon23 workshop.

I have a simple example fortran code (Poisson-Solver) which uses:

• openMP GPU offloading
• openACC GPU offloading
• as well as their CPU-threaded counterparts

I use CMake as a build system and it would be great if we could optimize for the whole code without needing to write fake data inputs or separate my kernel out of the main code base.

Possibly good to also support make.

Poisson-Solver provides a slightly more realistic example workload than vector add and I mostly use it to test compiler support of openMP and openACC GPU directive offloads.

I am really interested in energy consumption and performance (wall time) trade-offs so I would be extremely interested to see how much this simple example is effected by the optimizations suggested by kernel_tuner.

If you have any questions regarding my code, or if you need help building/compiling please let me know. Feel free to submit issues on my repo or reply to this issue.

[isazi]

Hi @TomMelt I will have a look at your code and try to tune it, will let you know here if I need some help or have some results available