Add PAPI functionality to Dr Hook

[ioanhadade]

No description provided.

[FussyDuck]

All committers have signed the CLA.

[wdeconinck]

safe_thread_num() normally returns 0 when _OPENMP is not defined.
So I thought it seems redundant to have j=safe_thread_num() wrapped with #ifdef _OPENMP.
However then the default value j=1 would become j=0.
That would bypass the further j==k condition when nthreads=1 as is the case without _OPENMP defined.
Is that the intention, why j=1?

[wdeconinck]

For sake of development, add C to the COMPONENTS in this line:
https://github.com/ecmwf-ifs/fiat/pull/18/files#diff-1e7de1ae2d059d21e1dd75d5812d5a34b0222cef273b7c3a2af62eb747f9d20aR26

[wdeconinck]

@MartynF any occurrence of long_long should be replaced with long long. See https://github.com/icl-utk-edu/papi/blob/master/src/papi.h#L554-L559

[MartynF]

Did fiat-print-bindings require OpenMP in C? It looks like it has pragmas!

[MartynF]

@MartynF any occurrence of long_long should be replaced with long long. See https://github.com/icl-utk-edu/papi/blob/master/src/papi.h#L554-L559

In version 7.... which isn't distributed with majority use OS's including current HPC right now ....

[wdeconinck]

Did fiat-print-bindings require OpenMP in C? It looks like it has pragmas!

Yes it does, but this is a standalone executable that does not even link to fiat.

In version 7.... which isn't distributed with majority use OS's including current HPC right now ....

Right let's keep it then. Thanks for clarifying.

[ioanhadade]

I would like to get this merged as it has been a while.
What is the holdup @wdeconinck?

[wdeconinck]

I would like to get this merged as it has been a while. What is the holdup @wdeconinck?

I will take this PR up further now.

[wdeconinck]

@ioanhadade @MartynF I have created a new PR #26 from the latest develop branch. Please check it still satisfies, and let's continue discussion there. I will close this PR at the moment. Thanks @MartynF for this nice feature!