Brice Videau

Hello, I am trying to reproduce the results showed here using CCS after merging with master. Rebasing the code is easy enough (see here: https://github.com/ytopt-team/kokkos-kernels/tree/feature/tuning_v4), but it seems that the...

I remember having to add it to THAPI/iprof because of one of the Intel runtime calling it through dlopen/dlsym (if I remember correctly). edit: or it could have been due...

Before the issue is closed, I would like to state that I am following it with attention...

Thanks for the detailed answer. The problem is not limited to the Sysman layer. Any tools in the table here: https://spec.oneapi.com/level-zero/latest/tools/PROG.html would be affected as well. How would those be...

@AstroVPK To be perfectly honest, I don't like the direction this is going. A flag at `zeInit` should be the way to access sysman functions, even if the `zeInit` call...

The base structures in the Core API do not have corresponding enum entries, and of the 5 base types in sysman, only base state has an enum entry.

This is mostly an issue with switch constructs on stype warning you that you have to check for it. That's how I stumbled upon it in the first place.

It should be noted that `ZE_RESULT_NOT_READY` is mentioned as a return code for the function in the `timeout` parameter documentation, but does not appear as a valid return code for...

Well you raise an interesting point: `cs._check_configuration_rigorous(x)` triggers for valid configurations: ```python for x in cs.sample_configuration(1000): d = dict(x) print(x) cs._check_configuration_rigorous(x) ``` outputs: ``` Configuration: p1, Value: '#pragma omp target...

This is the one I've always considered true: > - When the API was **last supported**? Aside from missing prefixes (mistakes?) I don't see incoherence with the above rule.