Tobias Meyer Andersen
Tobias Meyer Andersen
std::upper_bound() seems to be slower on my machine benchmarking a couple of runs on norne... Edit: To be a bit more precise, the property evaluation on single core norne on...
The measurements are probably within what can be explained by noise, so hopefully then benchmark can provide some consistent results
That was a bit underwhelming, but how are these times measured since there is absolutely no deviation on any case?
Can we get some clarifications as of what happened to the performance benchmark @blattms
My local measurements turned out to be a bit deceiving, and I dont really think the effect of skipping this if statement is measureable/significant with this level of noise. Should...
OPM-simulators mainly uses OMP for multithreading computation/memory transfers. For this reason it might be nice to use opm_get_max_threads()>1 to check whether or not we want to use the multithreaded parallelization...
The change with the OpenMP processor stuff is almost what I had in mind. It is still possible that the user has OpenMP available, but has explicitly run flow with...
This looks good to me.
This code is tested to work on an AMD card. All the cuda tests pass in their hipified form, two are rewritten as they contain types/function that do not exist...
jenkins build this hip please