miscellaneous fixes

[gcongiu]

• use pthread instread of omp in multi_monitor.cpp test
• fix path for hipcc
• use papi@master as default version of papi
• print the test name in run script

[eugeneswalker]

Thank you for this!

We are deploying E4S 23.08 on a number of systems right now. I will try this updated test and report back here ASAP.

[gcongiu]

@eugeneswalker I added a couple more fixes as I notices the run.sh script was not setting up the environment for the test correctly, as result the tests were not picking up the spack PAPI but the system installed PAPI.

[eugeneswalker]

@eugeneswalker I added a couple more fixes as I notices the run.sh script was not setting up the environment for the test correctly, as result the tests were not picking up the spack PAPI but the system installed PAPI.

One problem I just noticed is, we do not as a policy deploy branch versions of packages as part of E4S, so papi@master would not be available in our facility deployments. We only deploy tagged or versioned releases.

[gcongiu]

@eugeneswalker I added a couple more fixes as I notices the run.sh script was not setting up the environment for the test correctly, as result the tests were not picking up the spack PAPI but the system installed PAPI.

One problem I just noticed is, we do not as a policy deploy branch versions of packages as part of E4S, so papi@master would not be available in our facility deployments. We only deploy tagged or versioned releases.

Ok makes sense. We can “suspend” this PR until next Papi release.

[eugeneswalker]

Ok makes sense. We can “suspend” this PR until next Papi release.

Sounds good. I just converted it to a draft for now. Let us revive ASAP.

[wspear]

I see that we are releasing e4s with papi@7.1.0 now. Are we good to finalize and commit this change (maybe specifying papi@7.0.0: ?)