{chem}[intel/2023a] FHI-aims v240507

[sassy-crick]

(created using eb --new-pr)

[smoors]

@boegelbot: please test @ generoso

[boegelbot]

@smoors: Request for testing this PR well received on login1

PR test command 'EB_PR=20793 EB_ARGS= EB_CONTAINER= EB_REPO=easybuild-easyconfigs /opt/software/slurm/bin/sbatch --job-name test_PR_20793 --ntasks=4 ~/boegelbot/eb_from_pr_upload_generoso.sh' executed!

• exit code: 0
• output:

Submitted batch job 13708

Test results coming soon (I hope)...

- notification for comment with ID 2158999105 processed

Message to humans: this is just bookkeeping information for me,
it is of no use to you (unless you think I have a bug, which I don't).

[boegelbot]

Test report by @boegelbot
FAILED
Build succeeded for 0 out of 1 (1 easyconfigs in total)
cns1 - Linux Rocky Linux 8.9, x86_64, Intel(R) Xeon(R) CPU E5-2667 v3 @ 3.20GHz (haswell), Python 3.6.8
See https://gist.github.com/boegelbot/2c680794ff300f21ff5203337a665c60 for a full test report.

EDIT: i missed that it's a manual download

[sassy-crick]

Test report can be found here

[Flamefire]

To honor toolchain options oneapi and usempi:

Suggested change

	configopts = ' -DCMAKE_Fortran_COMPILER="$MPIF90" -DCMAKE_CXX_COMPILER="icpc" -DCMAKE_C_COMPILER="icc" '
	configopts = ' -DCMAKE_Fortran_COMPILER="$FC" -DCMAKE_CXX_COMPILER="$CXX_SEQ" -DCMAKE_C_COMPILER="$CC_SEQ" '

According to https://fhi-aims-club.gitlab.io/tutorials/cmake-tutorial/cmake_tutorial they need an MPI fortran compiler --> 'usempi': True in toolchainopts

[sassy-crick]

Thanks for that! I done a bit more digging and with the help of the FHI-aims team I think I got it all sorted so we can use oneAPI as the compiler now. From what I understand, there are only a few places where MPI is being used and thus when you link it is important to use MPI but not when compiling. I got told that eliminates problems with wrapper-scripts. Nonetheless, I used MPI for all three languages. Let me know what you think of the improvements done.

[Flamefire]

Are you sure any of this is required?

• -fc etc. should be set by EasyBuild via the I_MPI variables
• finline-functions is set by -O2 and above
• -std shouldn't be required as the defaults for those intel compilers are likely at least that

Removing might help updating the EC later.

[sassy-crick]

I took these flags from the template they provided. The -finline-functions comes from the dropping of -ip and that was the replacement I found.
That said: you are right, it works without them as well. I will update the PR and will include one which makes use of an external ELPA installation.

[sassy-crick]

I have included a second EasyConfig file which is using an external ELPA installation. The thing I am not sure how to handle best is: which ELPA kernel has being used? ELPA itself is setting the kernel but I don't know how to retrieve that information for the FHI-aims installation.