Toolchain 202501

[QuantumMisaka]

Reminder

• Have you linked an issue with this pull request?
• Have you added adequate unit tests and/or case tests for your pull request?
• Have you noticed possible changes of behavior below or in the linked issue?
• Have you explained the changes of codes in core modules of ESolver, HSolver, ElecState, Hamilt, Operator or Psi? (ignore if not applicable)

Linked Issue

• Software version update : ELPA, OpenMPI, MPICH, CMake, LibXC
• Add Compiler: AOCC (option: --with_amd=system)
• Add Mathlib： AOCL (option: --math_mode=aocl)
• Adjust minor description based on CP2K toolchain

Unit Tests and/or Case Tests for my changes

• A unit test is added for each new feature or bug fix.

What's changed?

• Example: My changes might affect the performance of the application under certain conditions, and I have tested the impact on various scenarios...

Any changes of core modules? (ignore if not applicable)

• Example: I have added a new virtual function in the esolver base class in order to ...

[QuantumMisaka]

@goodchong @dzzz2001 Community and me need a detailed tutorial for GPU-ELPA and ABACUS-GPU-LCAO installation, also for solve #5872

[QuantumMisaka]

Cannot Solve problem while using AOCC in ELPA compling

flang: error: unsupported option '--whole-archive'
flang: error: unsupported option '--no-whole-archive'
flang: error: unknown argument: '-soname'
flang: error: no such file or directory: 'libelpa_openmp.so.19'

I''ve no knowledge if ELPA support to be compiled by AOCC. Temporally give up

[QuantumMisaka]

#5982 Marks the error in AOCC/AOCL compiling while avoiding the problem from flang

Also, I need some help (or effort) to incorporate LAPACK in toolchain for Fully utilize the AOCL package

I have no idea to fix it. All these is this version of toolchain update

[mohanchen]

How about the following tutorials, maybe we should update them?

https://mcresearch.github.io/abacus-user-guide/abacus-gpu-lcao.htmlc
https://mcresearch.github.io/abacus-user-guide/abacus-hpc.html
https://mcresearch.github.io/abacus-user-guide/abacus-gpu.html

[QuantumMisaka]

How about the following tutorials, maybe we should update them?

https://mcresearch.github.io/abacus-user-guide/abacus-gpu-lcao.htmlc https://mcresearch.github.io/abacus-user-guide/abacus-hpc.html https://mcresearch.github.io/abacus-user-guide/abacus-gpu.html

I'll try it. This PR can be merged later, but we may need guide for installing ELPA-GPU, as the guide in ELPA official website is not very clear

[QuantumMisaka]

How about the following tutorials, maybe we should update them?

https://mcresearch.github.io/abacus-user-guide/abacus-gpu-lcao.htmlc https://mcresearch.github.io/abacus-user-guide/abacus-hpc.html https://mcresearch.github.io/abacus-user-guide/abacus-gpu.html

The reference page https://github.com/marekandreas/elpa/blob/master/documentation/INSTALL is 404, and the compliation tutorial (especially a robust tuturial suitable for all nvidia-GPU platform) of GPU-LCAO-CUSOLVERMP & GPU-LCAO-ELPA is also insufficient.

[QuantumMisaka]

LibXC 7.0.0 compliation problem is fixed by #5905. So I'll add 7.0.0 update

[QuantumMisaka]

With the help of @yizeyi18, the AOCL environment management is fully completed and will be commited in this PR, But the clang++ in AOCC still cannot be the compiler of ABACUS due to the same error in #5982

[QuantumMisaka]

Based on testing results, the AOCC-AOCL toolchain with ABACUS installation by GNU-AOCL can be done by toolchain while avoiding flang in AOCC. So this is the default toolchain plan. User can turn-on flang by --with-flang=yes if like to.

Reference Task: MPI16-OMP1 LCAO-genelpa for Fe5C2(510) [Fe80C36], ABACUS commit 1fa5e3a , Hardware: AMD-EPYC-7b12

• gcc-aocl (aocc-toolchain avoid flang): 38s/scf-step
• gcc-openblas 40s/scf-step
• gcc-aocl (gcc-toolchain) 41s/scf-step
• aocc-openblas-aocl （38-42）s/scf-step （unstable）
• intel-oneapi 46s/scf-step
• gcc-aocl (with flang): 50s/scf-step

[QuantumMisaka]

This version of toolchain will not update in this PR until severe bug is found

[QuantumMisaka]

@mohanchen I consider that this PR can be merged, and GPU-LCAO installation update will be the next target in the future PR

[mohanchen]

Excellent, I will merge it.