He/apple silicon

[ajwt]

Will's apple silicon update, hopefully ported.

[ajwt]

If someone could test this on apple silicon it would be great

[william-vanbenschoten]

Might be something to do at a later time, but I have -02 working on M3 (compared to -01).

[william-vanbenschoten]

Compiles fine on my MacBook Pro M3 running Sonoma version 14.6.1.

Compiler information:
gcc-14 (Homebrew GCC 14.1.0) 14.1.0
GNU Fortran (Homebrew GCC 14.1.0) 14.1.0
Lua 5.4.6
openmpi 5.0.3
hdf5 1.14.4-2

Note I installed coreutils, gcc, and cmake with brew.
OpenMPI and HDF5 was installed manually.

This was my cmake configuration script command:
python cmakeconfig.py --fc=mpifort --cc=mpicc --lua=lua --mpi --omp --det-size=64 --pop-size=64 --hdf5=/Users/vanbenschoten/Chemistry/codes/hdf5-1.14.4-2/hdf5 --type=relapplesilicon --uuid=False

I can run the test suite at a later date to confirm the compile is functional.

[william-vanbenschoten]

Here is test suite results:

All done. ERROR: only 356 out of 357 tests passed (1 warning, 2 skipped).
Failed test in:
fci/H2-RHF-cc-pVTZ-Lz_davidson/
Warning in:
fci/Ne-RHF-cc-pVDZ_davidson/
Skipped tests in:
ccmc/np1/Ne-RHF-cc-pVDZ-Lz-CCSDTQ5-multispawn-short-debug-logging/
fciqmc/np1/H2O-RHF-cc-pVTZ-debug-logging/

We had this issue before, there should be a commit that changes the Davidson test to lower its convergence criteria.

[william-vanbenschoten]

Here is the re-run test for FCI only with the updated Davidson benchmark.

All done. 28 out of 28 tests passed (1 warning).
Warning in:
fci/Ne-RHF-cc-pVDZ_davidson/

Seems to be in good order!