[Debichem-devel] Bug#687264: Bug#687264: aces3: MPI communication error

[Michael Banck]

severity 687264 serious
thanks

On Tue, Sep 11, 2012 at 12:01:19PM +0200, Michael Banck wrote:
> When I run a CCSD calculation (regular HF or MP2 calculation seem fine),
> I get the following error and the calculation aborts when using 2 (or
> more) MPI processes:
> 
>  Error: Cannot find array        32526  in served array table
>  ABORT JOB on server            1  SERVER TABLE:

I now ran the whole test suite (though reduced to be less demanding for
each test), and it turns out at the following 18 out of the 41 tests
fail:

1.2.3.1 1.2.3.2 1.3.1.1 1.3.2.1 1.4.3.1 1.4.3.2 2.2.3.1 2.2.3.2 2.3.1.1
2.3.2.1 2.4.3.1 2.4.3.2 3.1.2.1 3.1.4.1 3.1.4.2 3.2.2.1 3.2.2.2 3.2.3.1

Those fail even if I compile using gfortran-4.4, which helps a bit.  So
in essence, CCSD gradients, HF and MP2 hessians and any EOM-CCSD
calculations fail, leaving only HF/MP2 gradients and CCSD/CCSD(T)
energies (which do not need distributed arrays and MPI) functional.
This reduces the applicability of ACESIII so much that we should
consider this bug release critical.

The upside is that I've been working on a patch and I think I might be
able to fix it.


Michael