SPACK v0.17 on RH6

From v0.17, spack does not run on RH6.
The first reason is; python2.7 is needed, and the second; clingo needs to be installed by users, instead of bootstrapping.

I use miniconda for python2.7. Clingo can be installed via miniconda as
$ conda install -c potassco clingo

And also you may want to disable bootstraping as;

% spack bootstrap untrust github-actions

%  spack bootstrap disable

Please see
https://spack.io/changes-spack-v017/

Android 12 on Xiaomi A1

My A1 works on Pixel Experience Android 12. Looks good. Actually, feel better than Poco F3 with MIUI.

Invited talk

 We were invited to a session of the 181th meeting of ASA.

https://www.researchgate.net/publication/356367868_High-performance_computing_for_long-range_underwater_acoustics

Guest Investigator @Woods Hole Oceanographic Institution

 From this December, I am also a Guest Investigator at Woods Hole Oceanographic Institution.

I will work on HPC in underwater acoustics.

Hope I can provide some good achievements and make any feed-back to CTBT.

heFFTe

 Installation of heFFTe with MKL on SPACK;

% spack install heffte%intel@2021.4.0+mkl ^intel-oneapi-mpi@2021.4.0 ^intel-mkl threads=openmp

To use installed Intel OneAPI and CMAKE on SPACK

In order to avoid multiple installations of Intel-OneAPI and CMAKE, we can use installed ones.
Please install Install OneAPI and CMAKE via APT and put the packages.yaml under ~/.spack/ whose contents are as follows;

 $ cat ~/.spack/packages.yaml

packages:

intel-oneapi-mkl:

buildable: false

externals:

- spec: intel-oneapi-mkl@2021.4.0

prefix: /opt/intel/oneapi/

intel-oneapi-mpi:

buildable: false

externals:

- spec: intel-oneapi-mpi@2021.4.0

prefix: /opt/intel/oneapi/

cmake:

buildable: false

externals:

- spec: cmake@3.16.3

prefix: /usr

buildable: false prevents us from installing packages via SPACK.

MPI/OpenMP hybrid with Intel MPI

  mpirun -np 8 -genv OMP_NUM_THREADS=16 -hosts node2-ib,node3-ib,node4-ib,node5-ib -ppn 2 ./sol

To run an OpenMP/MPI hybrid code with IntelMPI, the above command is needed.

The meanings are;

8 MPI processes (-np 8),

16 OpenMP threads (-genv OMP_NUM_THREADS=16) ,

2 MPI processes per node (-ppn 2).