ABINIT Discussion Forums

Which compiler and version are you using?

Your compiler is looking for the Atlas library for linear algebra. You should also install it from Macports and/or make sure you have installed all its development-related components.

You're not the first who wants to rewrite fldiff and this is a topic that comes back periodically. However, Matteo is already on it and is taking care of preserving its integration within the test suite as well as its proper working on multicore architectures. If you wish to help, the best is probab...

I'll have a look at it as soon as I can.

When you do a "make multi", the multi_nprocs parameter can vary between 2 and the total number of processors, e.g. 8 in your case. How many you use only depends on you and how many resources you want to keep for the other applications you're running at the same time on this computer. The n...

The NetCDF version shipped with Abinit dates back from a few years ago. It is possible that GCC 4.8, which is stricter on some aspects than earlier versions of GCC, produces an error where formerly they were issuing a warning. Directly installing NetCDF from Macports or similar and pointing Abinit t...

This is actually a regression in the build system. I fixed incorrect detection of Intel compilers a few years a go. In any case, as Matteo says, this is not an issue, since abinit does not include C++ code.

Could you send us the output of "icpc -V" on your machine? Thank you in advance.

We could indeed use a heuristic based of the frequency of calls to a routine during a run, for which what you propose is a good approximation.

+1

This is indeed a great idea to break the old habits and design before coding, for once.

I'm not an expert in Mac, but I know that GCC 4.8 does provide the iso_c_binding module. You may thus be able to solve this issue by looking for the iso_c_binding.mod file on your hard disk and then use the following option to configure: FPPFLAGS_EXTRA="-I/directory/containing/the/iso_c_binding...

Could you send /home/jbaltrus/ABINIT/abinit-7.6.2/build/fallbacks/sources/wannier90-1.2.0.1/config.log as well?

You should not worry so much about these 2 errors. Abinit uses a single set of reference files and tolerances for its hundreds of tests, which means that there will always be architecture/compiler/optimizations combinations going out of the allowed intervals. The error messages you get are just tell...

FYI: the features of the future build system of Abinit are now frozen. I'll start the rewrite at the beginning of April and document it on the Abinit Wiki at the same time.

AtomPAW in Abinit is only used for some tests when a new version of AtomPAW is released. If you intend to use it for production, the best is to build it separately.

I'll look at the issue when I have some time, but it is possible that it has already been fixed in 7.7.

This is necessary for C / Fortran bindings and will become even more critical as we keep on improving the scalability of Abinit.

I cannot propose any solution for now. As a temporary measure, you could enhance the config/wrappers/wrap-fc.in, so that it removes unwanted libraries from the command line. I suggest you to consult Matteo Giantomassi if you need help.

Intel compilers do not automatically enable the correct build options when you use OpenMP and MPI at once. You may be able to solve this issue with the following flags: CFLAGS_EXTRA="-mt_mpi" CC_LDFLAGS_EXTRA="-mt_mpi" FCFLAGS_EXTRA="-mt_mpi" FC_LDFLAGS_EXTRA="-mt_...

Thanks for the confirmations. This is now scheduled for the build-system rewrite.

Hi, As I am currently prepraring a major upgrade of the build system, I would like to give you all the opportunity to let me know about the new features you wish to see in the upcoming version, those you consider that should be removed, and those which should be modified. Any kind of comment regardi...

I moved this topic to "Configuration", where it actually belongs. Here you will find all necessary information in the red "Forum rules" box, at the top of each page.

For the manual and more: http://www.abinit.org/documentation

As usual: which version of Abinit? Which version of ScaLAPACK? Which architecture? ...

A config.log would also be welcome.

The best would probably be to ask directly to the developers of BigDFT (see the AUTHORS file in the bigdft-abi-1.0.4 tarball), since such fixes depend essentially on them.

Did you have a look at the molecular dynamics tutorial?

http://www.abinit.org/documentation/hel ... oldyn.html

Thank you for telling us whether the information available there is sufficient.

I introduced the --enable-mpi-trace option upon request from CEA in June 2009. If nobody use it anymore, I'll be happy to remove it.