Re: [eigen] About dropping C++03 compatibility

[ Thread Index | Date Index | More Archives ]

Dear Daniel,

Am 20.02.19 um 10:12 schrieb Daniel.Vollmer@xxxxxx:

Let me try to explain:

We (occasionally) write scientific software that runs on HPC clusters (and is often distributed in source form). It has dependencies which we don't necessarily control. We may have to use buggy vendor compilers, which we cannot "just" upgrade (they were sold and validated for the system they're on).

At least for the big the German HPC systems I'm aware of, the situation improved a lot in recent years
with the introduction of the 'module' environment.

E.g, at the LRZ SuperMuc we have as today gcc/8 and intel/19.0 and big zoo of older versions, see 'module avail'.

The situation is often worse on smaller HPC systems run by institutes. However, at least I always
had good experience with the system administrators concerning requests for newer compiler versions.
This might not always be the case, but I don't expect it to be a large scale problem.
In addition there's a strong tendency towards  virtualized systems, packing everything into container like
docker or singularity, rendering the problem of outdated HPC software infrastructure irrelevant.

And if that doesn't work you can still compile and link statically. This can be a little bit tricky to get it run,
especially with the intel compiler and if you have to link to special MPI versions, but I've done it in former times.
Definitely not convenient, but often feasible.

I am still hesitant to move from Eigen 3.2.x to Eigen 3.3.x due to performance regressions (which have been improved upon by point releases, but are still present – and are very difficult to locate / boil down to a simple repo). I'd love the features (and bug-fixes) of 3.3!

Which actually fits nicely to Michaels suspicion.

Best regards,

Mail converted by MHonArc 2.6.19+