[Beowulf] value of parallel programming experience (was: Checkpointing using flash)

[Bogdan Costescu]

On Mon, Sep 24, 2012 at 6:57 PM, Andrew Holway <andrew.holway at gmail.com> wrote:
> In Germany, at present, there is I believe a
> fairly significant net surplus if compute resource as our scientists
> try to wrap their heads around parallel programming to take advantage
> of this exponentially increasing resource.

I beg to disagree on both parts of the phrase.

First, Germany has indeed a significant amount of HPC resources, but I
wouldn't call them "net surplus". If you know of HPC resources which
are lightly loaded, please let me know and I'll pass the info to the
people with a chronic lack of compute time :)

Secondly, there are quite a number of scientists in Germany who
already know parallel programming well. But I've listened to several
talks and following discussions on what needs to be done to take
advantage of their skills. Everybody agrees that something needs to be
done, to make these skills more valuable, but in the end not much is
felt by those scientists. Somehow adding "parallel programming
experience" to a CV doesn't seem to increase chances of getting hired
or a higher income. Other criteria seem much more important... but
these other criteria are often not correlated with HPC knowledge.
Which then results in "scientists try to wrap their heads around
parallel programming" as you mention. Some of the scientists with the
valuable knowledge choose to go away from Germany; some might
eventually come back but on a higher position (afterall, they have the
foreign experience!), where the parallel programming knowledge is not
important or the busy schedule doesn't allow using it in practice.
Sure, there are also exceptions... but as the problem is already
recognized and discussed, the exceptions remain few.

Not sure if this is limited to Germany. Any foreign opinions ?

Cheers,
Bogdan