Jim Kusznir wrote:
>Ok, trying now. This does bring up a question I've been meaning to
>ask. I need to find a way to analyze the efficiency of parallel jobs,
>and the only way I've known about in the past is through vampire. I
>do not have the software license presently (and last I looked, it was
>a commercial licensed product). Does this configure flag suggest that
>openmpi has native support for generating vampire trace files? Is
>there any open source tools for analyzing them (or another route I
>should be looking at)?
Writing up a FAQ entry on this subject remains on my "to do" list.
(Hasn't gotten dropped off the list, but hasn't gotten done either.)
Open MPI *does* have VampirTrace in it, but then VampirTrace is
available freely anyhow. The question is what you're going to do with
the trace data. You can get a Vampir license. If you want an
open-source tool for analyzing VT traces, I *think* you can use TAU, but
am not sure. There are also crude tools like otfprofile and otfdump.
Leo P recently raised some number of these issues on the "devel" alias
and was told by the Vampir people (per e-mail not reported to the devel
archives) that to look at VT data, you really need to get Vampir.
Evaluation licenses are apparently available.
I thought I had sent e-mail out on this subject before, but cannot find
it. Other open source performance tools supposedly include Jumpshot
(associated with MPICH, but usable with OMPI), mpiP, FPMPI, and IPM.
(I'm just dropping a bunch of acronyms on you here. I'm not that
familiar with any of these tools.)
If you're not so much interested in open source as in "free download",
then another option is Sun Studio Performance Analyzer, which uses VT
tracing and provides a GUI and command-line tool to look at the data.
Use was described in this message:
Personally, I'd recommend Sun Studio. (Fair disclosure: I'm a Sun