[Users] Slow reading of Carpet data in visit 2.13

[Erik Schnetter]

Yosef

There was a small change -- we added an additional attribute that makes it
easier (or possible) to reconstruct the buffer region, so that it can be
excluded from visualization. I don't think VisIt makes use of this
information yet. Performance wise this should make no difference at all.

HDF5 also allows setting parameters that influence caching. Maybe VisIt
differs in its behaviour there? But now I'm just guessing; I'm hoping that
others might know something for real.

-erik


On Mon, Jun 4, 2018 at 3:49 PM, Yosef Zlochower <yosef at astro.rit.edu> wrote:

> Thanks.
>    I was trying to load the same file under both versions of visit.
> Both version seem to use hdf5 version 9, which is the one automatically
> downloaded and installed by the build_visit script. The output file was
> created using the ET_2015_11 branch. Has the carpet output format
> changed since then?
>
> On 06/04/2018 03:36 PM, Erik Schnetter wrote:
> > Yosef
> >
> > Without knowing any details: Newer HDF5 versions (1.10 and later) are
> > supposed to be more efficient in handling metadata. I do not know
> > whether this applies to writing as well as to reading. It might be that
> > your VisIt versions are linked against different HDF5 library versions.
> > (I assume that you are testing against the same HDF5 file.)
> >
> > -erik
> >
> >
> > On Mon, Jun 4, 2018 at 2:43 PM, Yosef Zlochower <yosef at astro.rit.edu
> > <mailto:yosef at astro.rit.edu>> wrote:
> >
> >     Hi,
> >         I was trying to load CarpetIOHDF5 2D data with visit 2.10 and
> >     2.13. I
> >     compiled both using the build_visit scripts with the --hdf5 option.
> The
> >     2.13 takes much longer to load.
> >
> >     For example, with a file containing 3 timelevels, 2.10 takes under a
> >     second to load, and 2.13 takes about 7 seconds. This gets much worse
> as
> >     the number of time levels is increased. It took 2 minutes to open a
> >     file
> >     with 10 timelevels in 2.13, but under a second with  2.10
> >
> >     Has anyone else seen this? Is there a workaround?
> >
> >
> >     --
> >     Dr. Yosef Zlochower
> >     Center for Computational Relativity and Gravitation
> >     Associate Professor
> >     School of Mathematical Sciences
> >     Rochester Institute of Technology
> >     85 Lomb Memorial Drive
> >     Rochester, NY 14623
> >
> >     Office:74-2067
> >     Phone: +1 585-475-6103
> >
> >     yosef at astro.rit.edu <mailto:yosef at astro.rit.edu>
> >
> >     CONFIDENTIALITY NOTE: The information transmitted, including
> >     attachments, is intended only for the person(s) or entity to which it
> >     is addressed and may contain confidential and/or privileged material.
> >     Any review, retransmission, dissemination or other use of, or taking
> >     of any action in reliance upon this information by persons or
> entities
> >     other than the intended recipient is prohibited. If you received this
> >     in error, please contact the sender and destroy any copies of this
> >     information.
> >     _______________________________________________
> >     Users mailing list
> >     Users at einsteintoolkit.org <mailto:Users at einsteintoolkit.org>
> >     http://lists.einsteintoolkit.org/mailman/listinfo/users
> >     <http://lists.einsteintoolkit.org/mailman/listinfo/users>
> >
> >
> >
> >
> > --
> > Erik Schnetter <schnetter at cct.lsu.edu <mailto:schnetter at cct.lsu.edu>>
> > http://www.perimeterinstitute.ca/personal/eschnetter/
> >
> _______________________________________________
> Users mailing list
> Users at einsteintoolkit.org
> http://lists.einsteintoolkit.org/mailman/listinfo/users
>



-- 
Erik Schnetter <schnetter at cct.lsu.edu>
http://www.perimeterinstitute.ca/personal/eschnetter/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.einsteintoolkit.org/pipermail/users/attachments/20180604/7051ec66/attachment-0001.html