Re: [FLASH-USERS] Re: [FLASH-BUGS] HDF5 crash

Hi!

I compiled everything flash from scratch using prebuild hdf5-1.4.5 lib 64
bit and it is as you said, windtunnel produces error dump but keeps on
running - dumoing an error everytime it is supposed to write plotfiles,
sedov runs without any trouble. I removed the divide-domain from the
windtunnel and it still produced the errordump at the first plot and chkpt
file, then runs smoothly - haven't seen a second chkpt file yet though.
Thought removing devide domain will fix it, since I cannot see what is so
special about windtunnel other than its geometry, but not quite.

I will try to compile my own hdf5 now and will look for another pet
testcase.

Markus.

On Wed, 18 Jun 2003 19:02:05 -0500, JB Gallagher
<jbgallag@flash.uchicago.edu> wrote:

> Markus,
>
> It appears that this may be a bug in the windtunnel problem. I am
> able to
> reproduce the error using all combinations of hdf5 versions and i/o
> modules in
> Flash, even the parallel-netcdf i/o module reports a corresponding
> library error
> on this problem. Compiling 32 versus 64-bit makes no difference
> either. There
> appears to be something wrong in the problem setup. So the good news
> is you can
> quit worrying about your hdf5 setups/compiler combinations, it is
> safe to say
> that this is definitely not an hdf5 problem. It appears this library
> error
> happens in both 2d and 3d cases, but only on the first checkpoint
> file, on our
> SGI's anyway the code continues to run and produce additional
> checkpoint and
> plotfiles without producing the hdf5 library error output. I haven't
> looked at
> the data to see if it makes sense, but can you verify whether or not
> your run
> actually stops or not? I have included the flash.par I have used in
> these test
> cases, in both 2d and 3d 8 checkpoint files are produced and 7
> plotfiles are
> produced with this flash.par (I do 50 steps). In the 3d case I set
> maxblocks=2000
> and compiled it in 64-bit using hdf5-1.4.5 (not post2), in the
> 32-bit runs I'm
> using hdf5-1.4.3.
>
> I will look at this closer tommorow to see why the first checkpoint
> file is
> producing this error and let you know what I find out.
>
>
> --Brad
>
>
> Tomasz Plewa wrote:
>
> > Markus -
> >
> > It is true that serial HDF5 version is default in FLASH - you
> can
> > check in your Modules file (created by setup in FLASH root
> directory).
> > And I am positive serial is also default when building HDF5.
> >
> > I am copying Brad Gallagher on your message. Brad has lots of
> > experience with HDF and sharing rest of your observations with
> him
> > offers greatest chance to get some sensible answer. If that fails,
> then
> > I think flash-users is the place to post a summary.
> >
> > Brad - do any issue/question raised by Markus ring a bell?
> >
> > Tomek
> > --
> > On Wed, Jun 18, 2003 at 10:16:40PM +0100, M.S.Gross@hw.ac.uk
> wrote:
> > > Hi!
> > >
> > > I give it up for today. I think I had all combination by now.
> Even tried the
> > > prebuild binaries, same thing.
> > > Did you have a look into my flash.log file? The serial hdf5 is
> default,
> > > isn't it? At least I did the usual -3d -auto etc, nothing about
> i/o
> > >
> > > My last attempt is to try to get hdf5 running with fortran
> enabled and ifc -
> > > seems to be not my day today (the lift even dropped me off at
> the wrong
> > > floor earlier ;-)
> > >
> > > I got a testprogram running with hdf5 on sgi, worked fine until
> I queried it
> > > for the dimensions of a dataset, well it got it nearly right but
> one
> > > dimension was several thousand times off the mark. From what
> I've seen (and
> > > I have not understood a lot) in the error output generated by
> flash (of
> > > course rather hdf5 in flash) there was something about zero
> sized dimensions
> > > ....
> > > Would you mind asking your systems person what exact version you
> are using,
> > > pre-compiled or build fom source and if there were any tweaks? I
> mean it
> > > really shouldn't be difficult, hdf4 was a peace of cake ...what
> are they
> > > doing?
> > >
> > > Anyhow ... we'll see whats happening.
> > >
> > >
> > > Markus.
> > >
> > >
> > > On Wed, 18 Jun 2003 14:12:11 -0500, Tomasz Plewa
> <tomek@flash.uchicago.edu>
> > > wrote:
> > >
> > > > OK - thanks for the warning, Markus. I am copying our system
> person
> > > > on
> > > > this message so we would not suffer as you do.
> > > >
> > > > And let us know what happens when you use the older version.
> > > >
> > > > Tomek
> > > > --
> > > > On Wed, Jun 18, 2003 at 07:51:08PM +0100, Markus Gross
> wrote:
> > > > >
> > > > > We have the 7.4 compiler - should never ever have asked for
> it
> > > > ...
> > > > >
> > > > > On another system we have the 7.3.1.3m and it's compiling at
> the
> > > > moment. Went
> > > > > back to hdf5 1.4.4 as well, we'll see.
> > > > >
> > > > > So, you are warned not to upgrade your compiler ;-)
> > > > >
> > > > > Markus.
> > > > >
> > > > >
> > > > > On Wednesday 18 June 2003 19:43, you wrote:
> > > > > > > Just a short update: no joy.
> > > > > >
> > > > > > This is sad, actually.
> > > > > >
> > > > > > > I am trying an older compiler, we'll see.
> > > > > >
> > > > > > Hmm. Given we use it every day, I would expect -n32
> working
> > > > for
> > > > > > us.
> > > > > >
> > > > > > For the record, on our SGI systems we have 7.3.1.3m and we
> are
> > > > running
> > > > > > IRIX 6.5.16m.
> > > > > >
> > > > > > Tomek
> > > > >
> > > > > --
> > > > >
> _______________________________________________________________
> > > > >
> > > > > Markus Gross AMIMechE BEng (Hons.) Mechanical Engineering
> > > > >
> > > > > Heriot Watt University Edinburgh
> > > > > School of Engineering and Physical Sciences
> > > > > James Nasmyth Building
> > > > > Edinburgh
> > > > > EH14 4AS
> > > > > UK
> > > > >
> > > > > Member of IMechE, SPIE, CSME and VDI
> > > > >
> _______________________________________________________________
> > > > >
> > > > > further contact:
> > > > >
> > > > > Phone : +44 (0) 131 451 4737
> > > > >
> > > > > UNiX talk: talk markus@lasersim.mce.hw.ac.uk
> > > > >
> > > > >
> _______________________________________________________________
> > > > >
> > > > > "Plans are a place to begin," Grove said. "They rarely
> deliver
> > > > > you to where you expect. Make your plans knowing you are
> going
> > > > > to throw them away."
> > > > >
> > > > >
> _______________________________________________________________
> > > > >
> > > > >
> > > > >
> > > >
> > > > --
> > > >
> > >
> > >
> > >
> > >
> > >
> > >
> ________________________________________________________________
> > >
> > > DISCLAIMER:
> > >
> > > This e-mail and any files transmitted with it are confidential
> > > and intended solely for the use of the individual or entity to
> > > whom it is addressed. If you are not the intended recipient
> > > you are prohibited from using any of the information contained
> > > in this e-mail. In such a case, please destroy all copies in
> > > your possession and notify the sender by reply e-mail. Heriot
> > > Watt University does not accept liability or responsibility
> > > for changes made to this e-mail after it was sent, or for
> > > viruses transmitted through this e-mail. Opinions, comments,
> > > conclusions and other information in this e-mail that do not
> > > relate to the official business of Heriot Watt University are
> > > not endorsed by it.
> > >
> ________________________________________________________________
> >
> > --
>
> --
> Brad Gallagher
> ASCI Flash Center
> jbgallag@flash.uchicago.edu
>
>
>

________________________________________________________________

DISCLAIMER:

This e-mail and any files transmitted with it are confidential
and intended solely for the use of the individual or entity to
whom it is addressed. If you are not the intended recipient
you are prohibited from using any of the information contained
in this e-mail. In such a case, please destroy all copies in
your possession and notify the sender by reply e-mail. Heriot
Watt University does not accept liability or responsibility
for changes made to this e-mail after it was sent, or for
viruses transmitted through this e-mail. Opinions, comments,
conclusions and other information in this e-mail that do not
relate to the official business of Heriot Watt University are
not endorsed by it.
________________________________________________________________
Received on Thu Jun 19 16:46:04 2003