<div dir="ltr">Hi, <div><br></div><div>This is certainly a good idea. I can offer my examples too as and when I complete this small program.</div><div><br></div><div>Thank you very much for your help.</div><div><br></div><div>
Regards,</div><div>Nikhil<br><br><div class="gmail_quote">On Mon, Oct 13, 2008 at 5:02 PM, michael <span dir="ltr"><<a href="mailto:michael.bane@manchester.ac.uk">michael.bane@manchester.ac.uk</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;"><div><div></div><div class="Wj3C7c">On Mon, 2008-10-13 at 16:48 -0500, Robert Latham wrote:<br>
> On Mon, Oct 13, 2008 at 04:07:52PM -0500, Nikhil Laghave wrote:<br>
> > Hello All,<br>
> ><br>
> > I have been using Parallel HDF5 for our code to do parallel IO of huge files<br>
> > but the performance we're getting is not very encouraging. So I wanted to do<br>
> > the same with Pnetcdf, but I am not able find much tutorials on the web with<br>
> > simple examples to write out files in parallel. The ones that are there are<br>
> > the serial ones.<br>
> ><br>
> > So I was wondering if someone on this mailing list has a simple Pnetcdf<br>
> > example, that will help me in understanding how to use pnetcdf.<br>
><br>
> Hi<br>
><br>
> We really could use some more in-depth tutorials. Perhaps the best<br>
> examples right now are those in the 'test' directory. You can also<br>
> use the serial netcdf as a starting point: nfmpi_def_dim behaves just<br>
> like nf_def_dim, for example, though you might have to adjust the size<br>
> and types of some of your variables.<br>
><br>
> > Also, I am trying to create a simple file using pnetcdf and am getting this<br>
> > error. There seems to be some problem with the include file and I don't<br>
> > understand how to get rid of this error. My code is as follows:<br>
><br>
> I think I see what's wrong here. Two things, one small and one<br>
> larger.<br>
><br>
> > nfmpi_create(MPI_COMM_WORLD,FILE_NAME,<br>
> > & NF_NETCDF4,MPI_INFO_NULL,ncid)<br>
><br>
> The NF_NETCDF4 flag doesn't have any meaning in pnetcdf land, but I<br>
> don't think this why you are having problems.<br>
><br>
> > When i compile this using : ftn simple.f $PNETCDF, i get this error.<br>
> ><br>
> > PGF90-S-0021-Label field of continuation line is not blank<br>
> > (/usr/common/usg/pnetcdf/1.0.2/include/pnetcdf.inc: 1816)<br>
><br>
> line 1816 is:<br>
> #define NF_INT1_IS_C_SIGNED_CHAR 1<br>
><br>
> We require our fortran codes to use the C preprocessor. For many<br>
> compilers, this means using the '.F' suffix, or take Jim Edward's<br>
> advice.<br>
><br>
> People have used the portland group compilers with parallel-netcdf, so<br>
> I bet we can get you back on track in short order.<br>
><br>
> ==rob<br>
<br>
<br>
</div></div>I can dig out my examples and offer up, perhaps we could create a Wiki<br>
for people to upload (and improve!) examples?<br>
<br>
I don't know if there's {m}any UK {academic-related} people on the list<br>
but here's an event you may be interested in on 4 Dec 2008 in NW<br>
England: <a href="http://www.hpcx.ac.uk/about/events/parallel_io.html" target="_blank">http://www.hpcx.ac.uk/about/events/parallel_io.html</a> (albeit<br>
apparently more application than lower level focussed). Feel free, when<br>
booking/enquiring, to said I advertised this to you.<br>
<font color="#888888"><br>
M<br>
<br>
</font></blockquote></div><br><br clear="all"><br>-- <br>Regards,<br>Nikhil<br>
</div></div>