On Fri, Mar 4, 2011 at 5:24 PM, <span dir="ltr"><<a href="http://ning.an">ning.an</a>@<a href="http://ghiocel-tech.com">ghiocel-tech.com</a>></span> wrote:<br><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
Dear Knepley,<br>
<br>
Thanks for your quick response.<br>
<br>
Yes, I have not pre-allocated sufficient space for the matrix, since I don't know the max. bandwidth. I got this matrix from other person. My task is to test PETSc solver for huge matrix size. I'm going to ask the person to give me that number. base on your experience, how fast could it be if I use "call
MatCreateSeqAIJ(comm,n,n,rowmax,PETSC_NULL,&mat)" ?</blockquote><div><br></div><div>That will be fast as long as no row is greater than rowmax long. However, why not just count first? This is easy and fast.</div>
<div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;"> <br>
Since I don't have the PETSc matrix yet, I couldn't use MatView() to write it in the binary format. Do you have another suggestion to generate matrix file in PETSc binary format? I'm not familiar with C. I couldn't dig it out from the source code. It is so appreciated If you get me an example.<br>
</blockquote><div><br></div><div>The C is the only reference.</div><div><br></div><div> Matt</div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
Have a good weekend.<br>
<br>
ning<br>
<br>
----- Original Message -----
<br>
<div style="width:100%;background-color:#E7E7EF"><b>From:</b> <a href="mailto:knepley@gmail.com" target="_blank">knepley@gmail.com</a></div><b>To:</b> <a href="mailto:petsc-users@mcs.anl.gov" target="_blank">petsc-users@mcs.anl.gov</a>
<br>
<b>Sent:</b> 3/4/11 5:44 PM
<br>
<b>Subject:</b> Re: [petsc-users] how to write the symmetric matrix in PETSc binary<br>
<br>
<div style="border-left:2px solid black;padding-left:5px"> On Fri, Mar 4, 2011 at 4:40 PM, <span dir="ltr"><<a href="http://ning.an" target="_blank">ning.an</a>@<a href="http://ghiocel-tech.com" target="_blank">ghiocel-tech.com</a>></span> wrote:<br>
<div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0pt 0pt 0pt 0.8ex;border-left:1px solid #CCCCCC;padding-left:1ex"> Hello There,<br>
<br>
I am new to PETSc and programming with FORTRAN. For the huge sparse matrix (200,000x200,000 symmetric), it is so slow to set the matrix in PETSc by reading them in matrix market format, which is far beyond our patience. Therefore, I guess that it would be much fast to load the huge matrix, if the matrix is in PETSc binary matrix format <br>
directly.<br>
</blockquote><div><br>
</div><div>It is probably slow because you have not preallocated the matrix correctly: <a href="http://www.mcs.anl.gov/petsc/petsc-as/documentation/faq.html#efficient-assembly" target="_blank">http://www.mcs.anl.gov/petsc/petsc-as/documentation/faq.html#efficient-assembly</a></div>
<div> </div><blockquote class="gmail_quote" style="margin:0pt 0pt 0pt 0.8ex;border-left:1px solid #CCCCCC;padding-left:1ex"> Please help on how to write the symmetric matrix in PETSc binary matrix format directly? <br>
</blockquote><div><br>
</div><div>You just save it using MatView(). For example:</div><div><br>
</div><div> <a href="http://www.mcs.anl.gov/petsc/petsc-as/snapshots/petsc-current/src/ksp/pc/examples/tutorials/ex2.c.html" target="_blank">http://www.mcs.anl.gov/petsc/petsc-as/snapshots/petsc-current/src/ksp/pc/examples/tutorials/ex2.c.html</a></div>
<div><br>
</div><div> Matt</div><div> </div><blockquote class="gmail_quote" style="margin:0pt 0pt 0pt 0.8ex;border-left:1px solid #CCCCCC;padding-left:1ex"> Both the guidance and examples are welcome. <br>
<br>
Ning</blockquote></div>-- <br>
What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead.<br>
-- Norbert Wiener<br>
</div>
</blockquote></div><br><br clear="all"><br>-- <br>What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead.<br>-- Norbert Wiener<br>