<div class="gmail_quote">On Wed, Nov 10, 2010 at 20:55, SUN Chun <span dir="ltr"><<a href="mailto:Chun.SUN@3ds.com">Chun.SUN@3ds.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
<div id=":33n">Sorry this might doesn't make any sense but I have to ask ... Does PETSc have the capability to use out-of-core disk space if I set a cap for memory usage? I looked around in the src but I could be missing something.<br>
</div></blockquote><div><br></div><div>There is no explicit out-of-core capability, be warned that it will unavoidably make things much slower (even if written specially for out-of-core) but you could mmap a big file and have the matrices use that, or you can just create a big swap file and let the OS handle it. Note that kernels can actually do a pretty good job of handling paging, so it's not a clear win for an application to manage it themselves. See e.g. <a href="http://queue.acm.org/detail.cfm?id=1814327">http://queue.acm.org/detail.cfm?id=1814327</a></div>
<div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;"><div id=":33n">
What I can think of is to use MatShell if that's the only option...</div></blockquote></div><br><div>You can certainly do this, but if it makes sense to store your matrix in an assembled format, you should probably just assemble it and let the OS handle paging.</div>
<div><br></div><div>Jed</div>