[petsc-users] mumps running out of memory, depending on an overall numerical factor?

[Jed Brown]

Dominic Meiser <dmeiser at txcorp.com> writes:
> I don't doubt that. In fact I did run this same problem on clusters as 
> well. It's just that some users (or customers) don't have access to 
> clusters or don't want to deal with them. Sometimes this is for 
> non-technical reasons. In such cases it's nice to have the option of 
> doing out-of-core.

Yup, that happens.  I just wanted to point out that if your metric is
efficiency, you want to solve 6 of these problems per day, and you have
other users in the organization, you may as well buy 5 nodes without SSD
and solve each problem in 50 minutes, rather than having 5 SSD-equipped
machines each solving in one hour.  The small SSD-equipped workstations
are less versatile for PDE-type workloads.  This is why most computing
facilities (for HPC-type workloads, as opposed to web servers/data
mining/etc) choose not to equip their nodes with SSDs.  Out-of-core
algorithms are for a certain low-end niche; it doesn't make much sense
at the high end, despite attempts to sell it there.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 835 bytes
Desc: not available
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20140205/1237e04c/attachment.pgp>