On Mon, Nov 4, 2013 at 10:52 AM, Kirk, Benjamin (JSC-EG311) <benjamin.kirk@nasa.gov> wrote:
On Nov 2, 2013, at 2:42 PM, John Peterson <jwpeterson@gmail.com> wrote:

>
>> I'll still work to refractor the adjacency build, but that'll take more concentration than I can muster this morning!
>
> OK, but you might not even want to bother!  We were around 5.5-6 Gb total memory (on 2 cores) on the SFC and Centroid partitioners, so we're already in that same ballpark for Metis.

…and that is now done; the metis interface should build the CSR graph representation in-place now.  That should produce a line marginally lower than the current green 'best' implementation at https://docs.google.com/file/d/0B9BK7pg8se_iQXpnT0x6aEY5aTg/edit?usp=sharing

Nicely done yet again!  The new implementation ran faster, and cut .23 Gb of memory.  That is a pretty huge memory savings for a 2 Gb mesh, especially considering the graph is only actually formed on one processor now.

https://drive.google.com/file/d/0B9BK7pg8se_iSTZMNXFtUFZ3QTQ/edit?usp=sharing

--
John