[GRASS-user] problems with r.cost

[Hamish]

Jonathan Greenberg wrote:
> 
> Hamish (and all): Paul is working with me on this project, and the
> system we are running it on is a 64-bit debian system with plenty of
> RAM/scratch space (and the version we are using is the 64-bit grass
> release) -- why would memory problems be an issue?  We tried lowering
> the % memory but it doesn't seem to help...

does it work for a smaller region or coarser resolution? for example try
changing it from 3 to 30m res.

ie can you determine if is it a 64bit computational error or an out-of-
resources error? (does r.cost work for the spearfish DEM on your system?)

I can't answer why memory problems would be an issue, but that doesn't
mean they aren't! Large mem grass is still untested territory in many
places, maybe you have discovered something new.

Did r.terracost work? How does that compare to r.cost? (if it's good,
I'm thinking maybe Laura would allow us to merge it in CVS as a
companion to r.terraflow, or package as a GEM)
Or does that use GRASS 5 sites as starting/end points and need porting?

Hamish