[GRASS-user] Re: r.in.xyz memory error

[Hamish]

> Ken wrote:
> > I'm using r.in.xyz to import an ascii x/y/z file with about 2
> > million points.   While r.in.xyz is scanning data  I monitor it
> > using "top" and I can see a VIRT memory size of 863m and the RES
> > memory climbs fairly quickly up to that amount and then I get the
> > following error: ERROR: G_calloc: out of memory
> >
> > I have 1GB physical ram on my system + 1GB swap space.   Is this 
> > expected behaviour for r.in.xyz? Any suggestions for getting it to
> > run successfully within the ram constraints of my system?
> >
> > Or, perhaps there's a better way to do what I'm trying to do?  I
> > have a postgres table with 10 million x,y,z points (which will grow
> > to about 100m).  I want to perform queries to extract various
> > regions of points and create raster layers in grass from those
> > points.  Currently I'm writing my query results to an output file
> > and then importing with r.in.xyz.   Is there a better way to do
> > this which might solve my memory issue above?
..
> ooops... should read the man pages more carefully.  "percent="


The memory requirements depend on: the raster region size (g.region);
the type of output map (CELL,FCELL,DCELL), and the statistical method
used.

The number of input points shouldn't matter as long as it can open the
file.

see also the new (6.3-cvs) v.in.ascii -r flag, "Only import points
falling within current region". (similar v.out.ascii -r)



Hamish