[GRASS-user] Configuration options

Massimiliano Alvioli nocharge at gmail.com
Fri Sep 13 05:37:21 PDT 2019

Hi all,

> Last evening I re-imported a huge (4.4G) .gdb of all wetlands in Oregon
> adding 'snap=1e-0.7' to the command line because grass suggested adding the
> snap option when it completed the initial import (which took about 3 hours
> on that host). Running overnight, the re-import took just less than 10 hours
> to complete. I forgot to increase the available memory for v.in.ocr to use.
> I'd like to have grass modules take advantage of the hardware to the extent
> they can.

I doubt there is anything one can do to speed up any vector import operation,
both because it is mostly read/write (original file on disk/map in
vector folder)
and because it uses a database - both of them are inherently serial operations,
at least on any commercial hardware. OpenMP works well on distributing CPU
usage, but if you need to read/write you can only access so much disk space
per unit of time. The same goes for database access, I believe. For sure this is
the case in SQLite in which the whole DB is a single file.


More information about the grass-user mailing list