[GRASSLIST:9230] problems with r.watershed

[Wolf Bergenheim]

Hi!

I have a raster map (DEM) with a resolution of 10x10m and a total area
of 10x10 km, which gives me a raster with 1000 x 1000 = 1 000 000
pixels. When I run r.watershed on it it starts out nicely but then the
programs grinds to a halt in SECTION 5.

GRASS 6.1.cvs > r.watershed elevation=kevodem_fill3 flow=flowdir_fill2
basin=basin_fill stream=stream_fill visual=lammi_basin_visual threshold=40
Running: /opt/gis/grass-6.1.cvs/etc/water/ram el="kevodem_fill3"
ov="flowdir_fill2" t=40 ba="basin_fill" se="stream_fill"
di="lammi_basin_visual"

SECTION 1a (of 5): Initiating Memory.

SECTION 1b (of 5): Determining Offmap Flow. Percent Complete:

SECTION 2: A * Search.                      Percent complete:

SECTION 3: Accumulating Surface Flow.       Percent complete:

SECTION 4: Watershed determination.

SECTION 5: Closing Maps.

No when I do a ltrace on the process it gives me that it is running this:
G_get_color(22709, 0xbfc13408, 0xbfc13404, 0xbfc13400, 0xbfc1334c)
                                   = 1
G_get_color(22710, 0xbfc13408, 0xbfc13404, 0xbfc13400, 0xbfc1334c)
                                   = 1
G_get_color(22711, 0xbfc13408, 0xbfc13404, 0xbfc13400, 0xbfc1334c)
                                   = 1
G_get_color(22712, 0xbfc13408, 0xbfc13404, 0xbfc13400, 0xbfc1334c)
                                   = 1
G_get_color(22713, 0xbfc13408, 0xbfc13404, 0xbfc13400, 0xbfc1334c)
                                   = 1
G_set_color(22713, 157, 237, 217, 0xbfc1334c)
                                   = 1

This also tels me that it is running inside a while loop in
close_maps2.c lines 46 and 48. The first number, the r parameter is
constantly increased. Do I have to wait for this to reach the full
1000000 pixels? According to my calculations it will take some 47
days(!), give or take a few hours. Is there any way I can speed up this
process? I'm running on Debian linux with 1.4 GHz Pentium-M CPU. Or
should I use some other program to find the basins?

This is just a single step in a course excercise which is supposed to be
doable in 3 hours with Arc/INFO. Please help! I'm trying to show my
professor that GRASS is at least as good as Arc/INFO which they are
using here at the university.

--W

-- 

<:3 )---- Wolf Bergenheim ----( 8:>