[GRASS-user] r.watershed crashing grass.

GRASS GRASS user
1

Dear all,

I am trying run a r.watershed command with one 7200x7200 DEM map.
But after one day running, the system crash without finish the output generation.
Please, see below (2 warnings before crash).
By the way I am running grass 6.4.0 SVN (rc4?) under Msys/Vista
on a 6Gb/64 bits machine.

Thanks for the help,

milton

GRASS 6.4.0svn (newLocation):c:/Users/famiglia > g.region -p
projection: 0 (x,y)
zone: 0
north: -23.99986111
south: -26.00013889
west: -49.00013889
east: -46.99986111
nsres: 0.00027778
ewres: 0.00027778
rows: 7201
cols: 7201
cells: 51854401

GRASS 6.4.0svn > g.region rast=FabioASTGTM_S2526W04948_dem
GRASS 6.4.0svn > r.watershed.exe -m 4096 elevation=FabioASTGTM_S2526W04948_dem stream=FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100 threshold=100 --o
WARNING: Subprocess failed with exit code 255
WARNING: category information for
[FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100] in
[PERMANENT] missing or invalid

2

Do you have enough hard disk space? I have had a problem running
r.watershed on very large datasets (my computer is much more modest).
I have had luck resampling the data to a larger resolution and then
running r.watershed. r.watershed crashes at 10m res but will run just
fine at 50m. 1GB RAM.

Stephen Sefick

On Fri, Jul 24, 2009 at 5:57 PM, Milton Cezar
Ribeiro<miltinho.astronauta@gmail.com> wrote:

Dear all,

I am trying run a r.watershed command with one 7200x7200 DEM map.
But after one day running, the system crash without finish the output
generation.
Please, see below (2 warnings before crash).
By the way I am running grass 6.4.0 SVN (rc4?) under Msys/Vista
on a 6Gb/64 bits machine.

Thanks for the help,

milton

GRASS 6.4.0svn (newLocation):c:/Users/famiglia > g.region -p
projection: 0 (x,y)
zone: 0
north: -23.99986111
south: -26.00013889
west: -49.00013889
east: -46.99986111
nsres: 0.00027778
ewres: 0.00027778
rows: 7201
cols: 7201
cells: 51854401

GRASS 6.4.0svn > g.region rast=FabioASTGTM_S2526W04948_dem
GRASS 6.4.0svn > r.watershed.exe -m 4096
elevation=FabioASTGTM_S2526W04948_dem
stream=FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100 threshold=100
--o
WARNING: Subprocess failed with exit code 255
WARNING: category information for
[FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100] in
[PERMANENT] missing or invalid

_______________________________________________
grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user

--
Stephen Sefick

Let's not spend our time and resources thinking about things that are
so little or so large that all they really do for us is puff us up and
make us feel like gods. We are mammals, and have not exhausted the
annoying little problems of being mammals.

                -K. Mullis

3

Hi Stephen,

Thanks for your reply.
I thank the disk space is not problem because I still have 100Gb free.
What was the dimension (row and columns) that you got success?!
May be 7200x7200 is to large. (?)

Cheers

milton

2009/7/24 stephen sefick <ssefick@gmail.com>

Do you have enough hard disk space? I have had a problem running
r.watershed on very large datasets (my computer is much more modest).
I have had luck resampling the data to a larger resolution and then
running r.watershed. r.watershed crashes at 10m res but will run just
fine at 50m. 1GB RAM.

Stephen Sefick

On Fri, Jul 24, 2009 at 5:57 PM, Milton Cezar
Ribeiro<miltinho.astronauta@gmail.com> wrote:

Dear all,

I am trying run a r.watershed command with one 7200x7200 DEM map.
But after one day running, the system crash without finish the output
generation.
Please, see below (2 warnings before crash).
By the way I am running grass 6.4.0 SVN (rc4?) under Msys/Vista
on a 6Gb/64 bits machine.

Thanks for the help,

milton

GRASS 6.4.0svn (newLocation):c:/Users/famiglia > g.region -p
projection: 0 (x,y)
zone: 0
north: -23.99986111
south: -26.00013889
west: -49.00013889
east: -46.99986111
nsres: 0.00027778
ewres: 0.00027778
rows: 7201
cols: 7201
cells: 51854401

GRASS 6.4.0svn > g.region rast=FabioASTGTM_S2526W04948_dem
GRASS 6.4.0svn > r.watershed.exe -m 4096
elevation=FabioASTGTM_S2526W04948_dem
stream=FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100 threshold=100
–o
WARNING: Subprocess failed with exit code 255
WARNING: category information for
[FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100] in
[PERMANENT] missing or invalid

grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user

–
Stephen Sefick

Let’s not spend our time and resources thinking about things that are
so little or so large that all they really do for us is puff us up and
make us feel like gods. We are mammals, and have not exhausted the
annoying little problems of being mammals.

-K. Mullis

4

I will look tomorrow, but I think it was about 1000x1000.
hth

Stephen Sefick

On Fri, Jul 24, 2009 at 9:47 PM, Milton Cezar
Ribeiro<miltinho.astronauta@gmail.com> wrote:

Hi Stephen,

Thanks for your reply.
I thank the disk space is not problem because I still have 100Gb free.
What was the dimension (row and columns) that you got success?!
May be 7200x7200 is to large. (?)

Cheers

milton

2009/7/24 stephen sefick <ssefick@gmail.com>

Do you have enough hard disk space? I have had a problem running
r.watershed on very large datasets (my computer is much more modest).
I have had luck resampling the data to a larger resolution and then
running r.watershed. r.watershed crashes at 10m res but will run just
fine at 50m. 1GB RAM.

Stephen Sefick

On Fri, Jul 24, 2009 at 5:57 PM, Milton Cezar
Ribeiro<miltinho.astronauta@gmail.com> wrote:
> Dear all,
>
> I am trying run a r.watershed command with one 7200x7200 DEM map.
> But after one day running, the system crash without finish the output
> generation.
> Please, see below (2 warnings before crash).
> By the way I am running grass 6.4.0 SVN (rc4?) under Msys/Vista
> on a 6Gb/64 bits machine.
>
> Thanks for the help,
>
> milton
> ===
> GRASS 6.4.0svn (newLocation):c:/Users/famiglia > g.region -p
> projection: 0 (x,y)
> zone: 0
> north: -23.99986111
> south: -26.00013889
> west: -49.00013889
> east: -46.99986111
> nsres: 0.00027778
> ewres: 0.00027778
> rows: 7201
> cols: 7201
> cells: 51854401
>
>
> GRASS 6.4.0svn > g.region rast=FabioASTGTM_S2526W04948_dem
> GRASS 6.4.0svn > r.watershed.exe -m 4096
> elevation=FabioASTGTM_S2526W04948_dem
> stream=FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100
> threshold=100
> --o
> WARNING: Subprocess failed with exit code 255
> WARNING: category information for
> [FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100] in
> [PERMANENT] missing or invalid
>
>
> _______________________________________________
> grass-user mailing list
> grass-user@lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/grass-user
>
>

--
Stephen Sefick

Let's not spend our time and resources thinking about things that are
so little or so large that all they really do for us is puff us up and
make us feel like gods. We are mammals, and have not exhausted the
annoying little problems of being mammals.

                                                           \-K\. Mullis

--
Stephen Sefick

Let's not spend our time and resources thinking about things that are
so little or so large that all they really do for us is puff us up and
make us feel like gods. We are mammals, and have not exhausted the
annoying little problems of being mammals.

                -K. Mullis

5

Milton Cezar Ribeiro pisze:

Dear all,
I am trying run a r.watershed command with one 7200x7200 DEM map.
But after one day running, the system crash without finish the output generation.
Please, see below (2 warnings before crash).

little strange, I run r.watershed (but on ubuntu not windows) with raster file 4500 x5000 and it takes no more than 2 minutes.
Tell us more about grass version you use (probably you use old version of r.watershed)
Jarek

By the way I am running grass 6.4.0 SVN (rc4?) under Msys/Vista
on a 6Gb/64 bits machine.
Thanks for the help,
milton

GRASS 6.4.0svn (newLocation):c:/Users/famiglia > g.region -p
projection: 0 (x,y)
zone: 0
north: -23.99986111
south: -26.00013889
west: -49.00013889
east: -46.99986111
nsres: 0.00027778
ewres: 0.00027778
rows: 7201
cols: 7201
cells: 51854401

GRASS 6.4.0svn > r.watershed.exe -m 4096 elevation=FabioASTGTM_S2526W04948_dem stream=FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100 threshold=100 --o
WARNING: Subprocess failed with exit code 255
WARNING: category information for
         [FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100] in
         [PERMANENT] missing or invalid

_______________________________________________
grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user
  

6

Hi Jarek,

I downloaded and compiled grass 6.4 SVN last week. So I suppose my version is not so old.
Is there a way of I know the version of a r.(command)?

May be some issue with Vista (again!) ?

Thanks for the reply

milton

2009/7/25 Jarek Jasiewicz <jarekj@amu.edu.pl>

Milton Cezar Ribeiro pisze:

Dear all,
I am trying run a r.watershed command with one 7200x7200 DEM map.
But after one day running, the system crash without finish the output generation.
Please, see below (2 warnings before crash).

little strange, I run r.watershed (but on ubuntu not windows) with raster file 4500 x5000 and it takes no more than 2 minutes.
Tell us more about grass version you use (probably you use old version of r.watershed)
Jarek

By the way I am running grass 6.4.0 SVN (rc4?) under Msys/Vista
on a 6Gb/64 bits machine.
Thanks for the help,
milton

GRASS 6.4.0svn (newLocation):c:/Users/famiglia > g.region -p
projection: 0 (x,y)
zone: 0
north: -23.99986111
south: -26.00013889
west: -49.00013889
east: -46.99986111
nsres: 0.00027778
ewres: 0.00027778
rows: 7201
cols: 7201
cells: 51854401
GRASS 6.4.0svn > g.region rast=FabioASTGTM_S2526W04948_dem
GRASS 6.4.0svn > r.watershed.exe -m 4096 elevation=FabioASTGTM_S2526W04948_dem stream=FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100 threshold=100 --o
WARNING: Subprocess failed with exit code 255
WARNING: category information for
[FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100] in
[PERMANENT] missing or invalid

grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user

7

Try to run it all in memory:
r.watershed.exe elevation=FabioASTGTM_S2526W04948_dem stream=FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100 threshold=100 --o

Should finish in a few minutes.

Where is this 4096 coming from in the original command? Should that be memory=4096?
r.watershed.exe -m memory=4096 elevation=FabioASTGTM_S2526W04948_dem stream=FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100 threshold=100 --o

How far did it get? There are usually message like SECTION 1:, SECTION 2:, SECTION 3: etc.

Markus M

Milton Cezar Ribeiro wrote:

Dear all,
I am trying run a r.watershed command with one 7200x7200 DEM map.
But after one day running, the system crash without finish the output generation.
Please, see below (2 warnings before crash).
By the way I am running grass 6.4.0 SVN (rc4?) under Msys/Vista
on a 6Gb/64 bits machine.
Thanks for the help,
milton

GRASS 6.4.0svn (newLocation):c:/Users/famiglia > g.region -p
projection: 0 (x,y)
zone: 0
north: -23.99986111
south: -26.00013889
west: -49.00013889
east: -46.99986111
nsres: 0.00027778
ewres: 0.00027778
rows: 7201
cols: 7201
cells: 51854401

GRASS 6.4.0svn > r.watershed.exe -m 4096 elevation=FabioASTGTM_S2526W04948_dem stream=FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100 threshold=100 --o
WARNING: Subprocess failed with exit code 255
WARNING: category information for
         [FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100] in
         [PERMANENT] missing or invalid

8

With 7201 rows x 7201 columns, the in-memory version should use at most 1.6GB of RAM. There should be no problem running it all in memory (no -m 4096) on a machine with 6GB.

Markus M

Milton Cezar Ribeiro wrote:

Dear all,
I am trying run a r.watershed command with one 7200x7200 DEM map.
But after one day running, the system crash without finish the output generation.
Please, see below (2 warnings before crash).
By the way I am running grass 6.4.0 SVN (rc4?) under Msys/Vista
on a 6Gb/64 bits machine.
Thanks for the help,
milton

GRASS 6.4.0svn (newLocation):c:/Users/famiglia > g.region -p
projection: 0 (x,y)
zone: 0
north: -23.99986111
south: -26.00013889
west: -49.00013889
east: -46.99986111
nsres: 0.00027778
ewres: 0.00027778
rows: 7201
cols: 7201
cells: 51854401

GRASS 6.4.0svn > r.watershed.exe -m 4096 elevation=FabioASTGTM_S2526W04948_dem stream=FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100 threshold=100 --o
WARNING: Subprocess failed with exit code 255
WARNING: category information for
         [FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100] in
         [PERMANENT] missing or invalid

_______________________________________________
grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user
  

9

Hi Markus,

Thanks for the tips. In fact I made a mistake, because I thought that with “-m 4096” I was setting 4GB of memory, and unfortunatelly I not payed attention on memory=4096. Anyway, I tryed with both your suggestions (with and without -m memory=4096" and still ger error:

GRASS 6.4.0svn (newLocation):c:/Users/famiglia > r.watershed.exe -m memory=4096 elevation=FabioASTGTM_S2526W049
48_dem stream=FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100 threshold=100 --o --v
SECTION 1 beginning: Initiating Variables. 5 sections total.
ERROR: G_malloc: unable to allocate 800000 bytes at setup.c:64
WARNING: Subprocess failed with exit code 1
WARNING: category information for
[FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100] in
[PERMANENT] missing or invalid

Any other suggestions?

milton

2009/7/25 Markus Metz <markus.metz.giswork@googlemail.com>

Try to run it all in memory:
r.watershed.exe elevation=FabioASTGTM_S2526W04948_dem stream=FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100 threshold=100 --o

Should finish in a few minutes.

Where is this 4096 coming from in the original command? Should that be memory=4096?
r.watershed.exe -m memory=4096 elevation=FabioASTGTM_S2526W04948_dem stream=FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100 threshold=100 --o

How far did it get? There are usually message like SECTION 1:, SECTION 2:, SECTION 3: etc.

Markus M

Milton Cezar Ribeiro wrote:

Dear all,
I am trying run a r.watershed command with one 7200x7200 DEM map.
But after one day running, the system crash without finish the output generation.
Please, see below (2 warnings before crash).
By the way I am running grass 6.4.0 SVN (rc4?) under Msys/Vista
on a 6Gb/64 bits machine.
Thanks for the help,
milton

GRASS 6.4.0svn (newLocation):c:/Users/famiglia > g.region -p
projection: 0 (x,y)
zone: 0
north: -23.99986111
south: -26.00013889
west: -49.00013889
east: -46.99986111
nsres: 0.00027778
ewres: 0.00027778
rows: 7201
cols: 7201
cells: 51854401
GRASS 6.4.0svn > g.region rast=FabioASTGTM_S2526W04948_dem
GRASS 6.4.0svn > r.watershed.exe -m 4096 elevation=FabioASTGTM_S2526W04948_dem stream=FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100 threshold=100 --o
WARNING: Subprocess failed with exit code 255
WARNING: category information for
[FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100] in
[PERMANENT] missing or invalid

10

Milton Cezar Ribeiro wrote:

Hi Markus,
Thanks for the tips. In fact I made a mistake, because I thought that with "-m 4096" I was setting 4GB of memory, and unfortunatelly I not payed attention on memory=4096. Anyway, I tryed with both your suggestions (with and without -m memory=4096" and still ger error:
-----------
GRASS 6.4.0svn (newLocation):c:/Users/famiglia > r.watershed.exe -m memory=4096 elevation=FabioASTGTM_S2526W049
48_dem stream=FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100 threshold=100 --o --v
SECTION 1 beginning: Initiating Variables. 5 sections total.
ERROR: G_malloc: unable to allocate 800000 bytes at setup.c:64

Very very strange. Not even 781 KB of memory left? I don't know Vista very well, but I guess there is some sort of task manager where you can watch memory consumption. What does it say, is there still enough free memory available? Anyway, r.watershed should go into swap space (virtual memory) if there is not enough RAM free.

Can you do any other operations on the DEM and the given region, e.g. resampling or neighborhood filter or something like that?

Maybe there is a problem with your compilation of grass6.4.0.svn? Is everything compiled with msys? There were quite a few posts in the list about vista problems, maybe you find some hints there. I'm only using Linux, can't help there.

Markus

WARNING: Subprocess failed with exit code 1
WARNING: category information for
         [FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100] in
         [PERMANENT] missing or invalid
---------
Any other suggestions?
milton
2009/7/25 Markus Metz <markus.metz.giswork@googlemail.com <mailto:markus.metz.giswork@googlemail.com>>

    Try to run it all in memory:
    r.watershed.exe elevation=FabioASTGTM_S2526W04948_dem
    stream=FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100
    threshold=100 --o

    Should finish in a few minutes.

    Where is this 4096 coming from in the original command? Should
    that be memory=4096?
    r.watershed.exe -m memory=4096
    elevation=FabioASTGTM_S2526W04948_dem
    stream=FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100
    threshold=100 --o

    How far did it get? There are usually message like SECTION 1:,
    SECTION 2:, SECTION 3: etc.

    Markus M

    Milton Cezar Ribeiro wrote:

        Dear all,
         I am trying run a r.watershed command with one 7200x7200 DEM map.
        But after one day running, the system crash without finish the
        output generation.
        Please, see below (2 warnings before crash).
        By the way I am running grass 6.4.0 SVN (rc4?) under Msys/Vista
        on a 6Gb/64 bits machine.
         Thanks for the help,
         milton
        ===
        GRASS 6.4.0svn (newLocation):c:/Users/famiglia > g.region -p
        projection: 0 (x,y)
        zone: 0
        north: -23.99986111
        south: -26.00013889
        west: -49.00013889
        east: -46.99986111
        nsres: 0.00027778
        ewres: 0.00027778
        rows: 7201
        cols: 7201
        cells: 51854401
          GRASS 6.4.0svn > g.region rast=FabioASTGTM_S2526W04948_dem
        GRASS 6.4.0svn > r.watershed.exe -m 4096
        elevation=FabioASTGTM_S2526W04948_dem
        stream=FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100
        threshold=100 --o
         WARNING: Subprocess failed with exit code 255
        WARNING: category information for
                       [FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100] in
                [PERMANENT] missing or invalid

11

Markus Metz wrote:

There were quite a few posts in the list about vista problems,
maybe you find some hints there. I'm only using Linux, can't
help there.

IIRC it was more generic memory problems which only showed up on
MS Windows. In those cases valgrind reported errors, just on
Linux they end up being benign.

e.g. r.los (apparently fixed by Ben?), v.out.ogr not working in
current WinGrass (!!!), r.sim.water, details + maybe more
modules listed in the bug tracker (search on the wingrass
keyword).

Other wingrass problems seem to be mostly packaging issues.

Hamish

12

Hi Markus,

My system start with 1.5GB of memory usage, considering that
grass are loaded on memory, and my mapset is also loaded.
When I start the r.watershed, my system go up to 3.0GB, remain
running for few minutes, and then give the error that I reported,
and them release the memory. On several trying the system
goes only up to 3GB, but I have 6Gb. So my “-m memory=4096”
not make sense. In fact 3-1.5 = ~ 1.5GB as you suggested.

Any other idea?

cheers

milton

2009/7/25 Markus Metz <markus.metz.giswork@googlemail.com>

Milton Cezar Ribeiro wrote:

Hi Markus,
Thanks for the tips. In fact I made a mistake, because I thought that with “-m 4096” I was setting 4GB of memory, and unfortunatelly I not payed attention on memory=4096. Anyway, I tryed with both your suggestions (with and without -m memory=4096" and still ger error:

GRASS 6.4.0svn (newLocation):c:/Users/famiglia > r.watershed.exe -m memory=4096 elevation=FabioASTGTM_S2526W049
48_dem stream=FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100 threshold=100 --o --v
SECTION 1 beginning: Initiating Variables. 5 sections total.
ERROR: G_malloc: unable to allocate 800000 bytes at setup.c:64

Very very strange. Not even 781 KB of memory left? I don’t know Vista very well, but I guess there is some sort of task manager where you can watch memory consumption. What does it say, is there still enough free memory available? Anyway, r.watershed should go into swap space (virtual memory) if there is not enough RAM free.

Can you do any other operations on the DEM and the given region, e.g. resampling or neighborhood filter or something like that?

Maybe there is a problem with your compilation of grass6.4.0.svn? Is everything compiled with msys? There were quite a few posts in the list about vista problems, maybe you find some hints there. I’m only using Linux, can’t help there.

Markus

WARNING: Subprocess failed with exit code 1
WARNING: category information for
[FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100] in
[PERMANENT] missing or invalid

Any other suggestions?
milton

2009/7/25 Markus Metz <markus.metz.giswork@googlemail.com mailto:[markus.metz.giswork@googlemail.com](mailto:markus.metz.giswork@googlemail.com)>

Try to run it all in memory:
r.watershed.exe elevation=FabioASTGTM_S2526W04948_dem
stream=FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100
threshold=100 --o

Should finish in a few minutes.

Where is this 4096 coming from in the original command? Should
that be memory=4096?
r.watershed.exe -m memory=4096
elevation=FabioASTGTM_S2526W04948_dem
stream=FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100
threshold=100 --o

How far did it get? There are usually message like SECTION 1:,
SECTION 2:, SECTION 3: etc.

Markus M

Milton Cezar Ribeiro wrote:

Dear all,
I am trying run a r.watershed command with one 7200x7200 DEM map.
But after one day running, the system crash without finish the
output generation.
Please, see below (2 warnings before crash).
By the way I am running grass 6.4.0 SVN (rc4?) under Msys/Vista
on a 6Gb/64 bits machine.
Thanks for the help,
milton

GRASS 6.4.0svn (newLocation):c:/Users/famiglia > g.region -p
projection: 0 (x,y)
zone: 0
north: -23.99986111
south: -26.00013889
west: -49.00013889
east: -46.99986111
nsres: 0.00027778
ewres: 0.00027778
rows: 7201
cols: 7201
cells: 51854401
GRASS 6.4.0svn > g.region rast=FabioASTGTM_S2526W04948_dem
GRASS 6.4.0svn > r.watershed.exe -m 4096
elevation=FabioASTGTM_S2526W04948_dem
stream=FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100
threshold=100 --o
WARNING: Subprocess failed with exit code 255
WARNING: category information for
[FabioASTGTM_S2526W04948_dem_rwatershed_stream_thres0100] in
[PERMANENT] missing or invalid

13

Milton Cezar Ribeiro wrote:

Hi Markus,
My system start with 1.5GB of memory usage,

vista waste

considering that
grass are loaded on memory,

not much

and my mapset is also loaded.

not loaded to memory

When I start the r.watershed, my system go up to 3.0GB, remain
running for few minutes, and then give the error that I reported,
and them release the memory. On several trying the system
goes only up to 3GB, but I have 6Gb.

is grass compiled as 32bit or 64bit? 32bit applications can't use all 6GB of memory, only up to about 3GB.

So my "-m memory=4096"
not make sense. In fact 3-1.5 = ~ 1.5GB as you suggested.
Any other idea?

Not really, only diagnostics: the G_malloc error was caused by the segment library, not r.watershed directly.

Hamish said this is a generic memory problem on MS windows (vista).
I would 1) close grass, 2) reboot windows, 3) make distclean, configure, make, 4) check if it's now working, 5) if not check if a similar bug has been already reported, file a new bug if necessary, give up and have a nice weekend, IOW I'm out of ideas, sorry!

Markus

14

Milton Cezar Ribeiro wrote:

> On several trying the system goes only up to 3GB, but I have 6Gb.

Markus Metz wrote:

is grass compiled as 32bit or 64bit? 32bit applications can't use all
6GB of memory, only up to about 3GB.

Milton,

you mention you have 6GB of RAM. I assume that you know a 64-bit
Operating System (besides the applications being 64-bit) is required to
break the 3GB RAM limit.

* If you bought the machine with 6GB already equipped then they _must_
have installed the 64-bit version of sVista [*]. Then follow Markus'
suggestions.

* If you did a RAM upgrade by yourself (e.g. from 2GB to 6GB) then did
you check that you have a 64-bit sVista?

Nikos
---

[*] svista means in greek "erase them" ;-p

15

Hamish said this is a generic memory problem on MS windows
(vista).

eh? my point was mearly that there are some existing memory bugs
which haven't been fixed yet as they only present themselves when
run under MS Windows, and we are just starting to discover those
as more people try native WinGrass.

I've got no idea if this problem belongs to that class or not,
nor if XP vs. Vista makes any difference to GRASS. Just that
Valgrind is a nice tool for helping to deal to bugs stemming
from memory code mistakes, or clearing that as a suspected
cause.

Hamish

16

Hamish wrote:

Hamish said this is a generic memory problem on MS windows
(vista).
    
eh? my point was mearly that there are some existing memory bugs
which haven't been fixed yet as they only present themselves when
run under MS Windows, and we are just starting to discover those
as more people try native WinGrass.

I've got no idea if this problem belongs to that class or not,
nor if XP vs. Vista makes any difference to GRASS. Just that
Valgrind is a nice tool for helping to deal to bugs stemming
from memory code mistakes, or clearing that as a suspected
cause.
  

"Hamish suggested that this may be a generic problem on MS windows (vista)" might have been better phrasing. Sorry for misinterpreting your words!

In this particular case, it may be a problem of a 32bit application running on a 64bit system with more than 3GB RAM, causing rather ugly out-of-memory errors because the 32bit application requesting more memory can't handle the memory returned by the OS. Just a wild guess, maybe a windows developer can say more about it.

Markus

17

Hi all,

In fact I have a dell machine, with Vista 64bits that came with the machine.
I compiled grass 64 svn using Msys. Is there a way of I know if my grass are running with 64bits? By the way, what I need to change to have sure that grass was compiled to 64bits?

Yesterday I preloaded about 2.5 Gb on memory, and started the r.watershed again. I monitored the memory usage on Task Manager, and I perfeived that memory usage gone ~4Gb. So it make me think that 2.5+(~)1.6 (suggested by Markus?)= ~4Gb. So my system are going behind 3Gbs without problem. But I am not sure grass do.

Thanks for your time

milton

2009/7/25 Nikos Alexandris <nikos.alexandris@felis.uni-freiburg.de>

Milton Cezar Ribeiro wrote:

On several trying the system goes only up to 3GB, but I have 6Gb.

Markus Metz wrote:

is grass compiled as 32bit or 64bit? 32bit applications can’t use all
6GB of memory, only up to about 3GB.

Milton,

you mention you have 6GB of RAM. I assume that you know a 64-bit
Operating System (besides the applications being 64-bit) is required to
break the 3GB RAM limit.

  • If you bought the machine with 6GB already equipped then they must
    have installed the 64-bit version of sVista [*]. Then follow Markus’
    suggestions.

  • If you did a RAM upgrade by yourself (e.g. from 2GB to 6GB) then did
    you check that you have a 64-bit sVista?

Nikos

[*] svista means in greek “erase them” ;-p

18

On Sat, Jul 25, 2009 at 5:45 PM, Milton Cezar
Ribeiro<miltinho.astronauta@gmail.com> wrote:

Hi all,

In fact I have a dell machine, with Vista 64bits that came with the machine.
I compiled grass 64 svn using Msys. Is there a way of I know if my grass are
running with 64bits? By the way, what I need to change to have sure that
grass was compiled to 64bits?

Did you enable it?
--enable-64bit
for configure...

Markus

19

Markus Neteler wrote:

> In fact I have a dell machine, with Vista 64bits that came with the machine.
> I compiled grass 64 svn using Msys. Is there a way of I know if my grass are
> running with 64bits? By the way, what I need to change to have sure that
> grass was compiled to 64bits?

Did you enable it?
--enable-64bit
for configure...

That switch doesn't do anything on Windows. Or on Linux, for that
matter. It is only used on (some versions of) AIX, HP/UX, Irix, and
Solaris.

If you want 64-bit Windows executables, you would need 64-bit versions
of gcc and binutils.

--
Glynn Clements <glynn@gclements.plus.com>

20

Hi Glynn

Just suppose I start Msys. Is there a way of i know if (1) my compiled version of grass support 64bits? or (2) if my gcc etc support 64bits?

Thanks a million…

milton

2009/7/26 Glynn Clements <glynn@gclements.plus.com>

Markus Neteler wrote:

In fact I have a dell machine, with Vista 64bits that came with the machine.
I compiled grass 64 svn using Msys. Is there a way of I know if my grass are
running with 64bits? By the way, what I need to change to have sure that
grass was compiled to 64bits?

Did you enable it?
–enable-64bit
for configure…

That switch doesn’t do anything on Windows. Or on Linux, for that
matter. It is only used on (some versions of) AIX, HP/UX, Irix, and
Solaris.

If you want 64-bit Windows executables, you would need 64-bit versions
of gcc and binutils.

–
Glynn Clements <glynn@gclements.plus.com>