[Bug] r.clump G_realloc error

[laurapoggio-sptools]

Describe the bug

I am using r.clump on a quite large file.

I get the error below.

1-band clumping with threshold 0
Pass 1 of 2…
Current region rows: 220000, cols: 240000
ERROR: G_realloc: unable to allocate 18446744065119617024 bytes of memory      at raster/r.clump/clump.c:633

System description

operative system of the singularity/apptainer image: almalinux:9.6-20250520

• grass --version GRASS GIS 8.4.1
• gdalinfo --version GDAL 3.11.3 "Eganville", released 2025/07/12
• proj --version Rel. 9.6.2, June 4th, 2025
• geos-config --version 3.13.1
• python --version Python 3.13.5

Additional context

comment from here :

annakrat
This seems to come from the reallocating the clump index. My guess is there is too large number of clumps. The nalloc variable is integer so there may be integer overflow going on. I recommend creating a bug report.

[metzm]

Are you sure that the current region with

Current region rows: 220000, cols: 240000

is really set correctly? This is really huge. If I got the numbers right, the system tries to allocate about 18,446,744 TB !!! of RAM.

On the other hand, there is most likely an integer overflow occurring when calculating the number of clumps. The maximum number of supported clumps is the maximum number of integer 32 bit = max 2,147,483,647 clumps.

[laurapoggio-sptools]

I checked the region and it is huge. I now found a workaround (I hope, still computing).

Initially I was lookign to a flag or option to use the disk intead of the RAM (a bit like in r.watershed)

Thanks for the feedback

[metzm]

I checked the region and it is huge. I now found a workaround (I hope, still computing).

You could check if the raster is actually that huge and that the region is set correctly to the raster.

Initially I was lookign to a flag or option to use the disk intead of the RAM (a bit like in r.watershed)

The most important problem is not memory consumption but integer overflow because too many clumps have been identified. The limit of max 2,147,483,647 clumps can not be raised.

[petrasovaa]

The most important problem is not memory consumption but integer overflow because too many clumps have been identified. The limit of max 2,147,483,647 clumps can not be raised.

But I would check you actually need that many clumps, 220000 * 240000 / 2147483647 gives 25 cells per clump on average so it's possible, but maybe check this makes sense for your data.