Hi,
just for your information:
I have created an OpenMP enabled version of r.series to parallelize the inner processing loop. The speedup is about 10%-15% for really large datasets (thousands of maps, each map millions of cells).
The code is available here as tar.gz:
http://www-pool.math.tu-berlin.de/~soeren/grass/
However, i will not commit the “improvement” to svn, since the speedup is so low and only relevant for huge datasets.
Best regards
Soeren