I am currently parallelizing raster algorithms of grass using OpenMP and I am supposed to evaluate their performance compared to original algorithm. Is there any standard way to do this?
Hi Shreyas,
That’s a great effort, would you be able to share the progress with us? Usually, it’s good if multiple people test it on different hardware, see for example a previous parallelization contribution [1]. To measure the performance, there is a benchmarking library [2] you can use. There are benchmarking scripts in the source code for some tools, see for example [3] or see a benchmarking example in a jupyter notebook [4].
Anna
[1] https://github.com/OSGeo/grass/pull/1782
[2] https://grass.osgeo.org/grass83/manuals/libpython/grass.benchmark.html
[3] https://github.com/OSGeo/grass/tree/main/raster/r.neighbors/benchmark
[4] https://github.com/ncsu-geoforall-lab/grass-workshop-gis-week-2023/blob/main/01_intro_to_GRASS_parallelization.ipynb
On Tue, Dec 5, 2023 at 3:28 AM Shreyas Udaya via grass-dev <grass-dev@lists.osgeo.org> wrote:
I am currently parallelizing raster algorithms of grass using OpenMP and I am supposed to evaluate their performance compared to original algorithm. Is there any standard way to do this?
grass-dev mailing list
grass-dev@lists.osgeo.org
https://lists.osgeo.org/mailman/listinfo/grass-dev