This is a speed comparison between Rasters and YAXArrays.
This uses the COSMO REA dataset.
It seems that YAXArrays has some overhead , but then it doesn't matter much whether it is the european or the restriction to the germany data.
I also tried to do this with a lazy Raster object, but then the computation didn't finish in many minutes. And when I do a subset of a lazy Raster the data got read at some point.
It seems that Raster is better suited for smaller arrays and then for larger Arrays the overhead of YAXArrays doesn't matter that much and it is becoming faster.
I would like to explore exactly where this tradeoff lays.
This is a speed comparison between Rasters and YAXArrays. This uses the COSMO REA dataset. It seems that YAXArrays has some overhead , but then it doesn't matter much whether it is the european or the restriction to the germany data. I also tried to do this with a lazy Raster object, but then the computation didn't finish in many minutes. And when I do a subset of a lazy Raster the data got read at some point. It seems that Raster is better suited for smaller arrays and then for larger Arrays the overhead of YAXArrays doesn't matter that much and it is becoming faster. I would like to explore exactly where this tradeoff lays.