In out project we read satellite image files to numpy array (1-D array, each cell contains a pair of coordinates and a band value). After that we load that data using multi threading in Python to multiple 1-D arrays in SciDB (so each thread loads to it’s own array);
When the loading is completed all data from temp arrays must be redimensioned (coordinates become array dimensions, so the result array is 2-D array) and saved in a resulting array, so for that purpose we use redimension() and store() functions. On this step it seems like on the step multi threading does not give any performance increase because of the locks in SciDB array.
So the question is, could you please suggest any other approach we can use to increase the loading and redimensioning performance. The source files are usually in GeoTiff format. Thank you.