Task #9652 (new)
Improve performance of HDF-based readers
|Reported by:||mlinkert||Owned by:|
See for example imaris/jake/*.
This is a very large file containing a pyramid of big images. It takes on the order of 40 hours to import, as 256x256 tiles are read at a rate of 4 per second.
The bottleneck appears to be in netCDF, though it may help some to adjust the optimal tile sizes to match the HDF chunk sizes.