Does anyone have guidelines of what the general purpose optimal image dimensions in pixels are for any given type? (3 band 8bit RGB, 1 band 16bit signed, ...)
At some point the time it takes to do anything in memory on a whole image exceeds the time it takes to read the next chunk from storage and repeat. I'm curious how to figure out what the point of diminishing returns is. The quest behind the question is determining a good default tile size for our image repository. There's a hundred variables that's going to make it a different answer for every organization, and that it will change over time as hardware and software changes. My hunch is there's a close enough approximation that works for most situations much of the time. Before I embark on series of trials to figure out what our local optimum is I figured I'd ask and see what other work in this area there might be. Your thoughts? Matt Wilkie Geomatics Developer & Administrator Environment | Technology, Innovation and Mapping T 867-667-8133 | Yukon.ca<http://yukon.ca/> Hours: 08:30-16:30, Mon-Wed: Office, Thu: Remote, Fri: Away.
_______________________________________________ gdal-dev mailing list gdal-dev@lists.osgeo.org https://lists.osgeo.org/mailman/listinfo/gdal-dev
