Is there a document anywhere specifying the best practices for parallel writes to a GDAL raster? I have an embarrasingly parallel problem that I would like to parallelise with MPI, but I'm not sure what I am allowed to do in parallel. I would like to assign blocks exclusively to worker threads to read and write concurrently, but I am unsure how I might do this safely.
Any suggestions would be greatly appreciated. Kind Regards, Francis Markham On 9 February 2011 09:14, Even Rouault <[email protected]> wrote: > Le mardi 08 février 2011 23:07:36, Stefano Moratto a écrit : > > Thanks for the suggestion, > > > > I have to read png from openstreetmap. I'm trying to render an A4 paper > > size sheet so I have a lot of data to move. > > It may be better to divide the area to be read in blocks and to give to > > each thread a block. Foreach block a thread should read all the bands. > > Probabilly I should allign the block to the tile's boundary > > Unfortunately the PNG format is such that it is not possible to seek to an > arbitrary line without decoding the data of all previous scanlines, so this > approach won't work. The same would be true for JPEG or GIF. If you can > fetch > the data as TIFF, you could however try that approach. > > > > > Stefano > > > > > Le mardi 08 février 2011 22:41:50, Stefano Moratto a écrit : > > > > Hello, > > > > > > > > given an opened GDAL dataset, is it possibile to perform > > > > > > > > gdalrasterio calls for each band in a dedicated thread per band in a > > > > thead-safe way? > > > > E.g. I have a raster image with 3 band (r,g,b). Since I have 4 core I > > > > > > would > > > > > > > like to I run a thread for a band r, one for band g and one for band > g. > > > > > > This has been discussed recently. See > > > http://lists.osgeo.org/pipermail/gdal- > > > dev/2011-January/027567.html and the following messages. > > > > > > (I'd note that if you process a pixel interleaved image, there are > > > optimizations in the GTiff driver to make the fetching of other bands > > > almost > > > 'free' if you read block by block, so there's probably little to gain, > > > but the > > > results of your experiments are welcome) > > > > > > > Regards, > > > > Stefano > _______________________________________________ > gdal-dev mailing list > [email protected] > http://lists.osgeo.org/mailman/listinfo/gdal-dev >
_______________________________________________ gdal-dev mailing list [email protected] http://lists.osgeo.org/mailman/listinfo/gdal-dev
