Performance of raster package
Thanks for pointing this out. Some functions in 'raster' are indeed slower/faster than others. Some seem to be reasonably quick. The focal* functions are slow, and indeed, as Rainer alludes to, good candidates for a C-implementation of at least the basic cases (e.g. fun=mean). Although I recently made them a bit more efficient, so it would be good to know which version of raster you are using. When reporting speed problems, please also report the number of rows/columns of the object and the exact function (focal, focalFilter, focalNA?) used. In the case of the focal functions it could matter a lot which "fun" you used to compute the focal values; as that might be what further slows things down. rasterToPolygons is also slow; in part because I did not pay much attention to it. I just made it a bit quicker (almost twice as fast), but much more could probably be done. That is, speed is not only a function of R vs. C code, but also of the quality of the algorithm used. The value of the 'raster' package over "GIS" software is obviously not speed of execution, but the ease (speed) of use and the benefit of the integration with everything else in R. Robert
Am 24.05.2011 10:17, schrieb Luca Morandini: Folks,
I have played a bit with the R "raster" package, and found it has an impressive list of features; however, its performance seems to be less than optimal when confronted with sizeable rasters (say, a 250MB GeoTIFF): a focal statistics that took less than a minute in GRASS, took 7 hours or so in R. Is there anything I could do to speed it up (I already tried readAll before invoking the focal statistic function) ? Regards, Luca Morandini http://www.lucamorandini.it
_______________________________________________ R-sig-Geo mailing list [hidden email] https://stat.ethz.ch/mailman/listinfo/r-sig-geo
Dear all, I discovered such performance problems concerning computation time, too. Using the function rasterToPolygons() for a 30Mb, 1Byte integer GeoTiff took 10 times longer than using a common GIS software. Why is that the case? I had to change to other R implementet GIS functionallity like in the package RSAGA. When posible reading data into memory will also accelerate processing time.
-- View this message in context: http://r-sig-geo.2731867.n2.nabble.com/Performance-of-raster-package-tp6397653p6399759.html Sent from the R-sig-geo mailing list archive at Nabble.com.