Accelerating batch processing of spatial raster analysis using GPU
Batch processing of raster data performed by geographic information systems (GIS) is a time consuming procedure. Modern high performance GPUs are able to perform hundreds of arithmetical operations in parallel. These GPUs can help to reduce the computing time of such operations. In addition, most of the commonly used raster operations are I/O-bounded. Memory transfer between hard disk and RAM takes up more time than computations. The scope of this paper is to present an efficient two-level caching strategy for raster data and an acceleration of selected raster operations using the GPU, which were implemented as a plugin for the open source software GRASS. An example data flow based on a real world use-case will be presented and the obtainable and practically expectable speedup will be measured and discussed. âº Accelerating raster operations using GPU. âº Accelerating batch processing of spatial raster analysis by caching. âº Implementations for the open source GIS GRASS. âº Speed comparison between standard GRASS raster operation and our accelerated ones.