CUDPP is the CUDA Data-Parallel Primitives Library. CUDPP is a library of data-parallel algorithm primitives such as parallel-prefix-sum (“scan”), parallel sort and parallel reduction. Primitives such as these are important building blocks for a wide variety of data-parallel algorithms, including sorting, stream compaction, and building data structures such as trees and summed-area tables. The first beta release of CUDPP is now available, as is the searchable online documentation.
This is a beta release, and as such, it is work in progress (and interfaces will change!). The library should already be very useful, but we have plans for much more, including fast segmented scan primitives and algorithms built upon them, such as quicksort and sparse matrix operations.
This is a joint project between The University of California, Davis and NVIDIA.