1. A very fast hash implementation on GPUs from
"Real-Time Parallel Hashing on the GPU" is planned to getting merged sometime in January.
2. Added reductions cudppReduce().
3. Double precision support coming at least now has double support for (2)
Added CUDPP_DOUBLE type enum, since cudppReduce() supports double precision.4. CuDPP autotuning interface for plans added
http://code.google.com/p/cudpp/wiki/cudppTuneDesignNotes
Follow questions and progress in:
http://groups.google.com/group/cudpp
http://code.google.com/p/cudpp/updates/list
0 comments:
Post a Comment