GPUlib 1.4 is now available! See the GPULib blog for a rundown of the new features. Particularly exciting for me are:

  1. new API using operator overloading for GPU variables (if you have IDL 8.0)
  2. support for streams so memory can be transferred between GPU and CPU while executing kernels
  3. sort and histogram routines
  4. a variety of memory leaks and other bugs have been fixed

Full disclosure: I work for Tech-X Corporation and have worked on the IDL bindings for GPULib.