Plz help me.I want to program an aplication in Cuda to sort arrays.The idea is to compare GPU vs CPU performance toward array sorting.Any ideas which sorting algorithms to use?

The simplest way is to use the Thrust library, see

There are also two new frameworks on NVLabs:

Both have interesting sorting implementations.

Thank you!!I really appreciate your answers…but I’m looking not for the simplest way, I want to program sorting in parallel using CUDA C and then comparing the perfomance of CPU versus GPU.Please any new ideas?

What is your target? Learning or performance?
In the latter case, with respect, it would be difficult to outperform the libraries that have been suggested :-)