The current version only sort 32 bits values.
Regards
You have to know that sorting on several kind of GPU and several kind of CPU is not an easy task and so, our goal is to provide, in this first phase, a set of stable algorithms.
In the second phase we will improve them to support more 'types'. But it request a lot of work.
We have choosed int32 because in most cases it is what is required even if some other cases exists, some peoples are interested by float, double or even structure (in this case we will propose other solution for performance purpose)
Krys
It is already on the Roadmap...
But you have to know that for now, nobody work on the project... so, we have to find someone to work on it before !