Progress

Well, I found some ( read : very very few ) implementations of bucket sort. Two of them were OpenMP based.
1) http://link.ece.uci.edu/~weiran/documents/eecs221_w08.pdf

2) http://www.moon.sannet.ne.jp/okahisa/sort/node32.html

3) http://snippets.dzone.com/posts/show/4365

The Third One:

It uses a library called MCSTL. It can be downloaded from here. It's corresponding research paper can be downloaded from here. Interestingly, my version of Ubuntu, viz. 10.10 or Maverick, does not have support for MCSTL. Even if I downloaded MCSTL and copy pasted the .h files at my "include" folder ( /usr/include ), i did not work somehow. Apparently, there is no way to install MCSTL, so I had to do this ugly copy-paste. Bottom-line, it did not work. Later I found that MCSTL requires g++-4.2.2 which Maverick doesn't support, so, conclusion is, forget MCSTL.

Fortunately, I found an interesting piece of information at the STXXL library page. It says STXXL internally may use MCSTL or libstdc++ parallel mode, where the later being a successor of MCSTL.

So, that's what I need. I need the libstdc++ parallel mode library. This page is a decent piece of documentation about that. Being part of Libstdc++, it has been very well documented at many places.

EDIT: Got some documentation for mcstl.

Other interesting Pages:

Parallel In-Place Radix Sort Simplified

32 OpenMP Traps For C++ Developers