Parallel Programming @APOGEE 2014

[Saiyedul Islam]

Dear all,

I hope you all are doing great !

As you know that project presentation format for APOGEE has changed and we are already one month into the semester, I think this is the right time to start doing some things "in parallel".

For the last two years we were focussed on CUDA only but this time I intend to introduce some other forms of parallel programming like multicore and/or cluster computing. So, there won't be any more CUDA classes, though if there are enough number of interested students, I may take few introductory classes on OpenMP and/or MPI.

One of the easiest way to exploit multicore processor for parallel computing is OpenMP (google few sample codes). For cluster of systems connected over LAN, we normally use Message Passing Interface (MPI). Both these technologies are widely used in High Performance Computing industry.

As far as projects are concerned, you should start searching for "little less explored compute intensive problems" and then discuss how to parallelize it using CUDA, OpenMP, MPI or combination of these.

I request anyone who has done any kind of parallel computing in previous APOGEE or in his/her COP/SOP/RP/PS or just for fun, to come forward and share his/her experience/expertise with us.

Anyone who wants to discuss anything (absolutely anything) related to parallel computing is welcome to reply here or mail me or meet me in Lab 1103 in FD I :-)

Any suggestions for doing something more in this SIG are also most welcome.

Regards,

Saiyedul Islam

PhD Scholar,

BITS-Pilani

(saiyedu...@gmail.com , sis...@pilani.bits-pilani.ac.in)