cuda research papers

Question

i am currently doing my BS in computer science and i am interested in graduate studies. i realize that most universities ask for student research experience and publications. i am very interested in cuda programming. so my question is: how can i write papers about cuda. i searched a lot on Google and did not find a lot of research papers about the topic (or found topics that i am not interested in). so will the following leads to a publication: if i wrote a sequential solution for an NP-Complete problem, and i wrote its equivalent cuda code, and showed that after testing the cuda program runs much faster than the sequential algorithm. Will this leads to a publication

Answer 1

You shouldn't need publications to get into grad school, and once you're there, you shouldn't write papers just to waste paper ;v) .

ALWAYS solve problems because they are interesting and publish because others would benefit from your work.

CUDA is just a toolchain and machine architecture to help run parallel code quickly. See the help-wanted section of NVidia's forums for ideas about who is doing what with CUDA. Maybe you'll even find a job which could help you get into grad school later if the first round of applications doesn't work.

Answer 2

well the most important requirement for graduate school is the number of publications you already have. because this proves how good researcher you are.

Do NOT publish papers just to get to graduate school, publish because they are interesting and others would benefit from your work. "as Potatoswatter said".

and yes this is what you should do to publish. find an interesting NP-Complete problem, check some exact algorithms for it, write its sequential and parallel (cuda) code, compare results, and publish your paper if you think that you got good results.