i'm a graduating CS student at University of Bologna (Italy) and i'm currently working as software developer at INFN (Nuclear Phisics National Institute) in Bologna. I'm involved in a research project about scientific computation through hardware acceleration, focusing my work on linear algebra functions and OpenCL programming, due to its portability.
I started developing my own linear algebra library, that relies on a collection of OpenCL kernels i'd written to accelerate matrix operations. Since i'm the only developer actually working on this project and still practicing with OpenCL, i was facing the fact that it will take a very long time to me for porting all general linear algebra methods to OpenCL routines and get kernels both performant and most computationally correct.
Thus, i decided to make my library relies on a more stable, accelerated matrix algebra OpenCL library, and clMAGMA seems to be all i need.
I made some researches on current clMAGMA release (0.3.0), i realized that clMAGMA needs code contributors for multi-GPUs environment algorithms and dynamic run-time/scheduling system to improve code performance. I have some experience in concurrent programming and I would be very glad to get in touch with clMAGMA developers to help providing this (and eventually more) features to the library.
So, here are some questions:
1. Is it possible to get involved in clMAGMA development and how to do it?
2. Where i cand find tasks and outlines in clMAGMA development program?
3. What is the very current focus of clMAGMA development team?
4. Which part of the code of clMAGMA at the moment is related to kernel scheduling?
Thanks in advance for your attention.