the OpenBlas author has always been super responsive and helpful whenever I've had questions or bugs.
For less than the cost of a single user commercial license for Revolution R, Matlab, or Intel MKL, his hardware costs for continuing to provide a great BSD BLAS/Lapack will be covered for the next 2 years (if he raises the amount of money asked for)
I'd never heard of this project before now. While I'm not ever likely to use these libraries directly, I do use a lot of software that could benefit greatly from a BLAS offering similar performance to MKL.
In particular, if you use Julia, you are most likely using OpenBLAS since that's the default BLAS that we ship with. While MKL still has better performance in many situations, OpenBLAS is not far behind and is catching up. Definitely a worthwhile project to contribute to.
http://www.cs.utexas.edu/~flame/pubs/GotoTOMS_revision.pdf