OSKI is being actively developed by members of the Berkeley Benchmarking and OPtimization (BeBOP) Project, a research program on automatic performance tuning and performance analysis. We welcome additional development contributions from the user community!
 
The primary OSKI developers are Rich Vuduc, James Demmel, and Kathy Yelick. OSKI was based in large part on foundational work on the Sparsity framework for automatically tuning sparse matrix kernels.
 
The following individuals made significant code contributions to OSKI: Rajesh Nishtala (cache blocking), Benjamin Lee, (symmetric implementations), and Shoaib Kamil and Jen Hsu (triangular solve).

Oski: "Go Bears!"