Using mixed precision for sparse matrix computations to enhance the performance while achieving 64-bit accuracy, (2008)

by A Buttari, J Dongarra, J Kurzak, P Luszczek, S Tomov
Venue:ACM Trans. Math. Softw.