Future linear-algebra libraries

The ultimate development of fully mature, parallel scalable libraries will necessarily depend on breakthroughs in many supporting technologies. Scalable library development cannot wait, however, until all the enabling technologies are in place for two reasons: The need for such libraries for existing and near-term parallel architectures is immediate, and progress in all the supporting technologies depends on feedback from concurrent efforts in library development. Four major research issues that need to be addressed in developing parallel scalable linear-algebra libraries are identified; inadequate user-library interface, necessity of object-oriented programming for developing portable libraries, the importance of work on algorithms, and the lack of language standards.