Sparse BLAS

Development of community standards for Basic Linear Algebra Subprograms,
computational kernels for matrix computation which enable portability
and performance in core linear algebra software.
Staff:
R. Pozo, K. Remington
With:
Cray Research, University of Minnesota, BLAS Technical Forum
 Pozo chairs Sparse Subcommittee of BLAS Technical Forum.
 Sparse BLAS Toolkit: reference implementation of core matrix multiplication
and triangular solution operations for a variety of matrix data structures.
 Sparse BLAS Lite: lightweight interfaces for each combination of operation
and data structure, automatically generated from templates. Used by Cray
to develop special versions for T3D.
 Webbased source code generation service.
