Some opportunities for dynamic scheduling in distributed mathematical programming libraries
1 : Department of Mathematics [Stanford]
-
Website
Stanford University, Department of Mathematics, building 380, Stanford, California 94305 -
United States
Opportunities for improving upon current best-practice techniques for block synchronous distributed-memory parallelizations of (fixed and arbitrary-precision) dense and sparse-direct linear algebra, Interior Point Methods, and lattice reduction will be given from the perspective of pitfalls of MPI library development within Elemental.
| Subject : | : | oral |
| Topics | : | Dynamic scheduling |
| PDF version | : |  PDF version |
