Partitioning sparse matrices with eigenvectors of graphs
SIAM Journal on Matrix Analysis and Applications
The Stanford Dash Multiprocessor
Computer
Shared virtual memory on loosely coupled multiprocessors
Shared virtual memory on loosely coupled multiprocessors
Hi-index | 0.00 |
The programming of distributed memory architectures can be significantly facilitated by abstract, application-oriented models. The Promoter model reconciles generality and easiness of use by a new concept of high-level data parallelism that leads to distributed types with parallel operations. The model enables the user to describe the spatial structures for a wide spectrum of numerical and non-numerical application in a uniform algebraic formalism, so that the compiler can readily generate the optimized message-passing program.