Departmental Metacomputing ML
We have designed a functional data-parallel language called BSML for programming bulk-synchronous parallel algorithms, a model of computing. Dead-locks and indeterminism are avoided and the execution time can be then estimated. For large scale applications where parallel processing is helpful, more than one parallel machine is needed. One speaks about metacomputing. A major problem in programming parallel application for such platforms is their hierarchical network structures: latency and bandwidth of global ones are orders of magnitude worse than those of local networks. Here we consider how an extention of the well-suited BSP model might be for these kinds of computing and how extended our functional language can be for this.
Members
- Frédéric Gava, Université Paris-Est Créteil
- Frédéric Loulergue, Université d'Orléans
Publications
- Frédéric Gava and Frédéric Loulergue. A Functional Language for Departmental Metacomputing. Parallel Processing Letters, 15(3):289-304, 2005. DOI:10.1142/S0129626405002222
- Frédéric Gava. Design of Departmental Metacomputing ML. International Conference on Computational Science, pages 50-53 2004. DOI:10.1007/978-3-540-24688-6_8
Status
The project is currently inactive.
