Tradeo s Between Size and Time in Recon gurable MeshesYosi
نویسندگان
چکیده
The reconngurable mesh model (RN-mesh) for parallel computations exhibits some \strong" results of algorithms involving a constant number of broadcasting steps. Naturally, one can ask whether or not these constants are tight. Consider an nn directed reconngurable mesh (DRN-mesh) that computes a function f(n) in a constant number of steps T (directed indicates that the edges of the mesh are directed so that signals (at a given edge) ow in one direction only). In this paper we show that T can always be reduced to a single step, still using a polynomial size DRN-mesh (n 2 T). Moreover, we show that this is in fact a general tradeoo, i.e., by expanding the size of the DRN-mesh, one can reduce the number of steps to any value between 1 and T. We also show that this size expansion can be further reduced, when f is a composition of functions.
منابع مشابه
Reconngurable Computing: Architectures, Models and Algorithms Current Science: Special Section on Computational Science
ion Mapping Parameters Scheduling Figure 2: Traditional Design Synthesis Approach and the Model-based Approach One major problem in using FPGAs to speed-up a computation is the design process. The \standard CAD approach" used for digital design is typically employed (see Figure 2). The required functionality is speci ed at a high level of abstraction via an HDL or a schematic. FPGA libraries sp...
متن کاملConstant Time Algorithms for Computing the Contour of Maximal Elements on the Recon gurable Mesh
Elements on the Recon gurable Mesh M. Manzur Murshed and Richard P. Brent Computer Sciences Lab, Research School of Information Sciences & Engg. The Australian National University, Canberra ACT 0200, Australia E-mail: fmurshed,[email protected] Abstract There has recently been an interest in the introduction of recon gurable buses to existing parallel architectures. Among them Recon gurable...
متن کاملSelf Con guring Binary Multipliers for LUT addressable FPGAs
In this paper we present a self con gurable multiplication technique allowing vari able con guration time for a class of LUT based Field Programmable Gate Arrays FPGAs which exist today We show this technique to be implementable on FPGA architectures allowing internally addressable RAM primitives to be directly mapped to the Logic Elements LEs of the logic resource This provides run time read w...
متن کاملRun-Time Management of Dynamically Recon gurable Designs
A method for managing recon gurable designs, which supports run-time con guration transformation, is proposed. This method involves structuring the recon guration manager into three components: a monitor, a loader, and a con guration store. Di erent trade-o s can be achieved in con guration time, optimality of the con gured circuits, and the complexity of the recon guration manager, depending o...
متن کاملA Novel Dynamic Task Scheduling Algorithm for Grid Networks with Recon gurable Processors
The incorporation of recon gurable processors in computational grids promises to provide increased performance without compromising exibility. Consequently, grid networks (such as TeraGrid) are utilizing recon gurable computing resources next to general-purpose processors (GPPs) in their computing nodes. The near-optimal utilization of resources in a grid network considerably depends on the app...
متن کامل