Mutli-objective trade-off exploration for Cyclo-Static and Synchronous Dataflow graphs
| dc.contributor.advisor | Gerstlauer, Andreas, 1970- | en |
| dc.contributor.committeeMember | John, Lizy K. | en |
| dc.creator | Sinha, Ashmita | en |
| dc.date.accessioned | 2012-10-30T19:14:11Z | en |
| dc.date.accessioned | 2017-05-11T22:29:16Z | |
| dc.date.available | 2012-10-30T19:14:11Z | en |
| dc.date.available | 2017-05-11T22:29:16Z | |
| dc.date.issued | 2012-08 | en |
| dc.date.submitted | August 2012 | en |
| dc.date.updated | 2012-10-30T19:14:22Z | en |
| dc.description | text | en |
| dc.description.abstract | Many digital signal processing and real-time streaming systems are modeled using dataflow graphs, such as Synchronous Dataflow (SDF) and Cyclo-static Dataflow (CSDF) graphs that allow static analysis and optimization techniques. However, mapping of such descriptions into tightly constrained real-time implementations requires optimization of resource sharing, buffering and scheduling across a multi-dimensional latency-throughput-area objective space. This requires techniques that can find the Pareto-optimal set of implementations for the designer to choose from. In this work, we address the problem of multi-objective mapping and scheduling of SDF and CSDF graphs onto heterogeneous multi-processor platforms. Building on previous work, this thesis extends existing two-stage hybrid heuristics that combine an evolutionary algorithm with an integer linear programming (ILP) model to jointly optimize throughput, area and latency for SDF graphs. The primary contributions of this work include: (1) extension of the ILP model to support CSDFGs with additional buffer size optimizations; (2) a further optimization in the ILP-based scheduling model to achieve a runtime speedup of almost a factor of 10 compared to the existing SDFG formulation; (3) a list scheduling heuristic that replaces the ILP model in the hybrid heuristic to generate Pareto-optimal solutions at significantly decreased runtime while maintaining near-optimality of the solutions within an acceptable gap of 10% when compared to its ILP counterparts. The list scheduling heuristic presented in this work is based on existing modulo scheduling approaches for software pipelining in the compiler domain, but has been extended by introducing a new concept of mobility-based rescheduling before resorting to backtracking. It has been proved in this work that if mobility-based rescheduling is performed, the number of required backtrackings and hence overall complexity and runtime is less. | en |
| dc.description.department | Electrical and Computer Engineering | en |
| dc.format.mimetype | application/pdf | en |
| dc.identifier.slug | 2152/ETD-UT-2012-08-5995 | en |
| dc.identifier.uri | http://hdl.handle.net/2152/ETD-UT-2012-08-5995 | en |
| dc.language.iso | eng | en |
| dc.subject | Synchronous Dataflow graph (SDF) | en |
| dc.subject | Cyclo-Static Dataflow graph (CSDF) | en |
| dc.subject | Integer Linear Programming (ILP) | en |
| dc.subject | List scheduling | en |
| dc.subject | Mapping | en |
| dc.subject | Throughput | en |
| dc.subject | Latency | en |
| dc.subject | Buffer | en |
| dc.subject | Area | en |
| dc.title | Mutli-objective trade-off exploration for Cyclo-Static and Synchronous Dataflow graphs | en |
| dc.type.genre | thesis | en |