Throughput-Buffering Trade-Off Exploration for Cyclo-Static and Synchronous Dataflow Graphs

Abstract

Abstract-Multimedia applications usually have throughput constraints. An implementation must meet these constraints, while it minimizes resource usage and energy consumption. The compute intensive kernels of these applications are often specified as Cyclo-Static or Synchronous Dataflow Graphs. Communication between nodes in these graphs requires storage space which influences throughput. We present an exact technique to chart the Pareto space of throughput and storage trade-offs, which can be used to determine the minimal buffer space needed to execute a graph under a given throughput constraint. The feasibility of the exact technique is demonstrated with experiments on a set of realistic DSP and multimedia applications. To increase scalability of the approach, a fast approximation technique is developed that guarantees both throughput and a, tight, bound on the maximal overestimation of buffer requirements. The approximation technique allows to trade off worst-case overestimation versus run-time.

Iteration-Based Trade-Off Analysis of Resource-Aware SDF

“…RELATED WORK Much work has been done on analyzing SDFG throughput and on synthesizing schedules that minimize the resource of buffer sizes [3], [8], [10], [14], [18], [23]. Only recently, tradeoff analysis for SDFGs [17], [21], [22], [24] is investigated. [17] uses bottleneck analysis to explore distributed buffer size configurations efficiently while [22] applies this approach to the more general RASDFG model.…”

“…Only recently, tradeoff analysis for SDFGs [17], [21], [22], [24] is investigated. [17] uses bottleneck analysis to explore distributed buffer size configurations efficiently while [22] applies this approach to the more general RASDFG model. [23], [24] investigate tradeoffs between cost and performance by differently partitioning actors to software and hardware implementations, assuming software and hardware realisations have different cost and performance.…”

“…For example, the states I 3 and S 6,4 in Fig. 3 are (2,15), (3,17), (2,19)} mem { (1,14), (1,17), (1,19)…”

Modeling static-order schedules in synchronous dataflow graphs

“…Model-based design-flows (e.g., [1], [5]- [8]) model binding and scheduling decisions into the SDFG. This enables analysis of performance properties (e.g, throughput [9]) or resource requirements (e.g., buffer sizes [10]) under resource constraints. Fig.…”

“…For example, converting the SDFG of an H.263 decoder [10] to the equivalent HSDFG increases the graph size from 4 actors to 200 actors. The runtime of SDFG analysis algorithms depends amongst others on the size of the graph.…”