Static scheduling and code generation from dynamic dataflow graphs with integer-valued control streams

Abstract

Abstract

Modeling and optimization of dynamic signal processing in resource-aware sensor networks

“…Although CFDF is significantly restricted compared to EIDF, it is highly expressive as a deterministic dataflow model of computation. Indeed, it has been shown that Boolean dataflow actors [4] can be transformed into functionally equivalent CFDF actors [20]. From such a transformation and the established Turing completeness of Boolean dataflow, it can be demonstrated that CFDF is also Turing complete.…”

System-Level Modelling and Design Space Exploration for Multiprocessor Embedded System-on-Chip Architectures

“…The natural solution to this is to place the transformation mechanism in between the application and architecture models, that is, in Sesame's mapping layer. As shown in Figure 4.1, trace transformations which transform application event traces into (more refined) architecture event traces are realized by refining the virtual processor components in Sesame with Integer-controlled dataflow (IDF) [16] graphs. However, before going into details of IDF-based trace transformations, we first formalize traces and their transformations mathematically in the next section.…”

“…The only way to improve this is to improve the mapping strategy. In Sesame, we make use of Integer-controlled Dataflow (IDF) graphs [16] at the mapping layer for this purpose. With the introduction of IDF graphs for mapping, we move from the traditional TD approach to a new mapping strategy which we call the Enhanced Trace Driven (ETD) approach [38].…”

Concurrent models of computation for embedded software

“…Boolean dataflow [41,43] (BDF) and integer-controlled dataflow [44] (IDF) augment the model by permitting the number of tokens produced or consumed at a port to be symbolically represented by a variable. The value of this variable is permitted to change during execution, so datadependent control flow can be represented.…”