Trace-based KPN composability analysis for mapping simultaneous applications to MPSoC platforms

Abstract

Nowadays, most embedded devices need to support multiple applications running concurrently. In contrast to desktop computing, very often the set of applications is known at design time and the designer needs to assure that critical applications meet their constraints in every possible use-case. In order to do this, all possible use-cases, i.e. subset of applications running simultaneously, have to be verified thoroughly. An approach to reduce the verification effort, is to perform composability analysis which has been studied for sets of applications modeled as Synchronous Dataflow Graphs. In this paper we introduce a framework that supports a more general parallel programming model based on the Kahn Process Networks Model of Computation and integrates a complete MPSoC programming environment that includes: compilercentric analysis, performance estimation, simulation as well as mapping and scheduling of multiple applications. In our solution, composability analysis is performed on parallel traces obtained by instrumenting the application code. A case study performed on three typical embedded applications, JPEG, GSM and MPEG-2, proved the applicability of our approach.

Component-based waveform development: the Nucleus tool flow for efficient and portable software defined radio

“…Trace generation for these blocks works exactly as in the regular MAPS flow [7]. The CPN application is first instrumented and executed in parallel on the host (similar to the execution described in Sect.…”

“…Dynamic checking requires a parallel timing estimation, for which a schedule of the waveform has to be computed. For this, the mapping and scheduling component of the MAPS framework is used [7]. Details on how the Nucleus mapper is interfaced with the MAPS mapping and scheduling component are given in Sect.…”

Translating affine nested-loop programs with dynamic loop bounds into Polyhedral Process Networks

“…We are interested in the process network model because it provides a sound formalism, well suited for capturing and modeling of dataflow dominated applications in the realm of multimedia, imaging, and signal processing, that naturally contain tasks communicating via streams of data. Moreover, it has been already shown that process networks allow effective and efficient mappings of streaming applications to certain parallel execution platforms [7], [8], [9], [10], [11], [12].…”

A Unified Execution Model for Data-Driven Applications on a Composable MPSoC

“…Moreover, unlike the existing work, our approach is composable. Various definition of composability exist [11], [14]. Our definition however is more restrictive in that inter-application interference is completely prohibited.…”

“…The platform in [15] executes KPN processes by scheduling them on reconfigurable accelerators. [14] proposes an MPSoC platform that supports multiple real-time applications. The system performance is estimated using the individual application timing profile and a model for the inter-application interference.…”