Investigating Software Deployment on Heterogeneous Platforms
Paper in proceedings, 2016
Devices with computing capabilities are everywhere.
Physical components that once were restricted to its forms
and mechanics to provide functionality have now gained an
enormous number of possibilities through the implementation of
Embedded Systems (ES). However, together with the advantages
of ES come the challenges of dealing with both the increasing
complexity of software and the different constraints of hardware.
Moreover, it is becoming a common practice to deploy software
on computation units of different types (i.e. CPUs, GPUs and
FPGAs) in order to optimize data processing and resource
utilization. This paper reports on a PhD research project in
the area of software deployment on heterogeneous platforms,
focusing on how different deployment strategies contribute to
achieve certain non-functional properties, such as performance
and energy efficiency. As it is important to understand previous
practices and acknowledge possible gaps in the research area,
a systematic literature review is currently being conducted, and
its initial/planning stages are described. Further, recent contributions
in the context of data marshalling are also discussed, as
well as the planned next steps and the impact of this project in
an industrial context.