A system model and stack for the parallelization of time-critical applications on many-core architectures
Ref: CISTER-TR-141206 Publication Date: 21, Jan, 2015
A system model and stack for the parallelization of time-critical applications on many-core architectures
Ref: CISTER-TR-141206 Publication Date: 21, Jan, 2015Abstract:
Many embedded systems are subject to stringent timing requirements
that compel them to "react" within pre-dened time bounds.
The said "reaction" may be understood as simply outputting the results
of a basic computation, but may also mean engaging in complex interactions
with the surrounding environment. Although these strict temporal
requirements advocate the use of simple and predictable hardware
architectures that allow for the computation of tight upper-bounds on
the software response time, meanwhile most of these embedded systems
steadily demand for more and more computational performance, which
weighs in favor of specialized, complex, and optimized multi-core and
many-core processors on which the execution of the application can be
parallelized. However, it is not straightforward how event-based embedded
applications can be structured in order to take advantage and fully
exploit the parallelization opportunities and achieve higher performance
and energy-efficient computing. The P-SOCRATES project envisions the
necessity to bring together next-generation many-core accelerators from
the embedded computing domain with the programming models and
techniques from the high-performance computing domain, supporting
this with real-time methodologies to provide timing predictability.
This paper gives an overview of the system model and software stack
proposed in the P-SOCRATES project to facilitate the deployment and
execution of parallel applications on many-core infrastructures, while
preserving the time-predictability of the execution required by real-time
practices to upper-bound the response time of the embedded application.
Document:
3rd Workshop on High-performance and Real-time Embedded Systems (HIRES 2015).
Amsterdam, Netherlands.
Record Date: 12, Dec, 2014