International Journal of Advancements in Research & Technology, Volume 3, Issue 7, July-2014 142
ISSN 2278-7763
Copyright © 2014 SciResPub. IJOART
Equal and Unequal Memory Partitioning and Task Scheduling for Multi-Processor
System On Chip
Jenitha A
1
Dr. Elumalai R
2
1. Associate Professor, Dept of ECE, EPCET,
Bidrahalli, Virgonagar Post, Bangalore – 49. Karnataka.
Email : 28jenitha@gmail.com, Phone : 9449680553
2. Professor and Head, Dept of IT , MSRIT,
Mathekeri, MS Nagar , Bangalore. Karnataka.
Email : elumalai.epcet@gmail.com Phone : 9448826813
Abstract: - The growing trend in embedded
system is to deploy a Multiprocessor system on
chip (MPSoC). MPSoC has heterogeneous
processing elements, levels of memory hierarchy
and input/output component which are linked
together by an on chip interconnect structure.
Such architectures provide the flexibility to meet
the performance requirement of multimedia
application while respecting constraint on
memory, cost, size, time and power. The MPSoC
is an attractive solution for increase in
complexity and size of embedded applications.
While embedded system becomes increasingly
complex, the increase in memory access speed
has failed to keep up with processor speed. This
makes the memory access latency a major issue
in scheduling the task of an embedded
application on the processor and partitioning
the memory among the processors are two
critical issues in such systems. This paper
presents an integrated approach to task
scheduling and SRAM memory partitioning to
reduce the execution time of embedded
applications.
Keyword:- MPSOC, Task scheduling,
Memory partitioning, SRAM
I. INTRODUCTION
Due to clock and power constraints, architecture
with multiple processor on a single chip have
become attractive solution to achieve performance
in both high end and low end computing
Multiprocessor with large number of different
processing cores are now common for variety of
reasons, especially in embedded systems. MPSoC
consist multiple heterogeneous processing
element, memory hierarchy, input/output
components which are linked together by an on
chip interconnect structure. A critical component
of a chip multiprocessor is its memory subsystem.
This is because both power and performance
behaviour of a chip multiprocessor is largely
shaped by its on-chip memory. While it is possible
to employ conventional memory designs such as
pure private memory or pure shared memory, such
designs are very general and rigid, and may not
generate the best behaviour for a given embedded
application. Execution time predictability is the
critical issue for real time embedded applications;
this means that data caches are not suitable and
hard to model the exact behavior and to predict the
execution time of tasks. Software controlled
memories allow execution times to be predicted
with accuracy.
There is a large number of complex
embedded applications consisting of multiple
concurrent real Time tasks [1]. The problem of
scheduling is weighted directed acyclic graph
(DAG), also called a task graph or macro dataflow
graph, to a set of homogenous processor to
minimize the completion time. The tasks can be
scheduled on different processors. To alleviate
such problems, many modern MPSoC systems use
IJOART