An Adaptive, Energy-efficient Object Cache Architecture
S.S. Shekhar, Y.N. Srikant

IISc-CSA-TR-2006-1
(January 2006)

Available formats: [pdf]

Filed on January 11, 2006
Updated on January 11, 2006



An important trend in the design of embedded computing systems
is the proliferation of energy conscious architectures. Another important
trend is the emergence of Java as a popular platform of choice
for implementing various applications that run on mobile and hand
held devices. General purpose cache architectures that support Java
in embedded systems do not respond well to special properties and
behavior of object oriented programs and thus consume a significant
fraction of the total processor energy. In this paper we present an
adaptive, energy-efficient object cache architecture and a hardware
implementable reconfiguration algorithm to alleviate the problem of
increased energy consumption of cache architectures in embedded systems
that typically support runtime environments for object oriented
programs.

An Object-cache architecture consists of a data cache (reduced in
size) and an additional small cache structure that caches only the objects.
Certain special properties of object oriented programs such as a
high degree of temporal locality among object references and a short
life span of objects ensures good performance of such an architecture.
At the same time, the reduced active cache size, at times accompanied
by minor improvements in performance, leads to a significant
improvement in the energy behavior of programs. However, both energy
behavior and performance vary widely with the configuration of
the Object-cache. Thus we propose an adaptive object cache architecture
and a reconfiguration algorithm for dynamically reconfiguring
the size and associativities of the Object-cache such that it obtains
the best possible energy efficiency with the least impact on performance.
Using applications from SPECjvm98 benchmarks and a cycle
accurate simulation, the adaptive Object-cache architecture is shown
to reduce average data cache energy consumption by up to 43% and
also improve performance by about 1%.


Please bookmark this technical report as http://aditya.csa.iisc.ernet.in/TR/2006/1/.

Problems ? Contact techrep@csa.iisc.ernet.in
[Updated at 2009-10-22T06:42Z]