|
|
|
Current
Research
Energy consumption in computer systems has become of vital importance,
especially for battery-operated portable devices, where the lifetime of a
device is constrained by its battery capacity. Managing energy
consumption, can extend the battery life, or alternatively, allow more
capabilities in a device with the same battery capacity. Furthermore,
reduced energy can result in lighter devices and less expensive packaging;
ultimately reducing the total cost of building portable systems.
In my
research, I investigate energy management in processors and in memory
independently. This work is motivated by the high energy consumption in
these two subsystems (which consume about half of the of the total system
power combined). In processors, I propose a scheme that proves its
efficiency by almost doubling the battery lifetime in the system while
honoring real-time performance constraints on application's execution
time. To save energy in memories, I present a new memory organization that
achieves high energy savings (20% saving) while improving the overall
memory performance (11% improvement), in contrast to other techniques that
sacrifice part of the performance to reduce energy. Moreover, I study the
interaction between the two components (processor and memory) to achieve
higher energy savings by coordinating power management in both components.
My results show that energy savings can reach up to 28% over using
techniques for each component individually. These savings open the way for
further study of coordinated power management in different system
components.
Research
Project
|
|
|