Bus-invert coding for low-power I/O
IEEE Transactions on Very Large Scale Integration (VLSI) Systems
A Search Procedure for Hamilton Paths and Circuits
Journal of the ACM (JACM)
Proceedings of the 27th annual international symposium on Computer architecture
Fast cache and bus power estimation for parameterized system-on-a-chip design
DATE '00 Proceedings of the conference on Design, automation and test in Europe
Address bus encoding techniques for system-level power optimization
Proceedings of the conference on Design, automation and test in Europe
Reducing Address Bus Transitions for Low Power Memory Mapping
EDTC '96 Proceedings of the 1996 European conference on Design and Test
Memory access scheduling and binding considering energy minimization in multi-bank memory systems
Proceedings of the 41st annual Design Automation Conference
External memory page remapping for embedded multimedia systems
Proceedings of the 2007 ACM SIGPLAN/SIGBED conference on Languages, compilers, and tools for embedded systems
AMBER: adaptive energy management for on-chip hybrid video memories
Proceedings of the International Conference on Computer-Aided Design
Hi-index | 0.00 |
Power efficiency has become a key design trade-off in embedded system designs. For system-on-a-chip embedded systems, an external bus interconnects embedded processor cores, I/O peripherals, a direct memory access (DMA) controller, and off-chip memory. External memory access activities are a major source of energy consumption in embedded systems, and especially in multimedia platforms. In this paper, we focus on the energy dissipated due to the address, data, and control activity on the external bus and supporting logic. We build our external bus power model on top of a cycle-accurate simulation framework that quantifies the bus power based on memory bus state transitions. We select an Analog Devices ADSP-BF533 multimedia system-on-a-chip embedded system as our target architecture model. Using our power-aware external bus arbitration schemes, we can reduce overall power by as much as 18% in video processing applications, and by 12% on average for the test suites studied. Besides reducing power consumption, we also obtained an average performance speedup of 24% when using our power-aware arbitration schemes.