Loading...

Energy Efficient Dynamic Memory Allocation in Embedded Systems Equipped with Non-Volatile Memory

Ramezan Poursafaei, Farimah | 2016

754 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 48849 (19)
  4. University: Sharif University of Technology
  5. Department: Computer Engineering
  6. Advisor(s): Ejlali, Alireza
  7. Abstract:
  8. Energy consumption is one of the most important design constraints in embedded systems and a considerable fraction of the energy is consumed in memories. Memories play an important role in the systems since they occupy a large portion of the die area and they have a substantial impact on the performance of the systems. Therefore, managing their energy is of great importance. Moreover, as the technology scales down, the static energy becomes more important. For instance, in conventional memories like SRAM, the static energy becomes a considerable part of the total energy. Non-volatile memories (NVMs) have been proposed as a promising candidate for conventional memories. NVMs have appealing characteristics including low leakage power and high density. However, their employment in the memory hierarchy has two challenges. First, a write operation to NVMs incurs more costs in terms of energy and access latency. Second, NVMs have limited write endurance. In systems equipped with memory management unit (MMU), the tasks data is partitioned into pages at compile time and the data is managed at page level during runtime. In this proposal, we propose an NVM-aware dynamic memory management technique for multicore real-time embedded systems that are equipped with MMU. With the consideration of the access patterns of data as well as the characteristics of the different memories, the proposed technique specifies the allocation of the tasks variables to the memory pages and determines the mapping of the pages into physical memory. The experiments show that the proposed technique is able to improve the energy consumption of the memory system by 45% on average
  9. Keywords:
  10. Embedded System ; Energy Consumption ; Nonvolatile Memory ; Memory Management ; Dynamic Memory Allocation Algorithm

 Digital Object List

 Bookmark

...see more