Energy Efficient Synchronization Techniques for Embedded Architectures


Because many embedded devices run on batteries, energy efficiency is perhaps the single most important criterion for evaluating hardware and software effectiveness in embedded devices. In this project, we have looked into developing energy-efficient speculative approaches such as transactional memory to managed shared data on an embedded platform.  The challenge is to make transactional memory on an embedded platform cost-effective both in terms of throughput and energy. This year work has focused on translating these speculative techniques from multicore to many-core embedded system platforms that use network-on-chip architectures to configure shared memory.    This work is supported by NSF and SRC.

Current People Involved:

- R. Iris Bahar
Cesare Ferri
- Hamutal Moreshet (Swarthmore College)
- Maurice Herlihy (Brown, CS)
- Andrea Marongiu (post-doctorial researcher from University of Bologna, Italy).

Related Publications