1. This article proposes a novel on-chip DVFS technique for NoCs that is able to adjust per-region V/F level according to voted V/F levels of communicating threads.
2. This voting-based approach allows threads to influence the DVFS decisions independently by voting for a preferred V/F level that best suits their own performance interest.
3. Experimental results show that this voting based DVFS mechanism improves the network energy efficacy measured in MPPJ and MIPJ by about 17.9% and 26.3%, respectively, compared with a network without DVFS and a network metric (router buffer occupancy) based approach.
This article provides an interesting approach to dynamic voltage/frequency scaling (DVFS) in networks-on-chip (NoCs). The proposed technique is able to adjust per-region V/F level according to voted V/F levels of communicating threads, allowing each thread to influence the DVFS decisions independently by voting for a preferred V/F level that best suits their own performance interest. The authors provide detailed simulations of a 64-core CMP running a variety of multi-threaded PARSEC benchmarks, which demonstrate improved energy efficacy compared with other approaches.
The article appears to be well researched and reliable, as it provides detailed information on the proposed technique as well as experimental results from simulations of real world applications. The authors also provide references for further reading on related topics, which adds credibility to the article's claims. Furthermore, the authors present both sides of the argument fairly and objectively, providing evidence for both their proposed technique as well as other approaches they compare it against.
The only potential bias in this article is that it does not explore any counterarguments or alternative solutions beyond those mentioned in the paper itself; however, this does not detract from its overall reliability or trustworthiness since it focuses solely on presenting its own proposed solution rather than attempting to discredit other approaches.