Invasive TCPAs provide unprecedented energy efficiency for the parallel execution of nested loop programs by avoiding any global memory access such as GPUs and may even support loops with complex dependencies such as loop-carried dependencies that are not amenable to parallel execution on GPUs.
Vahid Lari is a researcher at the Department of Computer Science, Friedrich-Alexander University Erlangen-Nürnberg (FAU), since 2008. He defended his PhD degree on the topic of “Invasive Tightly Coupled Processor Arrays” in November 2015. He received his bachelor degree in computer engineering in 2005 from University of Isfahan, Iran, and master degree in computer architectures in 2007 from Sharif University of Technology, Iran. His main research interests include fault tolerance, programmable hardware accelerators, the design of massively parallel architectures and system level performance evaluation.
Innehållsförteckning
Introduction.- Invasive Tightly Coupled Processor Arrays.- Self-adaptive Power and Energy Management for TCPAs.- On-Demand Fault Tolerance on Massively Parallel Processor Arrays.- Conclusions and Future Work.