Academy of Finland
Funding decision
 
Name Lilius, Johan
Organisation Åbo Akademi University
Project title ESC - Efficient Stream Computing by Fitting Computations to Cores
Decision No. 308980
Decision date 20.06.2017
Funding,period 01.09.2017 - 31.08.2021
Funding 572 501
WebFOCUS Report
Project description
Stream Computing has been introduced as a paradigm in Cloud Computing to emphasise the streaming nature of modern computing applications. Recently heterogeneous architectures have been introduced (DSPs, GPUs) to process parts of the workload to achieve better energy-effectiveness at the given real-time constraints. In heterogeneous architectures more energy-efficient specialized compute-units are used, and the the design problem amounts to finding a decomposition of the problem into tasks together with a mapping onto the compute-units that achieves minimal energy consumption at a given performance level. The ESC project addresses this mapping problem by introducing the notion of Fitness. Fitness should characterise how much of the micro-architecture the task is able to exploit. Together with the metric we will propose a design methodology that supports the use of the metric at different of abstractions from instructions on CPU core level to many-core platforms.