| Title | A Unified Online Directed Acyclic Graph Flow Manager for Multicore Schedulers |
| Author | *Karim Kanoun, David Atienza (École Polytechnique Fédérale de Lausanne, Switzerland), Nicholas Mastronarde (State University of New York at Buffalo, U.S.A.), Mihaela van der Schaar (University of California, Los Angeles, U.S.A.) |
| Page | pp. 714 - 719 |
| Keyword | Directed Acyclic Graph DAG, Online task graph analyzer, Parallel processing, Multimedia embedded systems, Online energy-efficient scheduler |
| Abstract | The Directed-Acyclic Graph (DAG) monitoring solutions used by existing energy-efficient schedulers to analyze DAGs, make a priori assumptions about the workload and the relationship between the task dependencies. Thus, these schedulers are limited to work on a limited subset of DAG models. To address this problem, we propose a unified online DAG monitoring solution for all possible DAG models to assist online schedulers. We validate our approach using H.264 video decoding application and synthetic DAG models. |
| Slides |
| Title | Variation-Aware Statistical Energy Optimization on Voltage-Frequency Island Based MPSoCs under Performance Yield Constraints |
| Author | *Song Jin (Department of Electronic and Communication Engineering, School of Electrical and Electronic Engineering, North China Electric Power University, China), Yinhe Han (State Key Laboratory of Computer Architecture, Institute of Computing Technology, Chinese Academy of Sciences, China), Songwei Pei (Department of Computer Science and Technology, Beijing University of Chemical Technology, China) |
| Page | pp. 720 - 725 |
| Keyword | energy efficiency, process variation, voltage-frequency island, statistical design, performance yield |
| Abstract | Energy efficiency is a primary design concern for embedded multiprocessor system-on-chips (MPSoCs). Recently, Voltage-Frequency Island (VFI) -based design paradigm was introduced for fine-grained power management, which can seamlessly combine with the task scheduling algorithm to optimize system energy. However, the ever-increasing variabilities cause large uncertainty on delay and power. Such statistical nature in performance parameters easily makes deterministic energy optimization hard to achieve desirable performance yield, defined as the probability of the design meeting timing constraints of the system. In this paper, we propose a variation-aware statistical energy optimization framework, which takes account of performance yield constraints in energy-aware task scheduling, voltage assignment and VFI partitioning process. Energy optimization sensitivity, defined as energy variations of the task under voltage scaling, combines with the statistical slack of the task to guide the overall optimization flow. Experimental results demonstrate the effectiveness of the proposed scheme. |
| Slides |
| Title | QoS-Aware Dynamic Resource Allocation for Spatial-Multitasking GPUs |
| Author | *Paula Aguilera, Katherine Morrow, Nam Sung Kim (University of Wisconsin - Madison, U.S.A.) |
| Page | pp. 726 - 731 |
| Keyword | GPGPU, QoS, spatial multitasking, resource partitioning |
| Abstract | GPGPU computing is becoming widely adopted. Some GPGPU applications fail to fully utilize available GPU resources, motivating the use of spatial multitasking (partitioning resources between simultaneously-running applications). When applications have quality-of-service (QoS) requirements enough resources must be allocated to satisfy their requirements. Remaining resources can be disabled to reduce power consumption or used to accelerate other applications. We propose a runtime algorithm to dynamically partition GPU resources between concurrently running applications, when at least one has QoS requirements. |
| Slides |