Date: Thursday 27 March 2014
Time: 11:00 - 12:30
Location / Room: Konferenz 4
Chair:
Orlando Moreira, Ericsson, NL
Co-Chair:
Giuseppe Lipari, ENS - Cachan, FR
This session features new results in scheduling, allocation and management of real-time application in multi-core and distributed systems. The first paper presents a control algorithm for managing real-time tasks so to meet thermal constraints in a multi-core chip. The second paper proposes an algorithm for mixed-criticality task allocation in a multiprocessor platform. The third paper proposes a method for generating a schedule for a multi-mode application in a distributed system.
| Time | Label | Presentation Title Authors |
|---|---|---|
| 11:00 | 10.6.1 | THERMAL-AWARE FREQUENCY SCALING FOR ADAPTIVE WORKLOADS ON HETEROGENEOUS MPSOCS Speakers: Heng Yu, Rizwan Syed and Yajun Ha, National University of Singapore, SG Abstract For applications featuring adaptive workloads, the quality of their task execution can be dynamically adjusted given the runtime constraints. When mapping them to heterogeneous MPSoCs, it is expected not only to achieve the highest possible execution quality, but also meet the critical thermal challenges from the continuously increasing chip density. Prior thermal management techniques, such as Dynamic Voltage/Frequency Scaling (DVFS) and thread migration, do not take into account the trade-off possibility between execution quality and temperature control. In this paper, we explore the capability of adaptive workloads for effective temperature control, while maximally ensuring the execution Quality-of-Service (QoS). We present a thermal-aware dynamic frequency scaling (DFS) algorithm on heterogeneous MPSoCs, where judicious frequency selection achieves QoS maximization under the temperature threshold, which is converted to the thermal-timing deadline as an additional execution constraint. Results show that our frequency scaling algorithm achieves as large as 31.5% execution cycle/QoS improvement under thermal constraints. |
| 11:30 | 10.6.2 | PARTITIONED MIXED-CRITICALITY SCHEDULING ON MULTIPROCESSOR PLATFORMS Speakers: Chuancai Gu1, Nan Guan2, Qingxu Deng1 and Wang Yi2 1Northeastern University, CN; 2Uppsala University, SE Abstract Scheduling mixed-criticality systems that integrate multiple functionalities with different criticality levels into a shared platform appears to be a challenging problem, even on single-processor platforms. Multi-core processors are more and more widely used in embedded systems, which provide great computing capacities for such mixed-criticality systems. In this paper, we propose a partitioned scheduling algorithm MPVD to extend the state-of-the-art single-processor mixed-criticality scheduling algorithm EY to multiprocessor platforms. The key idea of MPVD is to evenly allocate tasks with different criticality levels to different processors, in order to better explore the asymmetry between different criticality levels and improve the system schedulability. Then we propose two enhancements to further improve the schedulability of MPVD. Experiments with randomly generated task sets show significant performance improvement of our proposed approach over existing algorithms. |
| 12:00 | 10.6.3 | GENERATION OF COMMUNICATION SCHEDULES FOR MULTI-MODE DISTRIBUTED REAL-TIME APPLICATIONS Speakers: Akramul Azim, Gonzalo Carvajal, Rodolfo Pellizzoni and Sebastian Fischmeister, University of Waterloo, CA Abstract A key problem in designing multi-mode real-time systems is the generation of schedules to reduce the complexities of transforming the model semantics to code. Moreover, multi-mode applications that require communications are prone to suffer from delays incurred during mode changes. We therefore aim to generate communications schedules that have low average mode-change delay for multi-mode real-time distributed applications. In this paper, we use optimization constraints associated to timing requirements to generate state-based schedules that are specific to multi-mode communication systems, and also demonstrate the workflow for generating schedules from specifications through a real-time video monitoring case-study. In terms of average mode-change delays, our experiments demonstrate that schedules generated using our method are more efficient than either a randomized algorithm or the well-known EDF scheduling algorithm. |
| 12:30 | End of session Lunch Break in Exhibition Area Sandwich lunch |