Bio-inspired distributed task remapping for multiple video stream decoding on homogeneous NoCs

Abstract

Centralised management of distributed systems require a significant amount of monitoring traffic to maintain an accurate view of the system global state. The communication overhead of these systems becomes a bottleneck as the number of processing elements in the network and workload increase. State-of-the art in decentralised resource management techniques address this issue by allowing individual or clusters of nodes to make decisions at runtime to manage the dynamic workload. The primary contribution of this paper is using a bio-inspired, distributed, task remapping technique to manage dynamic multiple video stream decoding workloads. Our proposed technique has a low-communication overhead and is used to reduce the cumulative job lateness of the video streams. Secondary contributions include, several improvements to an existing clusterbased resource management approach to introduce awareness of task blocking and relocation distance. We evaluate these two remapping methods by comparing the improvement of job lateness, communication overhead and distribution of utilisation via simulation of several workload patterns.

A Survey and Comparative Study of Hard and Soft Real-Time Dynamic Resource Allocation Strategies for Multi-/Many-Core Systems

“…reallocations are performed based on observed biological phenomenon. They have been well explored to balance communication loads in networks and distributed systems [Nishitha and Reddy 2012;da Silva Rego et al 2012], communication loads in many-core systems [Rowlings et al 2015], and both computation and communication loads in embedded systems [Mendis et al 2015]. These approaches usually employ distributed resource management to overcome the limitations of centralized and clustered (hierarchical distributed) managements for dynamic applications and large scale many-core systems.…”

“…Fig. 7 illustrates an example to reallocate some late tasks in a 3×3 many-core system by employing biological inspiration that is based on pheromone signalling mechanism as seen in social insects (e.g honey bees) in order to improve the overall performance [Mendis et al 2015]. At each reallocation interval, late tasks identification is done on each node to find late tasks in their task queues.…”

Dynamic Resource Allocation in Embedded, High-Performance and Cloud Computing

“…To reduce this job lateness of the admitted dynamic varying workload, we follow a bio-inspired distributed task-remapping technique with self-organising properties. This technique builds upon existing bio-inspired load balancing approaches by Caliskanelli et al [30] and Mendis et al [91].…”

“…A single centralised resource manager receives status updated from every slave PE in the network and performs periodic remapping as described in [91]. The manager notifies the task dispatcher of any remapping decisions.…”

“…• CCPRM V 2 -is a cluster-based management proposed in [91] as an improvement of the original work by Castilhos et al [33]. It is configured with a cluster size of 2 × 5 (i.e., 10 clusters).…”