[Paper Review] Heuristics for Routing and Spiral Run-time Task Mapping in NoC-based Heterogeneous MPSOCs

Mohammed Kamel Benhaoua, Abbou El Hassen Benyamina+1 more|arXiv (Cornell University)|Dec 19, 2013

Interconnection Networks and Systems22 references4 citations

TL;DR

This paper proposes a Spiral Dynamic Task Mapping heuristic for NoC-based heterogeneous MPSoCs that groups related tasks in a spiral pattern to minimize communication overhead. Combined with a modified Dijkstra routing algorithm, it reduces total execution time and energy consumption compared to state-of-the-art run-time mapping heuristics on an 8x8 NoC platform.

ABSTRACT

This paper describes a new Spiral Dynamic Task Mapping heuristic for mapping applications onto NoC-based Heterogeneous MPSoC. The heuristic proposed in this paper attempts to map the tasks of an applications that are most related to each other in spiral manner and to find the best possible path load that minimizes the communication overhead. In this context, we have realized a simulation environment for experimental evaluations to map applications with varying number of tasks onto an 8x8 NoC-based Heterogeneous MPSoCs platform, we demonstrate that the new mapping heuristics with the new modified dijkstra routing algorithm proposed are capable of reducing the total execution time and energy consumption of applications when compared to state-of the-art run-time mapping heuristics reported in the literature.

Motivation & Objective

• To address high communication overhead in NoC-based heterogeneous MPSoCs during run-time task mapping.
• To improve application performance by minimizing inter-task communication delays.
• To reduce energy consumption through efficient task placement and routing.
• To develop a run-time heuristic that adapts dynamically to workload variations.
• To evaluate the proposed heuristic against existing state-of-the-art mapping strategies.

Proposed method

• The Spiral Dynamic Task Mapping heuristic groups tasks based on their communication affinity, arranging them in a spiral pattern around a central point on the NoC mesh.
• A modified Dijkstra algorithm is used for path selection, prioritizing low-congestion and low-latency routes.
• The heuristic evaluates task relationships and assigns them to neighboring or nearby processing elements to reduce data transfer distance.
• A simulation environment is implemented to evaluate mapping on an 8x8 NoC-based heterogeneous MPSoC platform.
• The approach dynamically re-maps tasks at run-time based on current workload and communication patterns.
• Communication overhead is minimized by reducing hop counts and balancing load across the network.

Experimental results

Research questions

• RQ1 How can task mapping in NoC-based heterogeneous MPSoCs be optimized to reduce communication overhead?
• RQ2 Can a spiral-based task placement strategy improve performance and energy efficiency?
• RQ3 How does the proposed heuristic compare to existing run-time mapping heuristics in terms of execution time and energy consumption?
• RQ4 What impact does the modified Dijkstra routing algorithm have on path selection and load balancing?
• RQ5 To what extent does dynamic task mapping reduce inter-task communication delays in real-time workloads?

Key findings

• The proposed Spiral Dynamic Task Mapping heuristic significantly reduces total execution time compared to state-of-the-art run-time mapping heuristics.
• Energy consumption is reduced due to minimized inter-task communication distances and optimized routing.
• The modified Dijkstra algorithm contributes to better load balancing and lower path latency in the NoC.
• The simulation results demonstrate consistent performance improvements across applications with varying numbers of tasks.
• The spiral mapping approach effectively groups communication-intensive tasks, reducing overall network contention.
• The heuristic outperforms existing methods in both execution time and energy efficiency on the 8x8 NoC platform.