Power optimization for application-specific networks-on-chips: A topology-based approach

Elmiligi, Haytham; Morgan, Ahmed A.; El-Kharashi, M. Watheq; Gebali, Fayez

doi:10.1016/j.micpro.2009.03.002

Cited by 36 publications

(27 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This algorithm has an overall average runtime complexity of O(n log n). Another NoC-specific algorithm can be used to obtain optimum performance, either power optimization [22], area optimization [23], network reliability [24], genetic algorithmbased technique [25], or by using multi-objective task mapping [26]. Mapping a resultant task cluster to a location within the cluster grid is done according to Shell sorting.…”

Section: Cluster Managermentioning

confidence: 99%

Clustered Networks-on-Chip: Simulation and Performance Evaluation

Hassan¹

2017

IJCDS

View full text Add to dashboard Cite

In many-core Networks-on-Chip (NoC) systems, two topics have recently been researched; NoC clustering and NoC simulation. NoC clustering investigates grouping processing elements (PEs) based on common characteristics between them, like spatial locality or communication patterns. Research showed that NoC clustering achieved better performance and load balancing. Furthermore, it allows scalability of the NoC grid. The other topic, NoC simulation, provides tools for early evaluation of NoC systems. This allowed easy investigation of NoC systems with large number of PEs. Plenty of research has been done in both topics separately. In this paper, we try to fill the gap. We extend an existing NoC simulation tool by adding support for simulating NoC clusters and inter-NoC traffic. The presented NoC clustering simulator supports the modular design technique, which in turn offers the flexibility in configuring cluster parameters. Examples of configurable parameters that are supported by our modified simulator are: adding interconnection topologies to configure how the NoC entities are connected with each other, adding a cluster manager that maps tasks to NoC entities based on the inter-NoC communication pattern, adding latency factors of cluster links in the case of inter-NoC traffic, and adapting the routing algorithm for both inter-cluster and intra-cluster traffic. A clustered NoC simulation case study shows the effectiveness of the modifications made to the simulator. For example, a certain NoC setup is simulated twice, initially non-clustered then clustered, using the added features. Collected data shows that the average clustered NoC routers' energy consumption is 50% lower than the consumption in the non-clustered case.

show abstract

Section: Cluster Managermentioning

confidence: 99%

Clustered Networks-on-Chip: Simulation and Performance Evaluation

Hassan¹

2017

IJCDS

View full text Add to dashboard Cite

show abstract

“…Results were compared to Synopsys Power Compiler and Modelsim, showing an error of 2% for power estimation and 3% for throughput. Elmiligi et al [9] propose a topology-based methodology that explores the impact of the NoC topology on system power dissipation. It uses a partitioning algorithm that aims to achieve minimum inter-partition traffic.…”

Section: Related Workmentioning

confidence: 99%

“…2a shows an example of a system represented by a Traffic Distribution Graph (TDG) G = (V,E,ψ). Each edge e ij ∈ E has a weigh factor λ ij which represents the average number of packets per time step transmitted from v i to v j , 1 ≤ i,j ≤ n; where n is the number of PEs [9]. This graph can also be represented in a traffic distribution matrix form (λ), as shown in Fig.…”

Section: A a Graph-theoretic Approach For Application Modelingmentioning

confidence: 99%

“…This matrix represents the minimum number of links a packet goes through during its transition from the one node to another node [9]. For example, the connectivity matrix (C) of the ring network topology shown in Fig.…”

Section: A a Graph-theoretic Approach For Application Modelingmentioning

confidence: 99%

“…However, for some topologies this is not true (such as torus topology), there are some links do not have the same length as others. To address this problem, the set of entries corresponding to those links in the connectivity matrix C is multiplied by an adjustment factor to correct the length of those links [9]. At the system level, the total power consumed in the global links of a network topology can be estimated from [9]: ∑ ∑ (1) Where i and j are the source and destination node indexes respectively, λ ij is the average number of packets per time step associated with each logical link.…”

Section: B System Analysis Of the Power Consumption On The Global Inmentioning

confidence: 99%

See 2 more Smart Citations