The Networks-on-Chip paradigm has been seen as an interconnect architecture solution for complex systems. However, performance and energy issues still represent limiting factors for Multi-Processors System-on-Chip. Moreover, the execution of different applications requires flexible and transparent interconnection solutions, and this feature is best provided by a selfadaptable system. In this paper we propose HASIN, an architecture that explores the suitable switching architecture according to the traffic in each region of the system, in a hierarchical manner. The proposed interconnection allows adapting the network at runtime using three switching possibilities to reconfigure itself according to the floorplan information. HASIN allows increasing the throughput up to 77% and reducing the power consumption up to 76% when compared to a packet-switched mesh network-on-chip.
Hierarchical and Multiple Switching NoC with Floorplan Based AdaptabilityReconfigurable Computing: Architectures, Tools and Applications
PALERMO, GIANLUCA;
2013-01-01
Abstract
The Networks-on-Chip paradigm has been seen as an interconnect architecture solution for complex systems. However, performance and energy issues still represent limiting factors for Multi-Processors System-on-Chip. Moreover, the execution of different applications requires flexible and transparent interconnection solutions, and this feature is best provided by a selfadaptable system. In this paper we propose HASIN, an architecture that explores the suitable switching architecture according to the traffic in each region of the system, in a hierarchical manner. The proposed interconnection allows adapting the network at runtime using three switching possibilities to reconfigure itself according to the floorplan information. HASIN allows increasing the throughput up to 77% and reducing the power consumption up to 76% when compared to a packet-switched mesh network-on-chip.File | Dimensione | Formato | |
---|---|---|---|
ARC13.pdf
Accesso riservato
:
Post-Print (DRAFT o Author’s Accepted Manuscript-AAM)
Dimensione
292.73 kB
Formato
Adobe PDF
|
292.73 kB | Adobe PDF | Visualizza/Apri |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.