Ephedrine QoS: An Antidote to Slow, Congested, Bufferless NoCs

Datacenters consolidate diverse applications to improve utilization. However when multiple applications are colocated on such platforms, contention for shared resources like networks-on-chip (NoCs) can degrade the performance of latency-critical online services (high-priority applications). Recently...

Full description

Saved in:
Bibliographic Details
Published in:TheScientificWorld 2014-01, Vol.2014 (2014), p.1-11
Main Authors: Bao, Yungang, Sui, Xiufeng, Yao, Zhicheng, Fang, Juan
Format: Article
Language:English
Subjects:
Age
Antidotes
Computer Communication Networks - trends
Computer networks
Ephedrine
Information networks
Internet service providers
Management
Product introduction
Random Allocation
Science
Servers
Wireless Technology - trends
Workloads
Citations: Items that this one cites
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Datacenters consolidate diverse applications to improve utilization. However when multiple applications are colocated on such platforms, contention for shared resources like networks-on-chip (NoCs) can degrade the performance of latency-critical online services (high-priority applications). Recently proposed bufferless NoCs (Nychis et al.) have the advantages of requiring less area and power, but they pose challenges in quality-of-service (QoS) support, which usually relies on buffer-based virtual channels (VCs). We propose QBLESS, a QoS-aware bufferless NoC scheme for datacenters. QBLESS consists of two components: a routing mechanism (QBLESS-R) that can substantially reduce flit deflection for high-priority applications and a congestion-control mechanism (QBLESS-CC) that guarantees performance for high-priority applications and improves overall system throughput. We use trace-driven simulation to model a 64-core system, finding that, when compared to BLESS, a previous state-of-the-art bufferless NoC design, QBLESS, improves performance of high-priority applications by an average of 33.2% and reduces network-hops by an average of 42.8%.
ISSN:2356-6140
1537-744X
1537-744X
DOI:10.1155/2014/691865