國立高雄大學圖資館 |

Language: English

Back

Cost-effective lifetime and yield op...

Carnegie Mellon University.

Cost-effective lifetime and yield optimization for NoC-based MPSoCs.

Record Type:	Electronic resources : Monograph/item
Title/Author:	Cost-effective lifetime and yield optimization for NoC-based MPSoCs.
Author:	Meyer, Brett H.
Description:	163 p.
Notes:	Source: Dissertation Abstracts International, Volume: 70-09, Section: B, page: 5719.
Contained By:	Dissertation Abstracts International70-09B.
Subject:	Engineering, Computer.
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3376474
ISBN:	9781109373769

Cost-effective lifetime and yield optimization for NoC-based MPSoCs.
Meyer, Brett H.

Cost-effective lifetime and yield optimization for NoC-based MPSoCs. - 163 p.

Source: Dissertation Abstracts International, Volume: 70-09, Section: B, page: 5719.

Thesis (Ph.D.)--Carnegie Mellon University, 2009.

This research is focused on the design-time system-level architectural optimization of cost, lifetime and yield in embedded network-on-chip-based multi-processor-systems-on-chip (NoC-based MPSoCs). At the system level, the precise nature and timing of a fault is irrelevant because the fault results in the (possibly temporary) loss of an entire processor, memory, or interconnect module regardless. One advantage of managing failure at the computer system level is therefore that once the location of a failure has been identified, the cause can be abstracted away. In this case, failures of different types may be treated the same and addressed using the same techniques. Based on this observation, we employ system-level slack---excess capacity in processor and memory nodes available to accommodate additional tasks in the event that other processors or memories are lost---as a general technique for mitigating MPSoC failure in the presence of either component manufacturing defects or permanent component failures.

ISBN: 9781109373769Subjects--Topical Terms:

384375
Engineering, Computer.

Cost-effective lifetime and yield optimization for NoC-based MPSoCs.
LDR:05525nmm 2200301 4500 001 280734
005 20110119094941.5
008 110301s2009 ||||||||||||||||| ||eng d
020 $a 9781109373769
035 $a (UMI)AAI3376474
035 $a AAI3376474
040 $a UMI $c UMI
100 1 $a Meyer, Brett H. $3 492815
245 1 0 $a Cost-effective lifetime and yield optimization for NoC-based MPSoCs.
300 $a 163 p.
500 $a Source: Dissertation Abstracts International, Volume: 70-09, Section: B, page: 5719.
502 $a Thesis (Ph.D.)--Carnegie Mellon University, 2009.
520 $a This research is focused on the design-time system-level architectural optimization of cost, lifetime and yield in embedded network-on-chip-based multi-processor-systems-on-chip (NoC-based MPSoCs). At the system level, the precise nature and timing of a fault is irrelevant because the fault results in the (possibly temporary) loss of an entire processor, memory, or interconnect module regardless. One advantage of managing failure at the computer system level is therefore that once the location of a failure has been identified, the cause can be abstracted away. In this case, failures of different types may be treated the same and addressed using the same techniques. Based on this observation, we employ system-level slack---excess capacity in processor and memory nodes available to accommodate additional tasks in the event that other processors or memories are lost---as a general technique for mitigating MPSoC failure in the presence of either component manufacturing defects or permanent component failures.
520 $a Given an application and fixed NoC-based communication architecture, our goal is to cost-effectively perform slack allocation, distributing execution and storage slack such that with high probability when manufacturing defects or permanent component failure occurs, sufficient resources remain for the system to continue to operate. The design space for slack allocation is large and complex. The design space consists of every possible slack allocation (up to nm for a system with n components and m possible alternatives in the component library). Furthermore, evaluating the lifetime of any single design is computationally expensive, requiring performance, power, and temperature evaluation for every possible combination of component failures. In one example we considered, an MPEG-4 decoder with 21 processors, 5 memories and 10 switches, there are 1.6 billion possible slack allocations alone (given a fixed communication architecture) and each system lifetime evaluation took from 46.4 to 144.5 seconds.
520 $a To address the complexity of slack allocation, we have developed Critical Quantity Slack Allocation (CQSA), a novel, scalable, generalizable execution and storage slack allocation technique. CQSA takes advantage of the fact that the extra slack required to survive component failure is often quantized: around these quantized levels, less extra slack fails to protect against system failure while more extra slack is unnecessary and may even degrade system lifetime or yield. By focusing search on those quanta of slack expected to maximize survivable failures, called critical quantities, the best possible lifetime-cost and yield-cost trade-offs are efficiently exposed.
520 $a In this thesis we make the following contributions. (1) We are the first to define the idea of critical quantities of slack and use this concept to organize a novel, effective, efficient and scalable slack allocation technique for system-level lifetime optimization, Critical Quantity Slack Allocation (CQSA). In our experiments, CQSA outperformed several comparison approaches to find a wide variety of trade-offs within 1.35% of the optimal while exploring just 1.37% of the design space on average. (2) We are also the first to allocate storage slack to optimize the system lifetime and yield of MPSoCs, preventing system failure when on-chip memories are lost. By allocating storage slack as well as execution slack, in several cases it was possible to double the lifetime or yield improvement possible with execution slack alone. (3) Finally, we are the first to perform application-specific system yield optimization using slack allocation. Applied to yield-cost optimization, CQSA again outperformed a variety of comparison approaches to find a set of trade-offs within 7.34% of the design space while exploring just 2.17% of the design space on average, and further reduced error to 3.54% with just a 19.5% increase in design evaluations.
520 $a System-level design-for-resilience efforts must navigate large, complex, and computationally expensive design spaces to find the best trade-offs. Sophisticated tools are therefore needed that can both effectively as well as efficiently balance lifetime, yield and cost during system-level component selection and interconnect organization. By using critical quantities of slack to direct slack allocation design space exploration, CQSA finds near-optimal designs while exploring a small fraction of the design space, enabling future communication architecture synthesis tools to quickly evaluate candidate architectures in the search for globally Pareto-optimal resilient systems. (Abstract shortened by UMI.)
590 $a School code: 0041.
650 4 $a Engineering, Computer. $3 384375
650 4 $a Engineering, Electronics and Electrical. $3 226981
690 $a 0464
690 $a 0544
710 2 $a Carnegie Mellon University. $3 212563
773 0 $t Dissertation Abstracts International $g 70-09B.
790 $a 0041
791 $a Ph.D.
792 $a 2009
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3376474