Free Seminar/Webinar on GREEN RELIABILITY

 

Ops A La Carte, SinoRel, the IEEE Reliability Chapter, and the ASQ Reliability Division will be hosting a FREE seminar/webinar on GREEN RELIABILITY – some of the current trends as well as some of the issues and concerns around reliability in this rapidly expanding market. We will have the following presentations at the event.

Dr. Cheemin BoLinn, Peritus Partners: Getting the Bang – Green Agile Solutions Designed for Eco- Environmental Impact!

As significant investments in green technology continue, there are immense opportunities to demonstrate leadership in designing reliable green technologies and products. These green solutions can fulfill the need for increased power with a smaller or net zero carbon footprint.

However, robust reliability and maintenance practices will need to continue to be rapidly deployed.This focus will be required throughout the product life cycle of these green solutions, from concept design to prototype to manufacturing and throughout the ecosystem including supply chain and disposal. Even the traditional perspective of a data center has already given way to a broader ecosystem view that focuses on designing reliable energy efficiency not only within, but also among the data center, utilities, facilities, and surrounding environments. To receive optimal benefits, the "best practice" tools and processes utilized in other industries plus new agile innovations will be required for this new clean tech industry. These innovative solutions will be the bellwether that demonstrates new energy efficiency measures will not cost reduced reliability.

Green solutions, proactively designed from concept stage, with the fundamental precept of reliability and energy conservation and reuse, have a strong business value proposition. Green reliability drives increased profits which can be a new funding source for increased R&D to fuel continuous improvement, critical for sustainable product development. Green reliable solutions have a compelling value proposition not only for the environment but for the profitable growth of businesses.

This session will examine:

1) Trends and directions in the clean tech industry for green, reliable solutions
2) Hotbed investment areas
3) Ecosystem view of energy efficiency
4) Business and environmental impact of green reliability and efficiencies
5) Value proposition for green "carbon neutral" technical designs
6) "Greening" as an economic growth driver and implications for marketing

Alan Wood, Sun MIcrosystems:The End of Redundancy – alternative methods for achieving high reliability

Minimizing power consumption has become a very important topic in the design of components, computers, and data centers. Industry analysts predict that, within the next few years, the cost of power will surpass the cost of compute equipment in the data center. Dependability research has not traditionally considered the cost or availability of power to implement the proposed techniques. If redundant equipment was needed for fault-tolerance, it was blithely assumed that the power and other facilities overhead was negligible. That assumption is no longer valid, and dependability research and practice needs to change accordingly.

Many classical fault-tolerance techniques, such as voting, assume redundant compute resources. Having redundant hardware to derive duplicate results minimizes single point failures but is unlikely to be popular in a power-constrained data center. As an example, the IBM Z-series microprocessor has moved from duplication to error checking and sparing to save power. This type of dependability design shift is likely to accelerate. Active redundancy will not be affordable in many data centers. Even passive redundancy may be questioned because it takes space, and space usually equates to power, especially if the "sleep" mode for the passive components does not do a better job controlling leakage current.

At the chip level, the drive to lower voltages and power along with other factors will make logic more vulnerable to soft errors. Thus, the error rate is likely to increase as the redundancy decreases, making dependable computing a greater challenge. Techniques that detect/correct errors with minimal power consumption, such as information redundancy and self-checking logic together with some form of retry mechanism, will be favored. Power impact analyses of the proposed dependability technique may be come more important than the traditional analyses of the performance and chip area impact.

At the server level, it will be important to find different architectures and software approaches for dependability in the absence of hardware redundancy. It may be possible to consider trading recovery time for power. As flash memory is used as a partial DRAM replacement in some applications, it may be possible to consider storing checkpoints or other data in flash memory to improve dependability. More than ever, dependability research and practice will need to focus on the entire system to devise methods that do not require additional power.

At the data center level, operations may impact dependability. Equipment may be turned off to save power, causing more power and temperature cycling. Equipment with a sleep mode that uses minimal power but can quickly be activated when new tasks arrive will be valuable. Functions such as automatically throttling power when not needed and providing real-time information on power utilization will also be popular. Data center cooling may actually improve as power-saving features such as hot-aisle containment that reduce hot spots are implemented.

The dependability community is faced with the challenge of using less power while maintaining the same levels of dependability. Existing assumptions need to be altered, and new research is needed.

Subhasish Mitra, Stanford University:Globally Optimized Robust System Design

Most future systems, not only high-end mainframes and safety-critical systems, must be designed such that they can cope with imperfections in the underlying hardware. In contrast, today’s design methodologies assume perfect hardware during useful system lifetime. Classical redundancy techniques are too expensive and inadequate for this purpose.

This talk will present an overview of techniques that can enable a sea of change in robust system design:

1. Built-In Soft Error Resilience (BISER) for correcting radiation-induced soft errors;

2. Circuit Failure Prediction, together with on-line self-test, to overcome early-life failures and transistor aging;

3. Effective test and post-silicon validation techniques.

Opportunities for global optimization across multiple abstraction layers will also be discussed.

Bryan Stallard, Ops A La Carte:EDA Usage in Reliability Aspects of Green Technology

The use of EDA in GREEN situations, specifically in lowering energy consumption, is a natural extension of long-existing usage in two related contexts: thermal management in complex systems, and portable-device battery-limited usage. For years semiconductor companies, and their EDA suppliers, have battled power dissipation issues, driven by the twin facts that MOS ICs burn power proportional to clock frequencies (ignoring minimal idling currents), and that as majority-carrier devices, the MOSFET elements have channel resistances that increase with the absolute temperature, ultimately increasing RC time constants until they limit both raw speed and race-condition margins.

At board levels, the issue bifurcates between commercial usages, where convective cooling is dominant, and military/aerospace use where conductive cooling is primary. Both high-altitude and space applications essentially discount convection completely. In neither
group is there any appreciation of excessive power dissipation.

In portable devices, the ability to use a product without continuous recharging has long been a primary selling point for premium offerors, and a source of irritation for generic knock-off users. So, mobile devices are already biased (sic) toward energy efficiency.
Three topics are worth considering, or revisiting, in a GREEN light:
1) Conversion of existing circuitry to emphasize or improve energy-consumption performance;
2) Modification of production designs to meet RoHS or other material issues;
3) Development of ongoing engineering practices to institute or upgrade Green features to a level parallel with established customer "-ility" expectations (reliability, quality, availability, maintainability).

Email us at via our contact form for more information on any of these events above. Or just Email to office@sinorel.com.