Part of the  

Chip Design Magazine

  Network

About  |  Contact

Archive for April, 2009

Low Power Design Reverses Outsourcing Trend

Thursday, April 30th, 2009

Designing low-power products requires a greatly customized design which, in turn, requires a great deal of in-house engineering talent and less sharing with external suppliers.

More semiconductor companies are realizing that building products (chips-packages-boards) that consume less power is a market differentiator. But designing low power products isn’t easy. It requires a level of customization that – judging from current business trends – is not feasible in today’s outsourcing model.

One example is Apple’s recent strategy shift away from outsourcing to favoring the cultivation of in-house talent. According to a Wall Street Journal article, Apple has been hiring engineers to design -  really, customize – their multicore cell phone chips. By increasing their in-house design expertise rather than outsourcing it overseas, Apple hopes to create exclusive features through highly customized chip and board sets.

Apple’s strategy shift exemplifies the growing importance of customization.  Low power design requires a high degree of customization in order to meet ever shrinking power budgets that are married with ever increasing product feature sets. In this way, low power design is much like analog-RF design. Another similarity between the two is that customization is not confined to merely chip design. Rather, customization must occur across all levels of product design, including all life cycle phases as well as between domains – from chip to package to board to module to subsystem to final product.

In addition to building up in-house expertise, Apple has made other signs that support the goal of customization. Last April’08, Apple acquired PA Semi,  a key element to controlling power design for the iPhone. Another sign is Apple’s recent hire of the ADMs graphic division CTO, Raja Koduri. All of these trends mean that Apple will be capable of customizing the power, graphics and other key chips in a way that few competitors can match.

These moves toward increased customization capabilities have been mirrored by others in the industry. For example, Intel is pushing hard with its Atom low power embedded processor and recent ventures into the graphics world. ARM, certainly the low power leader in the cellular market, now offer a low power embedded processor for the growing nettop market. Even nVidia, a traditional graphic chip company, is considering building low power products around the x86 architecture.

Customization is a cyclical process (see Makimoto’s Wave). For low-power, it’s not enough to design low power chips. Companies must design low-power process across a number of domains. While not all EDA or semiconductor vendors realize this trend, many do [Chip Designers Scramble For Low-Power Solutions]

Companies that do not will be left “running out of power.”

Power and Placement – New Meanings in Green

Wednesday, April 29th, 2009

Sustainable electronic design – often collected under the general title of green electronics – has forced a focused on low power and life-cycle development in a way that has not been experienced since the days of the space race….

Do Readers Care Who Pays the Editor?

Friday, April 24th, 2009

Hi all. Sorry for the dry spell in my blogging. Sometime life intercedes and sidetracks our well-heeled routines.

Today I’d like to consider the question of recently displaced editors selling their brand w/o full disclosure. I know how the editorial community feels about such actions, but do readers really care?

Here the problem: What’s an industry-recognized and -respected editor to do once he/she is laid-off by a major publisher? There seems to be one of three career choices:
> Become a freelancer
> Become an editor for a corporation, i.e., EiC for a corporate magazine and/or online site.
> Become an editor for a corporation but give the appearance of still being an independent journalist.

It’s the later case that concerns me. Aside from ethical questions and eventual damage to the editor’s brand, do the readers really care? Can anyone cite examples of the effects on less than full disclosure to the readers (not the editorial community)? Appreciate your thoughts.

Power Trump’s Time-to-Market as Main Driver

Tuesday, April 14th, 2009

Learning about low power design takes a lot of energy. Over the last two months, I’ve interviewed chip companies, IP vendors, EDA suppliers, power organizations, standard bodies and even software development firms.

Each of these groups have a different perspective on the low power problem. Their solutions attest to the range and variety of these perceptions. Yet they also share a common understanding about the changing landscape for electronic products, namely, that power efficiency is now a critical part of the power budget. Let me explain.

In the past, architects would divide up available power – say, battery capacity for a given usage rate – and then allocate a portion of that total power (minus a small reserve) to each block in the chip or board design.

However, today’s power budgets come with an additional caveat: each block is expected to provide a power efficiency improvement as a way to reduce previous power levels. This mandate for efficiency is needed to offset the diverging rise in feature sets with a lack of improvement in battery technology. (See “Chip Designers Scramble For Low Power Solutions,” in the April 15th Low Power Design e-letter.)

This insistence on power efficiency has forced chip block designers to accelerate their collaboration with both board-level designers and software developers (device drivers, RTOS, OS and applications).

In the past, Time-to-Market (TTM) considerations have been the big driver for system-level awareness. TTM goals have been the main reason for collaboration between hardware chip-package-board and software design teams. But this has changed. While TTM is still important, power efficiency has eclipsed it as the more critical design driver.