Hey Michael,

We met about a year ago at a class that you presented.

I was wondering if you have noticed any movement in the embedded software community to consider multi-core chips and, if so, what OS?

I ask because I noticed ARM has a multi-core chip, Cortex A9, which is rumored to be in the next edition of the iPhone.

Do you see multi-core being valuable, or too expensive for most embedded products? Do you see any problems with going with a multi-core system?

Thanks,
Greg

Dear Greg,

The technology trend that’s driving all the talk about multi-core is that two or more slow processors on one die is becoming more cost effective to produce than a single “faster” processor.

Multi-core processors are thus of most interest toward the end of embedded systems with severe computational demands. Multi-core hardware brings challenges, especially for the software architectures and compilers and operating systems we mostly use. Thus it is unlikely that multi-core will be widely used in the near term by the majority of embedded developers.

Hope this helps!

Cheers,
Mike

With Dr. Dobb’s Journal going printless, the writing is on the (paperless) wall for other trade journals. This prompts me to reflect on the state of technical magazines and their web-based counterparts.

The thickest-ever print edition of Embedded Systems Design magazine (then Embedded Systems Programming) was published in September 2000. That issue had about 120 pages. I was the editor-in-chief at the time, and I distinctly remember the struggle to find enough high quality content for that issue and the impact that doing so had on my backlog of good articles.

In 2000, a full page ad in the magazine sold for about $10,000. And the formula for free-for-registered-subscribers trade journals was simple: publish one page of editorial content for each page of advertising sold. Some quick math shows that the 120 page issue probably generated about $600,000 of revenue. Of course it costs more to print and mail each additional page, but above a certain fixed break-even page count (say, 50 page issues), each additional ad dollar flowed mostly to the bottom line. I think it’s fair to say that the current 40 page print issues of Embedded Systems Design, with ads sold at much lower per page rates, is below that break-even.

By and large, we paid generously for the technical articles and columns we published in that more prosperous era. That incentive to authors plus the editorial funneling process meant that Embedded Systems Programming published lots of high quality articles over the years. Maintaining a reputation for quality content contributed to the publisher’s ability to add qualified subscribers and charge a premium for the ads.

But the online model is different. Because the per-eyeball ad revenue is far lower online, the editorial game has shifted from high quality content to a large quantity content. Even a poorly written technical article with bad advice can earn the publisher some money–if it the editor applies the right keywords.

As Jack Ganssle put it in the above article:

the publication will have more content than ever, no longer restricted by the physical constraints of printing

(To be clear: Embedded Systems Design and its sister website Embedded.com maintain a high standard for content to this day. The preceding comments concern the risks of a move from print to online publishing on any website.)

The move to online publishing is unstoppable, for obvious economic reasons. There are simply more cost effective ways for vendors to reach a large audience and engineers to search for information than the printing and mailing of periodicals.

It is unfortunate that a move toward a higher quantity of lower quality material is becoming the norm within the online editions.