When should EDN highlight new embedded processors relative to when they will be available? As EDN has evolved from a "print-hosted-on-the-Web" publication to a source of complementary print and Web content, the answer to this question may have shifted. During the purely print-centric days, there were page limits. If you saw a product in EDN, there was a high probability that the product was real and available.

A few times I covered a product that was not immediately available at the print date, I received angry emails, some you might call flames, urging me to continue focusing on real products. The feedback essentially was, "Make sure the product is real and available before covering it." One mechanism we use to filter premature announcements is to insist on pricing; this works to a certain extent. Freescale's announcement for a lower price PowerQUICC processor with Gigabit Ethernet connectivity and Texas Instruments' announcement of new Davinci digital video processors are two examples. I'll go into more detail on these below. I'm looking for your feedback as to when you want to hear about embedded processors relative to their general availability. Your answer will influence when processors are announced in the future.

I have observed that the timing between when embedded-processor manufacturers announce their processors and when those devices are available for production use varies in a cyclical manner. On one end of the scale, the manufacturers announce their processors more than 18 months in advance of when the first device will be available. On the other end, companies announce their processors only when they can deliver the silicon in production quantities and all of the software-development tools.

Most of the time, embedded-processor companies time their product announcements closer to when the processor, and usually the development tools, is available with production support. In contrast, the timing for general-purpose microprocessors that will end up in desktops, notebooks, and servers tends towards a longer lead-time between the announcement date and when the production-quantity availability.

The challenge is when the time between an embedded processor's announce date and its general availability increases beyond one or two months. When this happens, it seems to signal the start of increasing time intervals between announce and availability dates. In past cycles, the distance between the announce date and device availability quickly jumps from two to six, then nine, then 12 or more months. This process feeds on itself, with an early-adopter announcement from one company spurring even earlier adopter announcements from other companies. We seem to be entering another such cycle.

Freescale's PowerQUICC announcement leads general sampling by at least two months and production availability by at least five months (second quarter next year). Initial samples are available today, but the designers who will receive the initial samples are usually in early-adopter partnerships with Freescale. These designers are not finding out about these parts through early announcements.

TIs' Davinci announcement is similar, but the early nature of the announcement is subtle. The devices are sampling today, and production volumes are scheduled for second quarter next year. A telltale sign that this is an early-adopter announcement is that the software development tools for these devices, which will leverage the Davinci ecosystem, will not be out until the second quarter next year. Software developers should be able to use the existing C64x+ tools, but they will not be able to immediately take advantage of the Davinci ecosystem. The users of the first samples will be early adopters that will be working closely with TIs' pre-production support infrastructure.

In past announcement-availability cycles, coincident with my discussions with companies when they were announcing earlier and earlier in their design cycle, the early adopter announcements seemed to stop and the market returned to a shorter announcement-general availability lead-time. The most common answer I receive when I ask why companies are doing this is that it enables them to maintain leadership appearances against their competitors; they feel the market will perceive them as standing still, or worse, falling behind, if they do not match capability announcements with their competitors.

The majority of early adopters that will use these devices already know about them before the early-adopter announcements. I wrote a few articles a while back about early adoption. "Forge ahead" was a hands-on project that had working hardware, but the software I had to work with was hot off the bench—bugs and all. "Welcome to the jungle" addresses the risks, benefits, and considerations of being an early adopter in embedded designs.

Early adoption of a processor may make the hardware integration easier, or it may push the processing performance to enable new capabilities, but all of that usually comes at a price of more complexity for the software team. So my question to you is, how much lead time is too much? The metric I have used to filter early-adopter announcements has been that the announcement should not lead general (not initial) sampling by more than two months, and if the software-development tools will not be ready at that time, I mention it in the write-up. Should I change this now that the Web affords us more room to present this information?

This is your chance to meaningfully influence how and what I filter for you, and it is a chance to let processor manufacturers know if you appreciate these early announcements (and how early) versus announcements that are timed closer with production support. It's also an opportunity to express, in general, your satisfaction or concern about processor manufacturer's understanding and ability to provide full production support to both the hardware and the software teams that ultimately make the processor do something useful.