
EDN Europe's Editor Graham Prophet posts a selection of comments and insights prompted by the many items of industry news and rumour that cross the editorial desk or are gathered on his frequent travels to interviews, press conferences and events around Europe - and further afield - and somehow never find their way to the
magazine or the web site, recovering some of the information otherwise lost in the noise level...
Tuesday, February 17, 2009
Faith in the future
If you needed any confirmation of the way that the mobile sector now carries the hopes of a large section of the industry, look no further than the announcement yesterday (16th February) from ARM (see news section). That ARM IP should make its way to 32-nm processes is hardly a surprise. And where does the announcement of such a step forward appear? At Mobile World Congress, naturally. Why “naturally”?
One of the recurring themes in any examination of continuing progress in silicon technology is the financial underpinning of any products designed in ever-smaller-geometry rules. The last set of PowerPoint slides I saw on the subject argued that, by the time you take into account design costs, NRE (non-recurring-engineering) costs, mask costs, minimum order quantities from the silicon foundry, bills for coffee and Diet Coke for everyone involved, and so on, then in order to support the design of a mega-scale SoC, you need a market for the end-product of, at least, hundreds of millions of dollars (or euros).
Therefore, not many markets can support SoC designs in the newest processes, and that’s a trend that only gets worse as generations go by.
So MWC is not only the natural vehicle to show a 32/28-nm demonstrator chip – it’s very nearly the only one.
4G-mobile comms may only work if you have 32-nm silicon. 32-nm silicon will have few viable users other than 4-G mobile comms….. OK, that’s an exaggeration, but does anything about this sound precarious to you, given the parlous stgate of the global economy?
The ARM press release, by the way, describes the fabrication of the Cortex core on a 32-nm vehicle as “successful” but it doesn’t actually say that the processor itself works, or give even a hint of any metrics about just how much 32-nm buys in terms of speed/power gains…. Surely an oversight.
Post a comment
Note: fields with an asterisk(*) are required information.
All submissions are subject to review before they are posted live.