Dave and Intel Investors - Another StrongARM (AS-2) Article
ebnonline.com
Evidently the SA-2 announcement was the buzz of the day at the Embedded Microprocessor Forum today. SA-2:
1) skips directly from .35 to .18 technology, while most other players are going to .25 in this space. going to .18 will allow intel to apply its inherent "economy of scale" to this architecture.
2) .18 will allow some pretty severe performance. 600MHz with power dissipation ~450mW. Also, Intel has designed some of its traditional IA-32 type features into the traditional SA architecture to significantly improve the platforms performance.
3) The move indicates that Intel intends to compete vigorously in the embedded space. StrongARM has been perceived to be languishing under Intel's care since it acquired the technology as part of the DEC patent infringement suit. Intel has now demonstrated it can, has, and will, dedicate resources to this technology.
===================================================================
Intel applies manufacturing muscle to next-generation StrongARM
By Mark Hachman
Electronic Buyers' News
(05/03/99, 06:17:27 PM EDT)
Today, Intel Corp. formally announced its next-generation StrongARM embedded technology, the banner under which Intel will attack the market for portable computing devices.
The StrongARM fills the lowest tier of Intel's segmentation strategy, executives explained. Intel's Pentium III and Celeron lines are designed for high-performance and value-minded PCs, while its Xeon microprocessors are purchased by workstation and server OEMs. “The StrongARM family is the perfect fit for new and emerging systems in the handheld segment,” said Mark Casey, director of marketing for StrongARM processors in Austin, Texas.
The embedded chip will be used to upgrade communications systems that use Intel's older StrongARM SA-110 and SA-1100, as well as lure design wins into portable devices and set-top boxes, where embedded X86 processors and the StrongARM may become competitors.
“The Intel architecture and StrongARM technology will be used by OEMs depending on their focus,” Casey explained. “The StrongARM is designed for the best-of-class digital entertainment experience, with the secondary intent to access the Web. The Intel architecture is designed for Web-centric boxes, designed first for Web connectivity.”
But to date, the pace of Intel's march into the embedded space has been hindered not by the core itself, but by Intel's ability to manufacture it. The StrongARM was originally designed by Digital Equipment Corp., and was sold to Intel together with DEC's Hudson, Mass. fab in 1997. The first-generation StrongARM chips could only be manufactured upon DEC's process, and thus only at the Hudson fab.
Therefore, the StrongARM has remained a notable but second-tier player. Semico Research Corp., Phoenix, estimates that total StrongARM shipments during 1998 were between 500,000 and 700,000 units.
Intel's next-generation StrongARM chips, colloquially known as the SA-2, will be the first StrongARM processor to take advantage of Intel's massive manufacturing capabilities. The core will be the first designed exclusively for the Intel process, according to Casey, and could be theoretically produced in any Intel fab.
The technology will be first described at the Embedded Processor Forum tomorrow in San Jose, hosted by MicroDesign Resources Inc., Sebastopol, Calif. While Intel will not announce any products at the show, company officials will say that the core's speed will range from 150 to 600 MHz, generating 185 to 750 mips. Power consumption will scale from 40 mW at 0.75V to 450 mW at 1.3V, Casey said.
The chips will use a variant of Intel's 0.18-micron process, skipping the 0.25-micron generation altogether. But Intel will not be forced to sacrifice the 0.18-micron capacity the company is ramping now for its desktop microprocessors; the first SA-2 chips will appear in 2000, after the company has converted all of its production lines to 0.18-micron technology.
“The SA-2 will benefit from greatly improved functionality using the 0.18-micron process,” said Tony Massamini, microprocessor analyst for Semico.
Massamini added that while Intel's market share was quite small compared to other players, “it was rather notable that Intel decided to skip the 0.25-micron generation and move straight to 0.18-micron.”
|
