Intel does u-turn on Willamette and synchronous DRAM

Updated Sources close to Intel's plans have confirmed that the firm is readying a backup plan which will mean that its Willamette IA-32 processor, expected to debut the end of this year, will support synchronous memory as well as Rambus.

German magazine PC Welt has now published details of the Armador chipset, which you can find at this URL, and which, it says, is part of a backup plan. A diagram is also included in the article, which is short enough to be Babelfished.

The secret chipset project is intended to guard against Intel being caught by a suspected shortage of Rambus RIMM memories, and will allow it to launch to its correct timescale and save some face at the currently beleaguered microprocessor firm.

When we first heard of this information, rumours of which have circulated within the industry for some weeks, we contacted our own sources at PC companies who confirmed to us such a plan is in the offing. Just because Intel produces roadmaps and designs, however, does not necessarily mean the firm will release them. It changes its corporate mind all the time.

Nevertheless, the Armador project, something of a volte face for the firm, is likely to irritate Intel's partner Rambus, which managed to secure a promise from Dr Albert Yu at the firm's February Developer Forum and which suggested that Rambus and Willamette were inseparable bedmates.

Intel has always said that Willamette's elder brother, the server/workstation IA-32 chip known as Foster, will use double data rate (DDR) memory - and that caused many to scratch their heads in disbelief that essentially similar chips would use a different, and much cheaper, type of memory than Rambus.

Willamette will also be supported by a Rambus motherboard codenamed Garibaldi at launch time. The current chipset slated for the product is Tehema. But PC Welt says an SDRAM chipset design exists, at least on paper.

On the day that Yu demoed the Willamette microprocessor, expected to debut at 1.4GHz, his announcement caused Rambus share prices on Wall Street to soar.

Later this day, a German title, PC Welt, is expected to publish documents proving the existence of an SDRAM chipset for Willamette.

An Intel spokesman said that it was "speculation" that the firm was developing SDRAM (and DDR) for Willamette. That doesn't mean it isn't true, however. Remember, too, that Intel was denying to The Register for weeks that there was an i820 recall. And, last year, in the face of overwhelming evidence, Intel denied for a full five months that it would introduce a PC-133 chipset. It did.

No doubt the truth will eventually out, despite the best efforts of Intel's global spinolas. ®