Toshiba joins IBM's 32 nm pact, but may not bring Cell along

Already a partner with IBM and Sony in the development and production of the Cell BE processor that powers the PS3, Toshiba this morning stated it's joining IBM and its partners on HK+MG. But IBM says this is a different track.

Last week, IBM unveiled an expanded industry alliance that includes CPU manufacturer AMD, Chartered, Freescale, Infineon, Samsung, and new partner STMicroelectronics. Their objective is to pool their resources to make 32 nm CMOS chip production viable, using the high-k-plus-metal-gate technology that IBM raced against arch-rival to discover.

This morning, Toshiba enters the picture as a full partner in the new venture, which IBM said this morning will be devoted to perfecting 32 nm "bulk" (industry standard) CMOS lithography processes. Their development activities will now be centered at IBM's 300 mm fabrication facility in East Fishkill, New York.

Toshiba is no stranger to alliances with IBM. In fact, the two are close partners in the STI alliance which produces the Cell BE processor, along with Sony. It's Sony's PlayStation 3 which is by far the most visible production vehicle for the Cell CPU to date. Sony has already announced its plan to pull out of the production phase of the partnership, selling a major production plant to Toshiba, and leaving it and IBM to manage that part of the project from here on.

So will Toshiba's entry into the 32 nm alliance mean an accelerated ramp for possible future incarnations of the Cell? Today, BetaNews received responses from two different IBM spokespersons: One declined to respond citing lack of information, indicating an indefinite maybe. The other gave BetaNews a definite no.

The CMOS bulk technology track, the other IBM spokesperson told us, is completely separate from the Cell BE track, whose manufacturing process is currently SOI-based (silicon-on-insulator).

But last June in Kyoto, Toshiba showed an interest in moving the Cell to a bulk manufacturing process, demonstrating at a symposium in Kyoto that it had successfully created a variant of the Cell's SPE engine -- its principal core -- using 65 nm CMOS bulk. Such a process might not only make the Cell less expensive to produce, as Toshiba demonstrated then, but might actually give the company a reason to install it in some of its own products.

Doing that might help the Cell -- one of the most promising and innovative CPU designs ever to come even partly from IBM -- to rely less and less upon the PS3 for its success. Still, our second spokesperson was adamant that the Cell is not on the table for this latest venture.