Sony & Toshiba Disclose Cell Fab Plans

sean23007 writes "InfoWorld is running an article about Sony and Toshiba's plans for new fabrication plants to build the 'Cell' chip jointly developed by Sony, Toshiba, and IBM for use in the Playstation 3 and other home entertainment uses. The new fabs will be located in Nagasaki and Oita, and both companies plan to spend $1.7 billion over the next 3-4 years in their construction. They will be capable of using 300 mm wafers with a 65 nm process. The chip is slated to be the first 1 teraflop consumer device."

Where's IBM in all this?

[Ra5pu7in]

Why are only Sony and Toshiba reported? Does IBM not plan to produce and Cell chips or do they already have fab facilities?

Re:2007 Then

[sigep_ohio]

i don't think microsoft will wait until 2007 to release Xbox2. Nor will nintendo wait that long to replace the gamecube. I would say that cell will not be in the PS3. More likely sony will put say 2 emotion chips in the same box and in a way double the processing power of the current PS2. this would be a good stop gap before the Cell processor is ready.

Re:it looks like scea will lose their way

[sigep_ohio]

I always got the impression that Sony wanted to go the integrated home electronics route way before MS entered the game. I mean the original PS was always able to play music CD's if i recall correctly. Then PS2 was in developement before MS ever got in the video game business. So many of its capabilities have been in since the begining I am sure.

Plus, Sony would love to be able to cheaply manufacture and sell at their usually high markups, one box that can be a PVR(whose abilities they would control), dvd/cd player, video game system, and maybe even throw in a Home theater capability. This seems to be both Sony and MS's ultimate goal, along with subscription services for games and other online activities.

Re:cost justification

[sigep_ohio]

Well with the number of PS2's that have sold so far worldwide, I guess Sony thinks the demand for Cell will be even bigger. I know Cell won't be in the PS3, but since it is a scalar architecture it can be placed in anything from cell phones to video game machines to computers. This makes the market for Cell even larger than that of a Pentium or PowerPC or The Emotion Engine in PS2.

Re:Wowza

[Blaine Hilton]

Also, I should clarify, not just cell phones, but all electronics should have better methods of recycling them. When you went out to buy that PS2, what happened with the PS died? Circuit boards have many elements that should not be in a standard landfill.

Re:cost justification

[watzinaneihm]

Atleast some part of the decision may be due to the culture of Sony
Sony (the original one, not their music division) have always been about making a product and then finding a market for it. They pride themselves on being the lone-wolves and hence came up with a lot of products which are completely non-standard (unless the world accepted their standards)the oft used betaMax being a good candidate. Sony's memory products today are a good example, they are tiny but are not interchangeable with others (not all prodcuts, but there are examples like their USB cables etc.) .A completely new chip for a game machine seems to fit right in, compared to MicroSoft moving to Celerons.
But then again they used to be able to come up with a product so good that they opened up whole new markets.
This post is partially based on a reading of "Made in Japan" by Akio morita, and I understand that decisions of Billions are not always decided by the "culture" of a company.

Re:cost justification

[Enigma2175]

Usually, with initial chip production at this scale, they lease out fab production time from other companies.

True, but what fab has the capabilities to produce the 65nm parts that they require for the chip? AFAIK, there aren't any current fabs that could produce the chips they want (at least not in volume).

That beats Intels 90nm fab out of the water

[mnmn]

And so does including memory and GPU with the CPU. You could imagine the desktop computer market for these babies.... if released with Linux.

Thinking of that, I wonder if they would allow hookup to a DVI connector, or replacing the BIOS, or adding PCI/ISA slots, or even producing whole chips for third party taiwanese boards that would then be built into workstations. If the chip is up to the spec, on time and reasonably priced, theres already a big Linux-based market for it, meself included. Saddams gotta steal only a FEW of these to build nukes. Wonder if an anarchist teen could do that with this christmas present.

Re:2007 Then

[GweeDo]

This would be one of the worst things Sony could imagine doing. The EE (emotion engine) is already massively parallel and a pain the butt to program for since you have to have so many things synced. Double that you are asking to kill off programmers as they try and work. The Gamecube and the Xbox have proven to developers that it can be easy to develop for a console as well as be powerful. The design behind the PS2 was bad bad bad...Sony had 2 generations on the top...the PS3 might be their N64....

Re:cost justification

[Anonymous Coward]

I find it hard to believe that you know anything about the economics of the semi-industry. Semicondutor manufacturers build a Fab to support a process, not a product. The need to expand the manufacturing capacity for a particular process might be predicated on the demand for a family of devices. That is not usually the case with Fabs designed for new processes. And not the case here. Just because the new processes will be needed for the viable production of the Cell chip, which potentialy will be a big money maker, does not mean that the plants will not be used for other devises. Farthermore, Japanese companies tend to take a lot longer view of things and are not put off by short term economic sluggishness. Unlike manufacturers in the US, who clench their collective A-holes at the slightest murmerings of uneducated Wallstreet analists. Fab capacity must be catagorized by the processes the Fabs support. After all one usualy does not use Dachshunds to pull dogsleds, or Pitbulls to herd sheep ( though the latter has been done). The Fab capacity for a new process is obviously going to be limited.

BTW, 15+ years experience in Semiconducor Manufacturing in many different capacities including capacity managment.