Sony Plans Smaller PS2 Chip, Cell Manufacturing

Thanks to Reuters for their story discussing Sony's plans to start mass production on a smaller PlayStation 2 chip. The new chip is "using cutting-edge 90-nanometer processing technology", and monthly production of the chip, which is functionally identical, and "...combines the game console's microprocessor and graphics chip, will start at 'several' hundred thousand units before growing to more than a million units by next year." They also plan to start a test plant for their next-gen Cell processor, widely rumored as the basis for the PlayStation 3, and the Sony plants are "...expected to start mass production of 'Cell' in the second-half of 2005", hopefully far enough ahead of time to avoid the chip shortages that plagued the PS2 launch.

Comments

Steps to the PSP...

[H0NGK0NGPH00EY]

Sounds like there may be something to their claims that the PSP will have the power of a PS2, after all. I would say that combining the microprocessor and graphics chip into one smaller chip is a pretty good step toward fitting the whole thing into a handheld device, yes?

Re:Steps to the PSP...

[addaon]

There's something even more important, though. Originally, the PS1 (PSX has become an overloaded term) was a typical system; a bunch of chips tied together on a circuit board. Over the lifetime of the system they reduced the component count significantly; the final revision of the PS1 had the entire system (well, except for standard external stuff) on a single chip.

This was nice; it let Sony cut down the console costs and increase the profit margin. But it allowed something even better. It allowed Sony to include an entire PS1 inside the PS2, for the cost of one (relatively small) chip. Now that the PS2 (which contains a PS1...) is on a single chip, it becomes feasible to put an entire PS2 inside the PS3, even though the architectures are (presumably) very different. It is literally two (or, I guess, three) systems in a single box; multiple CPU's, of which only one is active at a given time.

This leads, of course, to backwards compatibility which, in my opinion, is the most important think Sony has over the competion.

Why not 0.13um????

[pagercam2]

Going to smaller geometries is a standard tactic, so this is surprising but the more supprising that they intend to skip 0.13um which is still leading edge, .09um is still pretty risky and expensive. 0.13um would have cut the die size in half, while .09 reduces by 75% which is better but the reliability of .09 and the extra effort to go to the latest process when you need 20 million parts/year this sounds like a major risk that shouldn't be allowed to limit the move of the PS2 marketing push. I find it somewhat strange that "Cell" is getting so much press and interest, nobody has built a chip at .065um yet and this may take years if they can fit 8 processors at 65nm they should be prototyping 4-6 processors at .09 or 2-4 at .13. The cell seems to be more of a marketing ploy rather that real technology or atleast that all thats getting out. Does a game system need 1 trillion ops/sec???? for game play? A workstation with this capability would make a lot more money, your own nuclear resarch on PS3????

Whats so new about the Cell processor?

[mnmn]

I have seen so much hype and no numbers. So what if each Cell is an overclocked ARM7TDMI and they use 16 of them in the box? I am more interested in how the memory is connected to each of the cells (northbridge), the memory specs (surely not the PC133), and how much of the OpenGL pipeline is offloaded further from the GPU to the chips. This is after all what they claim theyre doing.

I've seen dual and quad ARM cores in one chip, seen a simple GPU with an ARM in one chip and an ARM chip with a large FPGA on board. These are real production objects being used in cell phones and the likes. Everyone knows a dual Pentium2 500MHz doesnt necessarily beat a single Pentium3 at 1GHz, so how does this translate into cool gaming?