Toshiba, Sony close to 65nm sample production

[passerby]

Sony is definitely not waiting for PS2 to wind down first

Which reminds me that even the PSOne has not reached official end-of-life yet. With this as reference, PS2 will continue selling for quite some more time. No point trying to wait too long...

 

[Panajev2001a]

Since CELL is taped out, its logic is locked down - can't afford to sit on your hands while traditional GPU's improve.

Add more APUs, add more PEs, add more e-DRAM, increase clock-rate... the path to Figure 6 of the patent is not short or free of bumps.

 

[Brimstone]

I don't think anyone is suggesting that Sony is going to let their fabs be idle until the PS3 is ready. I say again the PSP and PS2 can probably absorb a lot of capacity. A version of CELL may start to get into production on .65 nm, but what products will it go into? If I remember correctly, it has been the mantra of many that CELL will find its way into a wide range of Sony products. High-end servers with several CELL cpu's running on a Linux operating system comes across as a more likely canidate than a PS3. Selling high end servers could consume the XDR Ram that supposed to come into production. The XDR Ram is going to go into more than the PS3. Networking equipment comes to mind.

I'm sure Sony can release a PS3 in 2005 if they wanted to. If Sony wasn't obliterating the competion like it is right now, I could see Sony being more eager to put something new out. If Sony really wants to pull off Blu-ray as a media standard, including it in a PS3 would help. The later the release date on the PS3, the cheaper the technology should get.

Moores law applys to Sony just like the PC industry. A 2007 PS3 could have .45 nm with SOI edram. That should allow for a much higher clock speed than a chip with edram not including the Toshiba SOI technology. The XDR Ram will also get produced on a smaller process allowing for a higher density and faster operational speed. Also the extra time will allow for better development software and tools for game developers.

 

[Paul]

A 2007 PS3 could have .45 nm with SOI edram. That should allow for a much higher clock speed than a chip with edram not including the Toshiba SOI technology. The XDR Ram will also get produced on a smaller process allowing for a higher density and faster operational speed. Also the extra time will allow for better development software and tools for game developers.

It's just not happening dude.

 

[cthellis42]

No one is waiting that long--it would be a bit suicidal to give multiple competitors THAT much launch advantage, not to mention game development priorities. Maybe if the PS2 has its current marketshare advantage along with Xbox's technical superiority they would have chosen to wait longer to ride the profits and press for maximum technical advantage, but if they wait it puts not one but two generations from their competitors looking better in comparison. Unless they were assured the the PS3 would cause a blowout of apocalyptic proportions, the longer they wait, the more ground they would be yielding to their competition rather than pressing their advantage.

Meanwhile, they'll likely be creating a PSTwo around the time the PS3 is launched, and will still have companies riding their old systems for quite some time anyway, so it's not like the profits from there would instantly dry up.

 

[Panajev2001a]

No one is waiting that long--it would be a bit suicidal to give multiple competitors THAT much launch advantage, not to mention game development priorities. Maybe if the PS2 has its current marketshare advantage along with Xbox's technical superiority they would have chosen to wait longer to ride the profits and press for maximum technical advantage, but if they wait it puts not one but two generations from their competitors looking better in comparison. Unless they were assured the the PS3 would cause a blowout of apocalyptic proportions, the longer they wait, the more ground they would be yielding to their competition rather than pressing their advantage.

Meanwhile, they'll likely be creating a PSTwo around the time the PS3 is launched, and will still have companies riding their old systems for quite some time anyway, so it's not like the profits from there would instantly dry up.

if Xbox launches by middle to late 2005 ( possible ), then a late 2006 launch for PlayStation 3 would not be impossible to conceive and would work out well.

PSP would get 2 nice years of spotlight and PlayStation 2 would be allowed to make tons of profits ( important ) during 2004 and 2005 while expanding their online strategy.

 

[cthellis42]

Late-2005 to early-2006 is pretty much when we're all expecting to see the next generation coming out. Just when is a question mark, as timetables can always extend and problems can always arise to push back launch. But planning on launching in 2007 (which in itself could be knocked back further from process problems) while both competitors would have at least a year's advantage...? Over a system already 1-2 years older than them from THIS generation?

'sides, if the scalability and design of CELL works as they want it to, it would be possible to move to .45 and adopt process advantages when they are solid to decrease costs, lower power output, maybe even kick up the performance somewhat... <shrugs> They're obviously not going to stop improving their processes, but there doesn't seem to be any overriding necessity to key PS3 to .45 over .65 while losing some of their momentum.

 

[Super Grafx]

it seems to me that a fall 2006 U.S. launch is the most likely.

fall 2005 would be pushing it, since the creation of complex games (graphically at least) on brand new hardware will take TIME.

on the other hand, a 2007 release seems too far off
(4 years!) and that I would concider a delay, but not 2006.

 

[London Geezer]

With all honesty, and i've said this before, if PS3 is delayed, it will be because Sony wants to wait until libraries/tools are ready. And they have 2 choices here, they can either do a PS2 and release PS3 without proper instructions to evelopers on how to work with the hardware, OR they could wait there until they know developers find it relatively easy to develop for the machine, in the meantime upgrading the machine as time goes by.

I'd go for a middle option really...

 

[cthellis42]

Still truckin' along, Toshiba is. Short mention on the Inq:

A REPORT SAID that Toshiba has beaten Intel by making transistors at 45 nanometers which use strained silicon.

According to the Nikkei Business Daily today, Toshiba has included the technique in transistors it is making at 45 nanometers, while Intel is currently restricting the technique to its next generation 90 nanometer process.

The report said that the p and n CMOS transistors use gate electrodes made using the strained silicon technique, and that means they switch over 10 per cent faster than non-strained silicon devices. µ

Sadly Nikkei Business Daily requires credit card registration...

 

[McFly]

I think the whole cell technology is perfect for scaling up after launch, so I would not wonder if the dev tools are designed so that the game-devs have to create games that can make better gfx with a better PS3 or added cell modules. I think that's one of the main ideas of cell technology.

That said, I think a late 2005 65nm PS3 and then one year later a better 45nm PS3 or/and module would be possible and it would give the other next gen consoles a very hard time.

Fredi

 

[V3]

More of the same...

Toshiba's embedded-DRAM pact to yield 65-nm technology

By Yoshiko Hara

EE Times
December 15, 2003 (8:09 p.m. ET)


TOKYO — Toshiba Corp. will begin offering by March engineering samples of a 65-nanometer embedded-DRAM technology jointly developed with Sony Corp. The two companies detailed the technology-which involves the smallest DRAM cell size ever reported, 0.11 square micron-at the International Electron Devices Meeting in Washington last week.

Their 32-Mbit memory device is fabricated on a 65-nm process called CMOS5 using six-layer copper wiring with low-k materials. "We are confident that our embedded-DRAM technology takes the lead by about two years among competitors," said Seiji Yamada, chief specialist in the Advanced CMOS Technology Group II at Toshiba's System LSI division.

Toshiba and Sony announced the key technologies-a transistor with a 30-nm gate, the embedded DRAM cell and SRAM cell, and multiple-layer wiring-for the 65-nm process last year. This year the team has come up with a practically operating device.

When DRAM cells shrink, it becomes essential to control the transistors' threshold voltage. To realize a high-performance DRAM cell, the threshold voltage should be maintained high and steady. That is realized by dense ion implantation at the transistor section, but at the same time, the dense ion implants cause leakage of electric charge at the memory capacitor, which deteriorates data retention.

The Toshiba and Sony team resolved this issue by means of a tilted implantation method that inserts a boron ion at a certain angle over the gates. Ions are implanted more densely on both sides of narrow gates; the opposite occurs with wide gates. This stabilizes the resulting threshold voltage, making it "almost flat," said Yoshinori Matsubara, a specialist in Toshiba's Advanced CMOS Technology Group II.

The method also serves to block ions from the memory capacitor section, the companies said. Thus, the team claimed that even the small, 0.11-square-micron cell has adequate threshold voltage and data retention.

Toshiba and Sony have been collaborating on 90-nm and 65-nm process technology development since April 2001. The second development phase, 65-nm process technology, is scheduled for completion in April.

Toshiba plans to begin sampling the 65-nm embedded DRAM in the first quarter, but Sony has not disclosed its fabrication plans. The company is now installing production equipment for 300-mm wafer lines at Nagasaki. Sony will use devices built on the CMOS5 process for its own products first, before selling on the merchant market.

 

[Gubbi]

That's pretty good. That means that the 64MB eDRAM of CELL will only take up 60mm^2.

Cheers
Gubbi

 

[Panajev2001a]

Still truckin' along, Toshiba is. Short mention on the Inq:

A REPORT SAID that Toshiba has beaten Intel by making transistors at 45 nanometers which use strained silicon.

According to the Nikkei Business Daily today, Toshiba has included the technique in transistors it is making at 45 nanometers, while Intel is currently restricting the technique to its next generation 90 nanometer process.

The report said that the p and n CMOS transistors use gate electrodes made using the strained silicon technique, and that means they switch over 10 per cent faster than non-strained silicon devices. µ

Sadly Nikkei Business Daily requires credit card registration...

Strained Silicon + SOI + e-DRAM without capacitor == good 45 nm node for Toshiba and Sony.

http://www.eetimes.com/story/OEG20021111S0024

This article talks about Strained Silicon + SOI ( Toshiba is not in this article, but Toshiba has already mentioned the fact their intentions were to move their 45 nm process to full SOI, including e-DRAM on SOI ).

 

[Brimstone]

Strained Silicon + SOI + e-DRAM without capacitor == good 45 nm node for Toshiba and Sony.

The 45 nm node could be the "Goldie Locks node" for Sony, where you get the right blend of clock speed for edram and logic, combined with a good transistor density that allows for more parallelism.

 

[overclocked]

I know there are a few, Pana, Vince and Paul etz that have high hopes for the PS3(i join the club) but what i feel that we have found is a argument vs different manufacturing and the core/base itself. Instead of asking "how many transistors can Sony fit on 65nm wafer" many i assume. The interesting stuff is "when. how good too competition in timescale".

IMHO the amount $ Sony invested and continue too invest is a good indication that PS3 will be all that hype.

 

[cthellis42]

I rather think CELL itself will be pretty impressive, considering the technologies behind it and the money invested in it and what Sony itself has riding on it and plans for the future... Just how it translates into a gaming experience through the PS3 implementation remains to be seen, though. I'm certain they've kept in mind to patch the kinds of problems they had hardware and software-wise with PS2's launch and what continues to give them problems even now, but PS3 seems to be a rather huge endeavor all-around, and there's likely to be some seams bursting here and there.

Of course it's the "huge endeavor" part that makes it so interesting!

 

[London Geezer]

I agree, something will have to go really really horribly wrong to make Cell any less than "impressive", seen the amount of resources (be it money, IBM-Tosh-Sony alliance, the time it's been in development etc) poured into it...
Hype aside (it's expected, being a Sony product, and especially a Sony product belonging to the Playstation family), the resources put into this architecture are REAL, and i'm not sure how they can possibly f**k this up, cause it would take A LOT to make this "not work".
The only way i can think of is in the case of Sony building PS3 and shipping it to devs without any kind of instructions on how it works. First of all, it's impossible, but for argument's sake, even in that case, there would still be big capable companies willing to spend time harnessing the architecture, hoping to then turn a profit in the long run... And of course that would have nothing to do with the capabilities of the hardware, which would still be "impressive", just a pain to program for.

 

[jvd]

I agree, something will have to go really really horribly wrong to make Cell any less than "impressive", seen the amount of resources (be it money, IBM-Tosh-Sony alliance, the time it's been in development etc) poured into it...
Hype aside (it's expected, being a Sony product, and especially a Sony product belonging to the Playstation family), the resources put into this architecture are REAL, and i'm not sure how they can possibly f**k this up, cause it would take A LOT to make this "not work".
The only way i can think of is in the case of Sony building PS3 and shipping it to devs without any kind of instructions on how it works. First of all, it's impossible, but for argument's sake, even in that case, there would still be big capable companies willing to spend time harnessing the architecture, hoping to then turn a profit in the long run... And of course that would have nothing to do with the capabilities of the hardware, which would still be "impressive", just a pain to program for.

London boy. You ever hear the quote " The best laid plans of mice and men " ? Yea . that applys to even sony

 

[zidane1strife]

Hopefully they'll be able to deliver... and surpass the expectations of the many... and of the media...