ManuVlad3.0
Regular
Sony selects Synopsys' phase shift mask software for 65nm node
- Thread starter ManuVlad3.0
- Start date
Cell was intended for 65nm to begin with. KK was disappointed to miss that target, but then again he was launching earlier than he intended too, I think. So a delay to November might see introductory Cells, in line with Sony's original intentions for the PS3. Of course, if they do launch on 65nm and manage higher yields, having one SPE deactivated for redundancy starts to sound like a waste.
A new process launch though always has some hiccups, so a redundant SPE would still have some utility early on.
Incidently Sony had wanted to basically be the first company on 65nm back in the day when they started the Cell project, so although I think they were just being optimistic and talking big thinking they could beat Intel - or really just talking - it doesn't surprise me that Nagasaki will be coming on line soon, considering the effort Sony's put towards that goal.
That said though, this article doesn't give indication as to when the 65nm node will start ramping this year, and I wouldn't be surprised if the initial PS3's were 90nm, with a quick transition to 65nm a little later. I mean 65nm at launch would be great, but I don't think we can derive that much from what's been stated here. Best to be conservative with our hopes and estimates.
Incidently Sony had wanted to basically be the first company on 65nm back in the day when they started the Cell project, so although I think they were just being optimistic and talking big thinking they could beat Intel - or really just talking - it doesn't surprise me that Nagasaki will be coming on line soon, considering the effort Sony's put towards that goal.
That said though, this article doesn't give indication as to when the 65nm node will start ramping this year, and I wouldn't be surprised if the initial PS3's were 90nm, with a quick transition to 65nm a little later. I mean 65nm at launch would be great, but I don't think we can derive that much from what's been stated here. Best to be conservative with our hopes and estimates.
archie4oz said:
Well I wasn't really trying to judge Sony on their past statements - by all means I only ever boil such things down to one central concept; in their case that they were going to try and get to 65nm as soon as possible. In that sense I don't doubt that Sony will reach 65nm favorably compared to the industry norm, I'm just wondering whether it can be ready in time to build the initial console design around that reduced heat/power profile rather than that of a 90nm Cell.
Megadrive1988
Veteran
if PS3 CELL had been introduced on 65nm that means it could've been designed for it, and we might have seen 2 PPEs and 16 SPEs (with 14 active SPEs), say 2 of those 14 SPEs for the OS, the remaining 12 free for use anywhere. just a little speculation.
and still half a "Broadband Engine"
and still half a "Broadband Engine"
Heres the old school article of STI getting to 65nm for cell in the PS3....
Toshiba, Sony close to 65nm sample production
New process technology vital for Sony's planned consumer electronics processor
By Martyn Williams, IDG News Service
December 10, 2003
TOKYO - Toshiba Corp. and Sony Corp. are close to beginning trial production of semiconductor chips using a manufacturing process more advanced than any in commercial use today, they said Thursday.
The technology is capable of making chips with features as small as 65 nanometers and its development is vital for Sony to produce its planned Cell microprocessor. The chip, which it is developing with Toshiba and IBM Corp., is expected to form the heart of its future PlayStation 3 games console and other digital consumer electronics products, but current production technologies are not yet sufficiently advanced to manufacture it in large quantities.
Toshiba's trial production of sample chips using the 65-nanometer technology will begin in March 2004, said Junichi Nagaki, a spokesman for Toshiba in Tokyo. At that time, the company will turn out system LSI (large scale integrated circuit) chips on a trial line at its Yokohama, Japan, factory and supply them to its customers for evaluation purposes, said the spokesman.
Commercial production of chips using the technology is not expected to begin until the first half of Toshiba's 2005 fiscal year, which is the period from April to September 2005, he said.
The base for that production is planned to be a new factory currently under construction at Toshiba's plant in Oita prefecture, Japan. Construction is scheduled to end in January 2004 and initial production on a 90-nanometer process is to begin in the middle of 2004 after which it will be upgraded to handle the 65-nanometer process. The plant will process 300-millimeter diameter wafers.
"This is the fundamental technology for 65-nanometer chips," said Shinji Obana, a spokesman for Sony in Tokyo. He said the two companies have already succeeded in producing a sample system LSI that also contains 32M bits of embedded memory. "We tested the device and it works correctly."
The system LSI chips, while falling short of a prototype of the Cell processor, amount to one of the first steps that Sony needs to take towards eventual mass production of the chip, he said.
Source: infoworld.com
Toshiba, Sony close to 65nm sample production
New process technology vital for Sony's planned consumer electronics processor
By Martyn Williams, IDG News Service
December 10, 2003
TOKYO - Toshiba Corp. and Sony Corp. are close to beginning trial production of semiconductor chips using a manufacturing process more advanced than any in commercial use today, they said Thursday.
The technology is capable of making chips with features as small as 65 nanometers and its development is vital for Sony to produce its planned Cell microprocessor. The chip, which it is developing with Toshiba and IBM Corp., is expected to form the heart of its future PlayStation 3 games console and other digital consumer electronics products, but current production technologies are not yet sufficiently advanced to manufacture it in large quantities.
Toshiba's trial production of sample chips using the 65-nanometer technology will begin in March 2004, said Junichi Nagaki, a spokesman for Toshiba in Tokyo. At that time, the company will turn out system LSI (large scale integrated circuit) chips on a trial line at its Yokohama, Japan, factory and supply them to its customers for evaluation purposes, said the spokesman.
Commercial production of chips using the technology is not expected to begin until the first half of Toshiba's 2005 fiscal year, which is the period from April to September 2005, he said.
The base for that production is planned to be a new factory currently under construction at Toshiba's plant in Oita prefecture, Japan. Construction is scheduled to end in January 2004 and initial production on a 90-nanometer process is to begin in the middle of 2004 after which it will be upgraded to handle the 65-nanometer process. The plant will process 300-millimeter diameter wafers.
"This is the fundamental technology for 65-nanometer chips," said Shinji Obana, a spokesman for Sony in Tokyo. He said the two companies have already succeeded in producing a sample system LSI that also contains 32M bits of embedded memory. "We tested the device and it works correctly."
The system LSI chips, while falling short of a prototype of the Cell processor, amount to one of the first steps that Sony needs to take towards eventual mass production of the chip, he said.
Source: infoworld.com
lentoPastel
Newcomer
Shifty Geezer said:
I have never really understood that SPE deactivated thing. What is going to happen in the future when they get a mature enough process to build flawless CELLs ?? We will get perfect CELL chips in the PS3 but artificialy crippled down. Or they could redesign the CELL with 7 SPE and none deactivated, but that sounds so weird, having two CELL models just because the initial yields.
Was it possible to design a temporary CELL with 9 SPE, one deactivated for redundancy, so when the process get mature they started to fab an standard 8 SPE CELL?? It will be just a little bigger and the effect in the yields with redundancy will be almost the same I guess.
I mean (clarification because my bad english), they will fabbing good CELL chips for 5 to 7 years and still deactivating (or simply not using it in PS3) one SPE just because the initial yields?
DudeMiester
Regular
Couldn't they also release an enhanced PS3, using the extra SPE for new functionality? It still wouldn't be available for games though.
Just for fun take a look at the Cell die:ERP said:
http://www.research.ibm.com/cell/cell_chip.html
I estimate that the SPEs are taking up about 50-60 % of the die space, let us say 50% to make the numbers easy. That is 6-7 % of die space for one single SPE.
If we reach 80% yield (not uncommon for a mature process in use for mass production) with 8 working SPEs we could improve it with roughly 50% (slightly less in reality) if the die anomalies are evenly distributed, that will give a yield of about 90% with at least 7 working SPEs.
So far it makes sense to keep the 8th SPE for redundancy, but if we reach 90% yield with 8 working SPEs then suddenly the 8th redundant SPE is not any longer improving the yields enough to justify the die space.
But if there is a big market for cells with 8 working SPEs, with demands of volumes similar to the PS3, the scenario will look different, then it may make sense to keep the 8th SPE.
deathkiller
Newcomer
Then they will change process again, 90->65->45->30 or something like that. At some point they will put CELL and RSX in the same die also.lentoPastel said:
I just listened to the IBM earnings conference call, and during the portion dealing with microelectronics performance they mentioned:
[The East Fishkill annex is now complete, supporting 90nm production and 65nm development.]
My question is how should the word 'development' be interpreted? Obviously since Sony provided a large amount of the funding for the 65nm line at the East Fishkill annex, 65nm production there would be nearly synonymous with 65nm Cell production.
However, 'development' != 'production' per se; any guesses as to the finer nuances of the meaning behind that word useage?
I would take it myself to mean that full 65nm production has not yet begun, but pilot production may have, or will sortly.
Thanks go to KoldFuzion by the way for providing the link to the conference call!
[The East Fishkill annex is now complete, supporting 90nm production and 65nm development.]
My question is how should the word 'development' be interpreted? Obviously since Sony provided a large amount of the funding for the 65nm line at the East Fishkill annex, 65nm production there would be nearly synonymous with 65nm Cell production.
However, 'development' != 'production' per se; any guesses as to the finer nuances of the meaning behind that word useage?
I would take it myself to mean that full 65nm production has not yet begun, but pilot production may have, or will sortly.
Thanks go to KoldFuzion by the way for providing the link to the conference call!
Last edited by a moderator:
