Sony's 300-mm fab starts test production of Cell processor
- Thread starter Gubbi
- Start date
IBM has always been listed as an intended participant in the manufacturing/fabbing of Cell chips, with their Fishkill plant. Sony/IBM have never indicated anything about beginning Cell production simultaneously at all of their plants so one was always bound to start before the others. Not sure there's much to read into why IBM may be first when it seems that Nagasaki will be close behind.
Since Fishkill sounds like its been ready longer than Sony's fabs, it just makes sense that they'd start there first.
Since Fishkill sounds like its been ready longer than Sony's fabs, it just makes sense that they'd start there first.
East Fishkill will be ready first, so production will start there first. The earlier sampling and higher-scale production can start, the more testing is done, and the more mature the process will be. The better-off the testing, the better and earlier the processes can be replicated in Nagasaki and Oita and follow similar lines--or improved ones depending on what is discovered in Fishkill's production. The earlier and better production is, the earlier and better they can test the chips externally. The sooner any refinement can be made to the lines if necessary. The longer they have to test the PS3 as a whole unit and make adjustments to anything ELSE. The more kits can be created. The more chips can be ready for a broader launch. The more CELL chips are available for separate uses and testing.
With Fishkill already set for CELL production, why spend the time and money to shift it? It would still have the most mature processes, and likely be the easiest to experiment with working with IBM engineers. (Not to mention why would Sony invest so much in Fishkill if it was expected to shift lines the moment Nagasaki and Oita #2 were up and running?) And extra production is always useful. If not all the lines of Nagasaki #2 and Oita #2 are needed specifically for PS3 chip production, then they'll be moved to other SoC solutions for other devices as well. (If as stated production for PS3 is aimed for 15,000 wafers/month, then they'd also be aiming to have headroom should problems arise. If capacity is improved over time--as the Oita article states, moving to 12,500 wafers/month itself by FY2007--then they may be hoping demand keeps going and they can provide it, or they're encorporating other lines or the same chips for other devices as well.)
Eventually any of the fabs could be retargetted--and perhaps Sony's agreement with IBM in East Fishkill has a maximum duration--but why is any of what we've seen thus far questionable? We've known how big an endeavor it is for them for a long while.
With Fishkill already set for CELL production, why spend the time and money to shift it? It would still have the most mature processes, and likely be the easiest to experiment with working with IBM engineers. (Not to mention why would Sony invest so much in Fishkill if it was expected to shift lines the moment Nagasaki and Oita #2 were up and running?) And extra production is always useful. If not all the lines of Nagasaki #2 and Oita #2 are needed specifically for PS3 chip production, then they'll be moved to other SoC solutions for other devices as well. (If as stated production for PS3 is aimed for 15,000 wafers/month, then they'd also be aiming to have headroom should problems arise. If capacity is improved over time--as the Oita article states, moving to 12,500 wafers/month itself by FY2007--then they may be hoping demand keeps going and they can provide it, or they're encorporating other lines or the same chips for other devices as well.)
Eventually any of the fabs could be retargetted--and perhaps Sony's agreement with IBM in East Fishkill has a maximum duration--but why is any of what we've seen thus far questionable? We've known how big an endeavor it is for them for a long while.
Pepto-Bismol
Newcomer
Megadrive1988 said:
I don't think so!
Just getting Cell to mass production sounds like a Herculean effort that's still another year or so in the making. Now it only stands to reason that any changes would only lengthen its deadline.
That's the plan when Cell was first announced.
The original plan was Cell on 100nm process from IBM. Since they invested money in IBM for better process, its only natural IBM has to show the result first. Then they'll duplicate the know how to the other fabs. It saves money that way.
Megadrive1988
Veteran
can we pin point roughly the time Cell went into development by STI?
I mean, yeah we know in early 2001 Cell was announced to the public but certainly it was underway before then.
I mean, yeah we know in early 2001 Cell was announced to the public but certainly it was underway before then.
overclocked
Veteran
When are PS3 set to launch, was it 2006middle to all regions or?
Ken Kutaragi said:
I find it amazing that some people* can take good news like this and spin it into yet another anti-Sony diatribe. As far as I'm concerned if we know E. Fishkill is to begin production before SCE-Nagasaki, and SCE-Nagasaki has begun test production now: how is this anything but good? It's 2004.
*And their name isn't DeadmeatX to boot!?!
One thing is for sure that Sony and Toshiba has done extensive early work @ 65NM. I would imagine that they would be further along than IBM....
"Toshiba and Sony have codeveloped the world's first 65nm CMOS process technology for embedded DRAM system LSIs - a major breakthrough in process technology for highly advanced, compact, single-chip system LSIs that will be only one-quarter the size of current devices while offering higher levels of performance and functionality." - electronicstalk.com
"Toshiba and Sony have codeveloped the world's first 65nm CMOS process technology for embedded DRAM system LSIs - a major breakthrough in process technology for highly advanced, compact, single-chip system LSIs that will be only one-quarter the size of current devices while offering higher levels of performance and functionality." - electronicstalk.com
Current guesstimates are late Q1 2006 in Japan and late Q2-to-early Q3 2006 in the US and Europe. No official releases have been mentioned. (And likely won't be finalized until next E3.)overclocked said:
No doubt any dates could be pushed one way or another, depending on how solid a launch they'd want, and how much extra cash they'd be willing to blow. If they indeed can have mass production started by the end of 2004, that could certainly push out something in Q3/Q4 2005. I just don't think they're wanting to push that hard, as the launch games would suffer, and they ready a better (and probably more capable machine in ways other than just CELL chippage) machine in 2006. I guess some of that depends on what Microsoft and Nintendo have in mind...
It's not like Sony and Toshiba's tech isn't touching the East Fishkill line or something--it's being readied for whatever STI is delivering. It is, however, the more mature fab--and the one that best encorporates IBM's process tech--which is exactly what Sony and Toshiba's engineers are leaning on.Mythos said:
