Why does each CELL core pack 8 VUs????

[nonamer]

Sorry, but when someone claims that S/I/T will launch the PS3 by mid-2005, I must respond and tell them it isn't likely.

I've shown in the other thread (one of them) link. Toshiba's fabs are around that that of TSMC's, the fab of NV and ATI currently, so they aren't that incredible.

Mid-2005 is about when Intel will get 65nm out, and a much smaller chip on it too (Nehalem, IIRC). I find it truly hard to believe that a lesser fab could produce a bigger chip at the same time, not to mention that they need higher volumes and lower costs than that of the CPU market.

I'm sure they could launch the PS3 by 2005, but yields will be absolutely dismal and for all practical purposes the launch will be too small to be meaningful.

You also have to factor in the developers. If the Cell is demonstrated in early 2004, then a mid-2005 launch would really give them about a year to make something, and only the biggest devs could afford that. That's very difficult, and at best you'll see a PS2 style launch will lackluster games.

I'm betting on a late 2005 launch in Japan, then we'll see the PS3 out for other places probably in 2006.

 

[Grall]

I don't claim to KNOW what's possible or not, unlike some other people

Then you dont know wether they are wrong or not either.

This is stupid.
I know that if they happen to be right, it's by sheer coincidence.

Relatively few people KNOW about these things, it's on the order of a few thousand at most. None of them happen to be frequenting this board, or at least not posting on it. If they did post, they'd most likely get fired for their trouble at the very LEAST. Ever heard of NDAs?

ANYONE can guess, MfA. It's pointless of you to start to argue semantics this way like you're trying to do, you're just trying to pick a fight. I could start spouting unfounded guesses too, if I do, will you suddenly get behind me instead of arguing with me?

Use your time in a more productive manner instead, thank you.


*G*

 

[MfA]

I argued against him in several threads ... but unfounded trashing of his opinion is as bad as him giving a poorly founded opinion (it is not unfounded). It is no more constructive than what you are accusing me of.

 

[Panajev2001a]

You also have to factor in the developers. If the Cell is demonstrated in early 2004, then a mid-2005 launch would really give them about a year to make something

Ever heard of developers starting to work on paper specs ( targetin g certain performance goals ) and using simulators much earlier than the real SDK unuts are released ?

Some developers already started conceptual R&D work as we speak: story, gameplay, music and art are things ou can start doing now without even knowing TOO much about the architecture aside the theroetical projected specs of the system.

Sure, these games will not maximize the PlayStation 3, but that is not required from a launch title on any platform.

About your comment about Sony, Toshiba and IBM's manufacturing process technology being on the level of TMSC I can say that I disagree

Risk production will start before the ~mid 2004 date, it should start as early as the end of January.

TMSC should be at least ~5-6 Months behind and I do not expect them to have a process as advanced as the one SCE and Toshiba have ( which is better than NEC's one as far as SRAM and e-DRAM cell sizes as Deadmeat posted in that other thread ).

I doubt TMSC has invested in Semiconductor technology as much as SCE has been investing: for the next three years they have already allocated over $8 Billions for Semiconductor related R&D ( $4 Billions are for CELL specifically ).

PlayStation 3 might launch a bit later than what I previewed ( so you can scale my timetable ), but that would not be a victory for TMSC because quite probably their time-table would be pushed back as well mantaining the same distance between the two companies.

 

[Panajev2001a]

Maf, I know why Grall is upset: Deadmeat has got under his skin and he is quite good at doing so.

Deadmeat has been posting in this SCE killer mode from mid 1999 onwards if I ecall correctly: the mix of "I am here to teach to the ignorant masses" + I dodge your counter arguments better than Neo dodges bullets + I am never wrong + SCe producs MUST suck in some way and they are all D00M3D got tired year after year.

Of course he posts a lot of good info and his posts are informative in that sense: I might disagree with his conclusions, but he usually posts quite a lot of meat to which base those conclusions on even if, in the past, he has done quite a bit of selective quoting.

Grall, there is no need for you and Maf to argue: you are both mature adults with some brain in their heads... I have been THAT angry with Deadmeat myself in the past, but it is more fun ( and more constructive for people who read these threads ) to proove him wrong, if he is wrong, that to make fun of him

 

[nonamer]

You also have to factor in the developers. If the Cell is demonstrated in early 2004, then a mid-2005 launch would really give them about a year to make something

Ever heard of developers starting to work on paper specs ( targetin g certain performance goals ) and using simulators much earlier than the real SDK unuts are released ?

Some developers already started conceptual R&D work as we speak: story, gameplay, music and art are things ou can start doing now without even knowing TOO much about the architecture aside the theroetical projected specs of the system.

Sure, these games will not maximize the PlayStation 3, but that is not required from a launch title on any platform.

Perhaps that is a bad reason, but it may have some merit. Assuming that the PS3 is a totally new design architecturely, it may be a while before devs can handle it. Sony may want to delay the launch just to give the devs some breathing room.

About your comment about Sony, Toshiba and IBM's manufacturing process technology being on the level of TMSC I can say that I disagree

Sony, Toshiba, and IBM, probably not. Sony and Toshiba, well, don't dismiss them too quickly. zidane1strife may think TSMC's tech is "cheesy," but in truth they're actually pretty good, so don't put it past them to match another pretty good fab.

Risk production will start before the ~mid 2004 date, it should start as early as the end of January.

TMSC should be at least ~5-6 Months behind and I do not expect them to have a process as advanced as the one SCE and Toshiba have ( which is better than NEC's one as far as SRAM and e-DRAM cell sizes as Deadmeat posted in that other thread ).

No, January is when the fab is physically built, risk production happens in the "latter half of first half of FY2004", according to the press release. Unless there is something I don't know, first half of FY2004 is around April 2004 to March 2005 according to this guy. So if it's Q2 2004, we're looking at July, August, September.

TSMC looks like it'll have 65nm by around Q42004, or October, November, December. At best, not least, Toshiba will have a 6 month lead, but it could be as little as 1 month.

I doubt TMSC has invested in Semiconductor technology as much as SCE has been investing: for the next three years they have already allocated over $8 Billions for Semiconductor related R&D ( $4 Billions are for CELL specifically ).

They certainly could not spend $4 billion only on Cell, and the $8bn on the 65nm process alone. It looks about right if they were to spend $4bn on the 65nm in general, and $8bn on all Semiconductor technology from 90nm to 45nm. They're not Intel.

PlayStation 3 might launch a bit later than what I previewed ( so you can scale my timetable ), but that would not be a victory for TMSC because quite probably their time-table would be pushed back as well mantaining the same distance between the two companies.

I agree with the first statement, but I don't think that there will be much of a gulf between Toshiba and TSMC at all. We're looking at just a few months here.

 

[incurable]

Risk production will start before the ~mid 2004 date, it should start as early as the end of January.

TMSC should be at least ~5-6 Months behind and I do not expect them to have a process as advanced as the one SCE and Toshiba have ( which is better than NEC's one as far as SRAM and e-DRAM cell sizes as Deadmeat posted in that other thread ).

No, January is when the fab is physically built, risk production happens in the "latter half of first half of FY2004", according to the press release. Unless there is something I don't know, first half of FY2004 is around April 2004 to March 2005 according to this guy. So if it's Q2 2004, we're looking at July, August, September.

Just to clarify: Toshiba's FY2004 runs from April 04 to March 05, so "the latter half of the first half of FY2004" for Toshiba is Q3/04.

I doubt TMSC has invested in Semiconductor technology as much as SCE has been investing: for the next three years they have already allocated over $8 Billions for Semiconductor related R&D ( $4 Billions are for CELL specifically ).

They certainly could not spend $4 billion only on Cell, and the $8bn on the 65nm process alone. It looks about right if they were to spend $4bn on the 65nm in general, and $8bn on all Semiconductor technology from 90nm to 45nm. They're not Intel.

One thing to keep in mind is that FABs are _very_ expensive and probably included in the grand total of $8 billion. I'd expect a decent-size 300mm 65nm FAB built from the ground up to cost $2.5+ billion.

cu

incurable

 

[Panajev2001a]

Risk production will start before Q2-Q3 2004.

New Facility

Building Structure Two-floor, steel-framed building
Building ground area 24,100m2
Floor area 48,800m2
Clean room area 15,700m2
Start of construction June 2003
Completion January 2004
Start of mass production in the latter half of first half of FY2004 (planned)

They complete the fab in January and yet they do not start limited risk production till Septemeber 2004 ?

Sony is going to spend $4 Billions on directly CELL related Semiconductor R&D and $4 Billions on general Semiconductor R&D ( non related directly to CELL ) for a total of $8 Billions to spend in the next 3 years.

Sony and SCE have licensed IBM manufacturing technology and they have been working with Toshiba very hard on their new 65 nm and 45 nm lines.

Sony has also announced they expect to see 65 nm devices in 2005 and while that does not rule out a later than mid 2005 launch, that should rule out the 2006 launch for 65 nm CELL chips.

 

[Panajev2001a]

TOKYO--Toshiba Corporation today announced that construction of an advanced 300mm-wafer clean room for System LSI at its Oita Operations plant in Oita prefecture, Kyushu would start in June this year. The fab will start mass production in the latter half of first half of FY2004, and once it reaches full production will have a capacity of 12,500 300mm wafers a month. Approximately 40 billion yen will be invested in the new fab and its clean room in fiscal year 2003, as the first stage of a four-year, 200-billion yen scale project.

Sony is basically splitting the costs of Oita #2 with Toshiba and 200 Billion Yen is equal to $1.7 Billions.

Say that Nagasaki #2 costs Sony $2 Billions and this adds up to $2.85 Billions

 

[Panajev2001a]

Sony Computer Entertainment and Sony Invest 200 Billion Yen
Over Three Years in Semiconductor Fabrication
Installation of fabrication line supporting 65 nanometer process technology
TOKYO, JAPAN, April 21, 2003 – Sony Computer Entertainment Inc. (SCEI) and Sony
Corporation (Sony) announced today that they would invest a total of approximately 200 billion
yen over three fiscal years from 2003 to 2005 in the installation of a semiconductor fabrication
line to build chips with 65 nanometer process on 300 mm wafers.
With this investment, SCEI will manufacture the new microprocessor for the broadband era,
code-named “Cellâ€￾, as well as other system LSIs, to be used for the next generation computer
entertainment system. This investment serves an important role not only for SCEI but also for
the Sony Group to develop future broadband network businesses.

http://www.scei.co.jp/corporate/release/pdf/030421ae.pdf

So, Nagasaki #2 ammounts to $1.7 Billions.

$1.7 Billions + $0.85 Billions = $2.55 Billions and this is the total for SCE as far as Oita #2 ( Toshiba + SCE ) and Nagasaki #2 ( SCE ) are concerned.

Nagasaki #2 1st floor is dedicated to CELL production while the second one is dedicated to 90 nm lines for the EE+GS@90 nm for the PSX and to 90 nm lines for the PSP SoC.

 

[Grall]

but unfounded trashing of his opinion is as bad as him giving a poorly founded opinion (it is not unfounded).

I've reputed a whole bunch of Deadmeat's statements with plenty of foundation. Problem is, he ignores all facts that speaks against his own opinions - which he incidentally treats as equivalent to the word of god - and apparantly god says cell is d00med and ps3 will suck, crash and burn. Then again, it seems he said pretty much the same thing about ps2 also...

It is no more constructive than what you are accusing me of.

...Which is what, exactly?

I just want to know what you meant by saying engineers had said there was no plan for the design of Cell. Could I please have an explanation?

You must have SOME basis for saying such a thing, right? Then what is it?


*G*

 

[zidane1strife]

I'm sure they could launch the PS3 by 2005, but yields will be absolutely dismal and for all practical purposes the launch will be too small to be meaningful.

Likely 1-2Million+... hardly small I'd say...

I'm betting on a late 2005 launch in Japan, then we'll see the PS3 out for other places probably in 2006.

if sony launches in 2005 in japan, expect it in US shores in 2005, even if they've got to do air-dev.(like they did for ps2.)... provided no disasters occur...
I'm not sure but I've heard the 300mm wafers, might allow for EVEN bigger chips than what's been expected.... So those numbers would be quite impressive indeed...

Risk production will start before the ~mid 2004 date, it should start as early as the end of January.

TMSC should be at least ~5-6 Months behind and I do not expect them to have a process as advanced as the one SCE and Toshiba have ( which is better than NEC's one as far as SRAM and e-DRAM cell sizes as Deadmeat posted in that other thread ).

I agree.

so don't put it past them to match another pretty good fab.

From what I've heard going to 65-45nm hasn't been easy, even for the likes of intel/ibm, to expect TSMC to achieve their best case scenario, and give us top stuff from the start, after all they've shown us, is...

IMHO, and from their comments, it seems more likely they'll start risk prod in early-mid 2005...

TSMC looks like it'll have 65nm by around Q42004, or October, November, December. At best, not least, Toshiba will have a 6 month lead, but it could be as little as 1 month.

Let us pray... miracles do happen after all...

 

[jvd]

From what I've heard going to 65-45nm hasn't been easy, even for the likes of intel/ibm, to expect TSMC to achieve their best case scenario, and give us top stuff from the start, after all they've shown us, is...

They moved to .13 very quickly and problem free. Well nvidia had problems but most have said it was nvidia's fault mostly . Then with ati haveing almost no problems i'd say they moved quickly .