BEHOLD, A CELL Q&A website!!!!!! (CELL to cost $100?!?!)

[Josh378]

I'm not saying this website is the official website, But it does answer some questions you may have about CELL and you may bookmark it for future updates: (and I'm still skeptical about the CELL costing $100 to make per CPU and other info on that website seems iffy to me,...but take it as it is )

http://www.cellsupercomputer.com/

-Josh378

(if this is old, I'm sorry)

 

[Carl B]

That seems a little expensive on the cost to manufacture side of things, but I guess they ARE stuck at East Fishkill at the moment.

 

[jvd]

Hmm I think thats a bit much , I can see it at 80ish if the yields are really bad .

Figure its 250m transistors ? That is a big chip for 90nm though ati and nvidia have gpus around 200m transistors on 110nm . So i dunno .

 

[Josh378]

Hmm I think thats a bit much , I can see it at 80ish if the yields are really bad .

Figure its 250m transistors ? That is a big chip for 90nm though ati and nvidia have gpus around 200m transistors on 110nm . So i dunno .

It might be less since thier using a much weaker version of CELL in PS3...So, we could see $60-80 per CELL manufactured (also, some of the article's info is also OLD...so things can change by June )

-Josh378

 

[Carl B]

It might be less since thier using a much weaker version of CELL in PS3...So, we could see $60-80 per CELL manufactured (also, some of the article's info is also OLD...so things can change by June )

-Josh378

I don't think that 'weaker' version should effect much, unless the quoted cost is for a full 8 SPE chip..

It's still the same die area - they will have higher yields though by being able to tolerate a defect on an SPE. Speed-wise, 3.2 Ghz and 4 GHz should both be achievable on the majority of chips - it's just a matter of voltage applied.

You know by the way, before I keep talking blindly here, can you link or quote to the exact text giving that estimate? Because I can't find it. :?

 

[Josh378]

It might be less since thier using a much weaker version of CELL in PS3...So, we could see $60-80 per CELL manufactured (also, some of the article's info is also OLD...so things can change by June )

-Josh378

I don't think that 'weaker' version should effect much, unless the quoted cost is for a full 8 SPE chip..

It's still the same die area - they will have higher yields though by being able to tolerate a defect on an SPE. Speed-wise, 3.2 Ghz and 4 GHz should both be achievable on the majority of chips - it's just a matter of voltage applied.

You know by the way, before I keep talking blindly here, can you link or quote to the exact text giving that estimate? Because I can't find it. :?

it's under the section:

How can one chip be so fast?

Since the cell will be the heart of the the playstation, they are immediately in largest volume production so each chip will always cost the the range of $100

 

[colinisation]

Hmm I think thats a bit much , I can see it at 80ish if the yields are really bad .

Figure its 250m transistors ? That is a big chip for 90nm though ati and nvidia have gpus around 200m transistors on 110nm . So i dunno .

I thought ati and nvidia were on 130nm at around 220m transistors. I also think about 300m transistors for R520 and NV50 (or whatever that is).

 

[mckmas8808]

Hey guys is it any possibility that Sony could hike up the speed of the CELL to 3.6 or 4.0 GHz before that PS3 is released? And when I say any possibility I mean any small like 1% chance

 

[DuckThor Evil]

Hey guys is it any possibility that Sony could hike up the speed of the CELL to 3.6 or 4.0 GHz before that PS3 is released? And when I say any possibility I mean any small like 1% chance

I think 4.0 is too much, but I wouldn't be too suprised to see 3.5, however I believe it is going to be 3.2 like they announced.

 

[Carl B]

it's under the section:

How can one chip be so fast?

Since the cell will be the heart of the the playstation, they are immediately in largest volume production so each chip will always cost the the range of $100

Oh ok - well I see the logic now; they're making a claim as to how inexpensive the Cell will be to produce compared to other lower-volume high-end chips and they just used the PS3 as the reason for it. That prediction of theirs was made in in Jan of 2005 before ISSCC; I strongly agree with their point of course, but I'm sure if they revisited the issue of cost today, they would agree that even $100 seems too high for this chip in terms of manufacturing costs.

To say nothing of process shrinks, that will make the price to produce plummet.

 

[Josh378]

it's under the section:

How can one chip be so fast?

Since the cell will be the heart of the the playstation, they are immediately in largest volume production so each chip will always cost the the range of $100

Oh ok - well I see the logic now; they're making a claim as to how inexpensive the Cell will be to produce compared to other lower-volume high-end chips and they just used the PS3 as the reason for it. That prediction of theirs was made in in Jan of 2005 before ISSCC; I strongly agree with their point of course, but I'm sure if they revisited the issue of cost today, they would agree that even $100 seems too high for this chip in terms of manufacturing costs.

To say nothing of process shrinks, that will make the price to produce plummet.

I want to know how much would a blu-ray player will cost Sony and the XDR and RSX....until then...

My Predictions:

I think that the RSX will cost as much as the R500

XDR-Ram at $50, GDDR3...much less

Blu-ray=$100

-Josh378

 

[Carl B]

You can come HERE if you want to discuss it or just see what some ideas are. Basically a thread I started on another forum devoted to predicting PS3's manufacturing costs. Some good things hidden in there, but of course a lot of blind guessing, since guesses educated or otherwise is all anyone has at the moment.

I think the RSX will cost more than the R500, depending on how expensive eDRAM is. The fact that it has it's eDRAM and the main core manufactured seperately though should give it a yield advantage to RSX's straight 300 million transistors.

I agree (roughly) with the RAM theory, but think you're pricing the blu-ray ROM drive a good bit too high.

 

[Josh378]

You can come HERE if you want to discuss it or just see what some ideas are. Basically a thread I started on another forum devoted to predicting PS3's manufacturing costs. Some good things hidden in there, but of course a lot of blind guessing, since guesses educated or otherwise is all anyone has at the moment.

I think the RSX will cost more than the R500, depending on how expensive eDRAM is. The fact that it has it's eDRAM and the main core manufactured seperately though should give it a yield advantage to RSX's straight 300 million transistors.

I agree (roughly) with the RAM theory, but think you're pricing the blu-ray ROM drive a good bit too high.

OMG, I thought that website died out...wow, I loved that site...looks like I'll be posting back there..once again (thanks for the link)

-Josh378

 

[Carl B]

You can come HERE if you want to discuss it or just see what some ideas are. Basically a thread I started on another forum devoted to predicting PS3's manufacturing costs. Some good things hidden in there, but of course a lot of blind guessing, since guesses educated or otherwise is all anyone has at the moment.

I think the RSX will cost more than the R500, depending on how expensive eDRAM is. The fact that it has it's eDRAM and the main core manufactured seperately though should give it a yield advantage to RSX's straight 300 million transistors.

I agree (roughly) with the RAM theory, but think you're pricing the blu-ray ROM drive a good bit too high.

OMG, I thought that website died out...wow, I loved that site...looks like I'll be posting back there..once again (thanks for the link)

-Josh378

My pleasure!

 

[mckmas8808]

I read on ga-forums that this is the whats registered.

Registrant:
Trounson [snip]

1365 West Brooks # 6
Ontario, CA 91762
US
Phone: [snip]
Fax: 999 999 9999(?)

Domain Name: CELLSUPERCOMPUTER.COM

Administrative Contact , Technical Contact :
Trounson, [snip]
[snipped]@YAHOO.COM
1365 West Brooks # 6
Ontario, CA 91762
US
Phone: [snip]
Fax: 999 999 9999 (?)

 

[jvd]

Hey guys is it any possibility that Sony could hike up the speed of the CELL to 3.6 or 4.0 GHz before that PS3 is released? And when I say any possibility I mean any small like 1% chance

Na i think 3.2 ghz is a good speed and they do need to keep yields up very very high .

This isn't like a server chip. They can't make a 1x8 at 4ghz with really bad yields and use the rest as 3.5 ghz ones and sell them cheaper u know what i mean ?

 

[version]

if sony start on 65 nm 4-5 Ghz possible

 

[jvd]

if sony start on 65 nm 4-5 Ghz possible

but they wont be . They will start on 90nm and the switch to 65nm wont be for awhile , most likely a year .

Sony wouldn't have announced what htey have already if that wasn't very close to what we were going to get

 

[version]

if sony start on 65 nm 4-5 Ghz possible

but they wont be . They will start on 90nm and the switch to 65nm wont be for awhile , most likely a year .

Sony wouldn't have announced what htey have already if that wasn't very close to what we were going to get

sony play foxy

 

[jvd]

na we all know if they seriously thought they could have 65nm chips ready they would have said it and would have used it on a slide :Using state of the art 65nm chip far more advanced than the 90nm process for the xbox 360 : .

Sony is most likely launching in japan sprin 2006 so they will need to start making systems in nov of this year or so to have a few million for launch there .

Not only that but if you build a chip that needs 65nm to attain its speeds at reasonable yields you wont really have major price cuts till 45nm . With a chip at 90nm with reasonable yields you will get your first major price drop on the 65nm process and we already see the cells flops raiting beating that of the x360 flops raiting so sony doesn't really need any more pr numbers in that catagory .


I can see them mabye going to 3.4 or 3.5 but i doubt it