45nm Cell in 2009

[eastmen]

http://www.engadget.com/2008/09/22/sony-and-toshiba-to-begin-mass-producing-45nm-cell-processor-in/

45nm cell in 09 it seems. Looks like they can get a price drop next year.

 

[ADEX]

45nm cell in 09 it seems. Looks like they can get a price drop next year.

This can be taken a rather different way:

IBM are making the 45nm Cell as planned on SOI whereas the Sony / Toshiba Cell is to be made on a bulk CMOS process.

As announced almost a year ago:

http://forum.beyond3d.com/showpost.php?p=1077612&postcount=1

So, that's a price drop this year and another next year
...and another after that when they get a 45nm RSX.

 

[eastmen]

The only thing i see in that is that the company will launch in april 2008. It doesn't mention when the chips will be produced though. But thanks for the information. I don't see them hitting 45nm till early next year.

It also doesn't apear that the 65nm RSX has made it into the consoles yet either , so we will most likely see that move first

 

[JPT]

The only thing i see in that is that the company will launch in april 2008. It doesn't mention when the chips will be produced though. But thanks for the information. I don't see them hitting 45nm till early next year.

It also doesn't apear that the 65nm RSX has made it into the consoles yet either , so we will most likely see that move first

But if IBM is already producing 45nm cell at Fishkill, then theoretically Sony could provide PS3's with 45nm Cell now. If IBM made chips are cheap enough and in a high enough quantity is another ball game

 

[eastmen]

But if IBM is already producing 45nm cell at Fishkill, then theoretically Sony could provide PS3's with 45nm Cell now. If IBM made chips are cheap enough and in a high enough quantity is another ball game

Yes they could , however it doesn't seem likely to me that IBM is making them for sony currently.

 

[Crossbar]

Yes they could , however it doesn't seem likely to me that IBM is making them for sony currently.

Why is that? Sony tend not to rely on single sources for key components and the process used for Cell production is far from a commodity.

Setting up the first 45 nm production line by the experienced Fishkill staff would make perfect sense to me.

 

[liolio]

There is a point I would want some of you to make clear.
Wasn't sony producing the cell @65nm? (and will for the RSX)

 

[Carl B]

The Cell going into PS3 will still remain SOI; they will be sourced from Fishkill and perhaps Nagasaki's lines as well. The CMOS development is for general Cell/SPE CE applicability and not necessarily for PS3.

http://www.beyond3d.com/content/news/514

http://www.beyond3d.com/content/news/582

Anyway whether IBM is fabbing these yet or not, Sony would still need to work through their existing 65nm Cell inventories before they made the transition. Remember that just because something is being manufactured doesn't mean it's making its way into retail yet.

 

[liolio]

Thanks for your response.
So Sony may already "outsource" the cell production, I was in fact wandering because Sony may lose some control of its production cost by buying cell to IBM.
But it appears that they already do it to some extend.

EDIT
Another thanks the articles adress my questions perfectly

 

[Carl B]

Sony helped fund the new SOI lines at IBM, so they have rights to some of the production there anyway. As for IBM's Cell production, most of it is intended for PS3 to begin with. In terms of Sony's manufacturing of Cell on 65nm, yes, but they sold those lines to Toshiba recently. So when we're talking about Toshiba potentially being a supplier also, it's on Sony's old lines that both companies run through a joint-venture.

Reading the first article I linked to and then the article linked within that should clarify the relationship between the three for you.

 

[Panajev2001a]

The Cell going into PS3 will still remain SOI; they will be sourced from Fishkill and perhaps Nagasaki's lines as well. The CMOS development is for general Cell/SPE CE applicability and not necessarily for PS3.

http://www.beyond3d.com/content/news/514

http://www.beyond3d.com/content/news/582

Anyway whether IBM is fabbing these yet or not, Sony would still need to work through their existing 65nm Cell inventories before they made the transition. Remember that just because something is being manufactured doesn't mean it's making its way into retail yet.

Well, having CELL on bulk CMOS might be an advantage if and when both it and RSX are ported to 32 nm (combo-chip ? You might save quite a bit of space simplyifying the FlexIO stuff too).

 

[Carl B]

I think 32nm one way or another will bring a big shift from the period up until now, because even if they were to stay on the IBM branch, I'd expect a true re-working of the chip associated with the transition to HKMG, so it should get rid of those 'dead space' areas associated with these automated shrinks thus far and the ability to finesse I/O; should be significant gains across a variety of criteria.

http://www.beyond3d.com/content/news/627

I agree that if they went CMOS instead, there would obviously be a whole other set of advantages associated. But if STI is developing Cell 32nm on an HKMG process, I would just have to think that Sony go that route vs CMOS. Since Toshiba is involved also in the non-STI HKMG alliance, maybe it all ends up converging anyway. With IBM's purported future ease of transition to 22nm, there may be that additional 'future advantage' aspect as well to staying on that process branch.