ATI to roll out R500 GPU (90nm) for Xbox 2 in 1Q 2005

[McFly]

http://www.digitimes.com/news/a20040930A7056.html

ATI Technologies is expected to roll out a new GPU, codenamed R500, for the Xbox 2 in the first quarter of next year, according to market sources.

The R500 will be built using a 90nm process at Taiwan Semiconductor Manufacturing Company (TSMC) and will deliver performance similar to ATI’s next-generation high-end graphics chip, the R520, the sources said.

ATI is cooperating with Microsoft to develop the chip, but the software vendor will be responsible for placing orders with TSMC. ATI will receive royalties from Microsoft based on the number of chips produced.

In related news, the sources said that ATI has completed the tape-out of its R480 chip, which is due to launch in the fourth quarter. The R480 is being manufactured at TSMC, using a 0.11-micron process.

So much about Xenon GPU at 65nm.

Fredi

 

[AzBat]

I hardly call "market sources" as facts, but yes, most likely it will be on 90nm. Most of this is just a confirmation of what we have already discussed before.

Tommy McClain

 

[Guden Oden]

Considering the fabs ATi uses haven't transitioned down even to 90 yet, anyone dreaming about 65 surely must be deluding himself severely...

 

[McFly]

Considering the fabs ATi uses haven't transitioned down even to 90 yet, anyone dreaming about 65 surely must be deluding himself severely...

from the linked and quoted article:

will be built using a 90nm process at Taiwan Semiconductor Manufacturing Company (TSMC)

Has ATI an own fab besides the external TSMC fabs?

Fredi

 

[bbot]

Would Dave Baumann like to comment on this article?

 

[jvd]

if they want a mass market launch for sept- november the chips have to start getting made in the first half of the year .

anyway a 90nm chip would be fine for the xenon .

 

[Guden Oden]

from the linked and quoted article:

will be built using a 90nm process at Taiwan Semiconductor Manufacturing Company (TSMC)

They still haven't done so YET.

 

[Vince]

Anyway a 90nm chip would be fine for the xenon .

I suppose this is dependent upon where you place the Xenon in the hierarchy of the next-generation consoles -- it's hovering somwhere around Dreamcast levels at the moment IMHO.

And the logic density differential is hardly something to shrug off JVD. As an example, One can, theoretically, take a full logic chip at 90nm (250mm2), shrink it to 65nm and add 128MB of eDRAM while retaining the initial, 90nm, die area. Or double up the logic or any combination thereof.

 

[McFly]

They still haven't done so YET.

I'm pretty sure they already did some test samples.

Fredi

 

[DeanoC]

And the logic density differential is hardly something to shrug off JVD. As an example, One can, theoretically, take a full logic chip at 90nm (250mm2), shrink it to 65nm and add 128MB of eDRAM while retaining the initial, 90nm, die area. Or double up the logic or any combination thereof.

Unfortunately 65nm is looking like post launch of any next-gen console.

 

[ERP]

And the logic density differential is hardly something to shrug off JVD. As an example, One can, theoretically, take a full logic chip at 90nm (250mm2), shrink it to 65nm and add 128MB of eDRAM while retaining the initial, 90nm, die area. Or double up the logic or any combination thereof.

Unfortunately 65nm is looking like post launch of any next-gen console.

Wow you have a date from Nintendo then ?????

 

[Vince]

Unfortunately 65nm is looking like post launch of any next-gen console.

According to... if you don't mind me asking.

 

[DeanoC]

Unfortunately 65nm is looking like post launch of any next-gen console.

According to... if you don't mind me asking.

Just rumours I'm afraid, I hope there wrong. But several different (very private) sources have said similar things about different chips and different consoles.

I have no proof, hopefully the sources are wrong or the bugs will be ironed out (I've heard that the early fab tests weren't great and its causing concern).

I should have said its just a very rough rumour when I originally mentioned it, sorry.

 

[DeanoC]

Wow you have a date from Nintendo then ?????

They still make consoles ;-) (just kidding)

As ERP points out, for all I know about Nintendo it could well be on .0001 nm technology

 

[McFly]

for all I know about Nintendo it could well be on .0001 nm technology

Is'nt that physicaly impossible?

Fredi

 

[Vince]

Just rumours I'm afraid, I hope there wrong. But several different (very private) sources have said similar things about different chips and different consoles.

I have no proof, hopefully the sources are wrong or the bugs will be ironed out (I've heard that the early fab tests weren't great and its causing concern).

Don't be sorry, but this is in direct opposition to what I've heard. I don't like mentioning it, as it's very JVD-esque, but from what I've heard the major problem was concerning back-end synthesis, the placement, timing and routing optimization, which is basically f*ed at 65nm at a prominent Chipmaker, but this shouldn't apply to Cell as much as, say, XCPU2 due to their respective timeframes and types of architectures. The process itself, from my rumblings, is quite solid.

Word is that there have been 65nm samples since 1H2004.

 

[PC-Engine]

Without knowing the die size of the Xenon's GPU, 90nm can be good or bad...

 

[cthellis42]

My sources say it's at least 1200mm^2.

 

[PC-Engine]

I AM the source!!

 

[jvd]

I suppose this is dependent upon where you place the Xenon in the hierarchy of the next-generation consoles -- it's hovering somwhere around Dreamcast levels at the moment IMHO.

from most powerfull to weakest

nintendo

ps3

xenon

Diffrence between xenon and ps3 will be slight.