ATI Talks Lithography

[Vince]

Shamelessly ripped from the 3D Graphics Forum topic entitled, "ATI: We want to go... 0.11 micron very aggressively - 2H2004"

http://www.xbitlabs.com/news/video/display/20030929004713.html[/url]]ATI Technologies’ co-founder and CEO, Mr. KY Ho said last week that his company was planning a very aggressive transition to 0.13 and 0.11 micron fabrication technologies. This is the first claim of such kind coming from an ATI top executive, as the company is generally very cautious and its decision not to manufacture the high-end RADEON 9700-, 9800-series of graphic processors using 0.13 micron technology, but stick to mature 0.15 micron process, allowed the firm to leave its main rival NVIDIA behind in terms of performance and production costs...

Transition to 0.11 micron technology will probably start in the second half of 2004, about a year and a half after the first 0.13 micron chips left TSMC foundry. Hence, all next-generation ATI’s VPUs will be made using 0.13 micron technology, while the future generation graphics products, such as code-named ATI R500, will be manufactured using 0.11 micron technology. It is also possible that ATI will make a less complex graphics processor for mainstream or value market segment using 0.11 micron technology for evaluation the process in the second half 2004.

This must be incorrect, no? I mean, I expected STI to be ahead in the process arena (eg. Density, Power attributes) due to some inherient reasons tied to the companies they are (eg. R&D expendatures) and physical design precedent - but this is an insane differential.

 

[Paul]

Oh man

 

[jvd]

It makes sense for ati. All the major players are talking up ati's tech so they will want to push it as far as they can. Hopefully they will be successfull with it. Or at least have a good back up plan . Or perhaps the 9700pro was thier back up plan.

 

[Panajev2001a]

Interesting...

STI seems undecided between 65 nm and 45 nm SOI with e-DRAM cells without the capacitor ( to tell the truth I think that 45 nm looks better for CELL [I think that the CELL processors in PlayStation 3 might be released with 65 nm technology even if they are not very economically feasible because the 45 nm chips would not be so far away] and the new plants they are building will be upgradable relatively quickly to 45 nm as Toshiba did state already ) and ATI is planning 110 nm for 2004 and ~90 nm for 2005.

 

[jvd]

Interesting...

STI seems undecided between 65 nm and 45 nm SOI with e-DRAM cells without the capacitor ( to tell the truth I think that 45 nm looks better for CELL [I think that the CELL processors in PlayStation 3 might be released with 65 nm technology even if they are not very economically feasible because the 45 nm chips would not be so far away] and the new plants they are building will be upgradable relatively quickly to 45 nm as Toshiba did state already ) and ATI is planning 110 nm for 2004 and ~90 nm for 2005.

Don't forget though ati wont be adding over 32 megs of edram to thier chip. So they can get away with a console chip that size and more money saved as it drops in micron process than the cell chip will .

 

[Paul]

Over 32mb on BE? Maybe; if they skimp on the External DRAM(XDR)

 

[jvd]

Over 32mb on BE? Maybe; if they skimp on the External DRAM(XDR)

Well i said over 32mbs cause i dunno what they'd have mabye i should have said at least 32mbs

 

[Paul]

Yea, I see ATLEAST 32mb e-DRAM(Easily over 300+GB/s) on PS3's Cell. Coupled with around 256mb of External DRAM. This and e-dram for Rasterizer.

Sony is going to take ENOURMOUS losses on PS3, I can see it coming. However they seem to be banking on the fact that they will get everything down to size really quick to normalize costs.

 

[jvd]

Yea, I see ATLEAST 32mb e-DRAM(Easily over 300+GB/s) on PS3's Cell. Coupled with around 256mb of External DRAM. This and e-dram for Rasterizer.

Sony is going to take ENOURMOUS losses on PS3, I can see it coming. However they seem to be banking on the fact that they will get everything down to size really quick to normalize costs.

Yes which may very well back fire big time . Although i think your wrong on the main memory size. I think it will be more like 128 megs.

 

[Paul]

Won't backfire. You have people working for Sony whose only job is to work on things like this.

128mb main Dram? Not a chance, this is only 4X, hardly enough.

http://www.gamesindustry.biz/content_page.php?section_name=dev&aid=1937

 

[jvd]

Won't backfire. You have people working for Sony whose only job is to work on things like this.

128mb main Dram? Not a chance, this is only 4X, hardly enough.

http://www.gamesindustry.biz/content_page.php?section_name=dev&aid=1937

People were working at nvidia to make sure the nv30 was a good product. Poo happens. The best laid plans of mice and men .... all that good stuff .

 

[Nexiss]

Umm, ok...

We really don't know enough about what kind of deal Nintendo has with ATI to comment on how this might affect that.
With MS, though, we know that they are licensing tech. I don't believe we know if they are getting tech based off the R500 or not, but I don't think it is the R500. Also, I don't think MS is planning on launching 2H2004, but more likely 2H2005.
So, to condense it, I don't think this really tells us much more than we already knew. As has been said before, it is looking like .09 for the GPU in XBox2.

 

[Paul]

People were working at nvidia to make sure the nv30 was a good product. Poo happens. The best laid plans of mice and men .... all that good stuff .

nv30 wasn't in development Since 1999, nor do I think it had billions of dollars thrown at it.

You can't compare a GPU which has a shorter development cycle to a console.

 

[Nexiss]

They aren't exactly the same situations, but the attitude is still the same. Truth is, you can't say there is no possibility for failure. Business is filled with risks, Sony is taking a risk, which may benefit them greatly, but there is also the possibility of failure (and what a spectacular failure it would be).

 

[jvd]

People were working at nvidia to make sure the nv30 was a good product. Poo happens. The best laid plans of mice and men .... all that good stuff .

nv30 wasn't in development Since 1999, nor do I think it had billions of dollars thrown at it.

You can't compare a GPU which has a shorter development cycle to a console.

I'm sorry how do we know its been in development since 1999. We know they scrapped thier cpu for the cell chip. How do we know they didn't scrap the graphic chip too. Btw the time in development does not mean it will be sucesfull. The rampage was in development for many years and it ultimately failed and never came out. The ithium cpu was in development and has been in retail for many years and has still failed.

 

[Paul]

I'm sorry how do we know its been in development since 1999.

Early Cell patent's were filed in 2000. Can't make a patent if your not working on something.

Oh but yes, Cell's main purpose(This is why they signed on) is for use in PS3. Have a look, IBM won the PS3 CONTRACT.

http://news.bbc.co.uk/1/hi/business/1216551.stm

We know they scrapped thier cpu for the cell chip.

This is wrong. Cell was and still is the only CPU destined for PS3. Since ever.

How do we know they didn't scrap the graphic chip too.

Scrap what? What is this other graphics chip you speak of?

 

[jvd]

I'm sorry how do we know its been in development since 1999.

Early Cell patent's were filed in 2000. Can't make a patent if your not working on something.

Oh but yes, Cell's main purpose(This is why they signed on) is for use in PS3. Have a look, IBM won the PS3 CONTRACT.

http://news.bbc.co.uk/1/hi/business/1216551.stm

We know they scrapped thier cpu for the cell chip.

This is wrong. Cell was and still is the only CPU destined for PS3. Since ever.

How do we know they didn't scrap the graphic chip too.

Scrap what? What is this other graphics chip you speak of?

When the ps2 was deisgned they had a plan. The gscube was one part and then the ee3 and gs3 were meant for the ps3. They then switched to the cell cpu. Which was designed for ibm first and foremost and then used by sony .

 

[Paul]

Your confused.

Gscube was nothing but a test for Paralell computing, this is what PS3 on the inside will be with Cell.

You can go see on one of the articles on the Gscube where they say this, that Gscube is just where developers will get good practice for future playstation consoles such as PS3.

Second.

Look at EE3 and GS3.

I have huge doubts that the CPU inside ps3 will be called "Cell", they WILL BE BUILT on cell tech, but it's doubtfull they will be called Cell. They will I'm guessing still be called EE3 and GS3.

My last point of evidence.

Back in 99 Sony claimed that EE3 and GS3 would have 1000X the performance of ps2.

Fast foward to GDC 2002.

Okamoto says that a PS3 based on Cell will have 1000X the performance of PS2.

Same performance claims, anyone else seeing the connection?

 

[Vince]

JVD, thanks for trolling my thread on lithogrophy with unrelated talk about an individual products success or failure. I mean, you know a threads been hijacked by lunes when "Rampage" is brought up... yet again. How about we get the mod in here and delete this dudes posts... oh, wait.

Ohh, and BTW: Before you talk, can you tell me the density of an eDRAM cell @ 65nm and the aggregate area for 64MB? I would consider this precursor knowledge before making a comment like:

Don't forget though ati wont be adding over 32 megs of edram to thier chip. So they can get away with a console chip that size and more money saved as it drops in micron process than the cell chip will

Perhaps you should read up on the impact that designing for 65nm SOI instead of 110nm Bulk has on an architecture. Hint: it's fucking huge.

EDIT: And then when you find the aggregate area the eDRAM will take up, tell me if it'll fit in the differential between ATI's largest core yet (eg. R3x00's 200mm^2) and Sony's GS (270mm^2). And then factor in how much more logic you can fit between the remainder on a 65nm process and a 110nm process. Thanks.

 

[nonamer]

This pretty much confirms my suspicion that ATI is the low-cost, low-risk GPU provider for Microsoft. If they are wait till 2005 for the 90nm, then frankly I don't see them very competitive in terms of power (probably comparable, like the NGC is to the XB, but you can't simply say the GC is equal/better). As it stands now, even if Cell is a lousy design, PS3 still has probably won the powergame already simply due to brute force design and process investments. I saw MS as a potential challenge, but unless a major change in their PR then I can't see MS seriously challenging Sony anymore.

Panajev, I'll bet you a $1000 that the PS3 will be out on 65nm. 45nm is waaaayyyy out there, like 2007/2008. There are major lithography problems at 45nm and leakage on that process could make SOI scream "mommy!" I am certain that S/T/I? will release at 65nm, maybe lose [lots of] money, and move to 45nm later when it is ready.

Vince, is Sony's SOI PD or FD?