ATI: We want to go... 0.11 micron very aggressively - 2H2004

T2k

Veteran
"?We do not have any problems with 0.13 margins. Actually, it is the opposite. We want to go into 0.13 and 0.11 very aggressively. If it was creating pressure, I don't think we would be banging our heads against the wall,? KY Ho told Reuters in an interview during Computex Taipei 2003 trade-show.
(...)
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."


Read the rest here: http://www.xbitlabs.com/news/video/display/20030929004713.html

Hmmm... 2ndH next year? Is this possible?
 
They might have done a great job with the RV350 using .13, but I'd like to see how they do with .13 on a more complex chip before I start jumping up and down :)
 
Ratchet said:
They might have done a great job with the RV350 using .13, but I'd like to see how they do with .13 on a more complex chip before I start jumping up and down :)

Yup. R420 should be the first "Complex" core from ATI on 0.13u....
 
AFAIK, NV will have 0.11u chips before ( or at least at the same time as ) ATI.


Uttar
 
Uttar said:
AFAIK, NV will have 0.11u chips before ( or at least at the same time as ) ATI.


Uttar

? Which one? Definitely it's not an option for the NV40 - NV50?
 
RussSchultz said:
Meaning, of course, that .11 can't be too far off if 90nm chips are already shipping. But, of course, FPGA chips are a different beast than logic chips.

True, and somewhat unreleated but to support what you're saying, you can buy a PS2 with a 90nm EmotionEngine & Graphic Synth integrated into one IC in Japan. This is not far off and kinda exciting. I for one can't wait to see what the respective IHV's use the higher density for.

Good move for ATI, although I'll keep my just mouth shut and smirk :)
 
Vince said:
RussSchultz said:
Meaning, of course, that .11 can't be too far off if 90nm chips are already shipping. But, of course, FPGA chips are a different beast than logic chips.

True, and somewhat unreleated but you can buy a PS2 with a 90nm EmotionEngine & Graphic Synth integrated into one IC. This is definatly not far off and kinda exciting.

Good move for ATI, although I'll keep my just mouth shut and smirk :)
Oh, did sony integrate the two chips already? (I avoid the console forum)
 
RussSchultz said:
Oh, did sony integrate the two chips already? (I avoid the console forum)

Yeah, Nagasaki and OTSS (fabs) started volume production in Spring. The integrated IC is down to like 86mm^2, kinda cool. Anyways, back to ATI....
 
Like like they might be on for 130nm at UMC soon as well...

http://www.digitimes.com/NewsShow/Article.asp?datePublish=2003/09/29&pages=A1&seq=1

UMC to see surging 12-inch wafer output at 0.13-micron node during 4Q
Samson Yu, Taipei; Jack Lu, DigiTimes.com [Monday 29 September 2003]


United Microelectronics Corporation (UMC) projects strong 12-inch wafer output using 0.13-micron processing in the fourth quarter, thanks to a major client’s aggressive migration to the advanced process, said sources.

UMC projects monthly output for 12-inch wafers at the 0.13-micron node will exceed 10,000 wafers by year-end, up from about 4,000 currently. UMC’s major clients have accounted for 50% of that figure, sources said.

Texas Instruments (TI) and ATI Technologies recently completed verifying UMC’s 0.13-micron process and may each input 2,000-4,000 12-inch wafers per month using the process during the fourth quarter, sources said. Xilinx is expected to significantly boost 12-inch wafer input at UMC during the second half of this year.

Compared to 8-inch wafers, using 0.13-micron processing on 12-inch wafers can produce more than twice the number of chips. Therefore, foundry customers have been aggressively migrating to 12-inch wafers despite 12-inch wafer prices remaining high, at over US$4,000, sources commented.

In the fourth quarter, UMC projects that 0.13-micron copper processing and 0.15-micron aluminum processing will respectively account for 70% and 30% of 12-inch wafer output at its Fab 12A, compared to an even mix during the second quarter, sources said.

By the fourth quarter, UMC could catch up to TSMC on 12-inch wafer output at the 0.13-micron node. But whether this happens will depend on whom ATI favors, sources commented.

For the 0.13-micron node, TSMC’s Fab12A currently outputs 5,000-7,000 wafers per month. TSMC’s major customers on the process include Nvidia and ATI. VIA Technologies and Altera are currently verifying the process, according to sources.
 
Curious

I remember a thread in here A few months back talkign about ATI's design for the next xbox system and how it would compare to the PS3 from sony. Since it seems liek teh next xbox will be based on R500 tech does that mean they are planing for .09 or .11? Also what is gonna be the size of the PS3 chip?
 
Re: Curious

CDAZ said:
I remember a thread in here A few months back talkign about ATI's design for the next xbox system and how it would compare to the PS3 from sony. Since it seems liek teh next xbox will be based on R500 tech does that mean they are planing for .09 or .11? Also what is gonna be the size of the PS3 chip?

That's almost impossible to guess at this point...mostly because we don't know where the x-box chip is going to be fabbed. If it's at TSMC, I'd guess at 0.09u. If it ends up at someone like Intel, it might be om 0.065....

I don't think it would be on any larger process than 0.09...
 
Why the shift to a more aggressive stance on smaller geometries on the part of ATI?

Has ATI hit some kind of sweet spot in respect to circuit design and process rules?

The rewards are potentially very good, but there are very obvious examples where shifting to a new process while simultaneously shifting to a new architecture can cause major headaches.

What if the coming hardware refresh cycles are their own renditions of the NV30 debacle for both leading IHVs? :oops:
 
I suspect you will not see them move to a new architecture and process at the same time. Look for them to have trialed a process on a smaller part beforehand IMO.
 
That makes sense. Would it kind of be like how the first .13 micron parts from ATI were the 9600s instead of the 9800s, or even more basic?
 
So, now that we know RV360 is 130nm low-k, lets take a wild stab at what process we think R420 will use... 8)
 
Back
Top