they want to make sure all the thermal issues are completely taken care of, the pushed out the 3.06 slightly faster then they should of because they had pledged it to the channel last spring. Now they don't feel much pressure to go up in speed. They want to make sure the fsb is stable and runs correctly. I know i'm not particulariy impressed with the 533 fsb processors.
they perform better tahn the 400fsb but there simply aren't any good (read payable or upgrdable) chipsets delivering that. 850i and RDARAM doesn't really provide you with a great upgrade path granite bay mobo's are expensive, still piv 533fsb is the top performer @ the moment if you don't like em it's because you want the 400 for oc reasons and even then, if you get a recent chip you can get it to 156-160 fsb without a problem which gives you a 600-640fsb that better for ya???? :-p
I do agree that the 533Mhz FSB comes with somewhat of a hefty price tag attached to it, if you want a really good system...
Canterwood will certainly make things much more interesting for Intel, won´t it? Highly upgradeable, and Springdale should be affordable, i.e. the value alternative to canterwood...
Actually, anyway, AMD-based systems are not in their peak offering of "upgradeability" right now either. So Bartons are a possible upgrade path for people who own any system, OK, but if you were to buy an amd-based rig now, you wouldn´t upgrade to much more than a Barton 3200+, best case scenario... With Hammer tech around... Which raises a question to which I don´t have an answer: Will there be a sucessor to the 3200+, still barton-core?
How many times have I repeated, 0.09m is being delayed on ALL fab owners due to tool problems!
Intel delayed the Prescott from early half 2003 to late half 2003, AMD followed with 2004.
EDIT: I just found one of the 90nm news problems here: <A HREF="http://www.theinquirer.net/?article=7532" target="_new">http://www.theinquirer.net/?article=7532</A>
So while Intel seems to be able to get through the money problem, it seems it doesn't stop here and that indeed there are design issues with 90nm. That SRAM 52Mbit cell they did was simply that, a cell, nothing complex. Maybe someone here (Juin?) remembers why exactly was 90nm delayed at the big fab firms.
This post is brought to you by Eden, on a Via Eden, in the garden of Eden. :smile: <P ID="edit"><FONT SIZE=-1><EM>Edited by Eden on 03/30/03 10:34 AM.</EM></FONT></P>
Just curious why you are so sure about tool issue delaying everyone. Also your post is definitely controdictory: first you say it is tool issue, then there is design issue. What is it exactly?
Also why did you say SRAM is nothing complex? SRAM has the tightest pitch (therefore smallest feature) in lithography, so the success in SRAM demonstrated the capability of 0.09um manufacturing. It has been over a year since Intel announced 52M SRAM. You think they are still stuck at the same place?
I like to hear some of your comments from your own technology point of view, if any. Is theinquirer.net the only source of your knowledge?
I wouldn´t trust the Inquirer <i>that</i> much, really... But maybe I´m too cynical anyway...
Even the Inquirer agreed, though, that Intel is "a big company with a lot of breath" (<i>by the way, the Inquirer has articles that aren´t well-written at all! That´s why I´m cynical!</i>) Intel really has a ridiculous amount of resources...
And if indeed there are troubles with tools required, then how the hell did that SRAM get done?... (one year from now, mind you!) I think Eden tried to explain that to me in another thread, but I forgot to answer (sorry bout that!) Am I forgetting something here or what? Are they still utterly unable to produce Prescott? Why do some Prescott die pictures exist, then??? Maybe the process is still too expensive for mass production or something... Besides, if Intel <i>could</i> get Prescott out earlier than announced, why would they want to? Hammer is their main concern now; AMD is somewhat out of weapons to fight with now... maybe they´re just perfecting Prescott to go vs Hammer or something... Any thoughts on that, anyone?
Moreso than anything lately, Inquirer has been right on.
I hate it when people always flame Inq when lately they have been telling nothing but the truth.
I realize I am not aware exactly of the problem itself, however it has been said that the 0.09m tools have not been entirely mastered, whether it's to acquire them financially or simply use them, I dunno. All I know is that Prescott has been delayed, and it hurts the likes of you who so wanted it on this first half. Intel's roadmap shows it as late H2.
AMD is in no better ground, they will be delaying their 0.09 production even further, placing what was supposed to come out this year as 0.09m, into next H1 04. Now imagine them refining that process, it will take a year easily. Whether IBM's tie with them will help or not remains to be seen.
Last I checked, cache in a CPU is not exactly much complex. Hell, it is one of the LEAST heat consuming part. (save for Trace Cache) Don't believe me?
Go explain to me why 256K L2 more for the Barton has merely raised the watt output by a few watts.
This post is brought to you by Eden, on a Via Eden, in the garden of Eden. :smile:
Intel 90 NM process use mix 247 wave lenght tool and 190 nm tool.The wave lengh at 190 nm have some rouble to are not relate to the manufacture itself but some issue that agilent or applied materail have make it as we are at the end of the UEV.Until we move to EUEV (extreme ultra viloet)(sorry for the misplelling).Thwy will got some issue making the equipement and the process.The process as gone faster that normal at intel but the equipement do not follow like they do.Intel is working close with equipement maker to solve the issu or find a others way to get it work.Also the design was a bit late they have finish 2 month ago.I suspect that most of corporation will move to 90nm in 2004 Q4 expection to Intel /IBM/Samsung/Toshiba.Intel will win the race with there presscott but a simple controleur from toshiba can win the 1 to produce but on a 100 million CMOS intel will come 1 with a nice 6 to 1 year adavance.
I may look a anti-AMD but i dont expect 90 nm from AMD before 2004 Q4 or 2005 maybe in the same time that Itanium move to 90 nm.
To Juin: Agilent Techology or Applied Materials donot make any lithography tools, whether or not it is 248nm or 193nm. If anything went wrong with those two companies, it shouldnot be a wavelength related issue.
To Eden: I donot know why you keep indicating SRAM is trivia. All generations of logic chip manufacturing start with SRAM. Since SRAM has the smallest lithography feature and highest density, everyone has to deal with pitch printing, defect problem, etc. Certainly SRAM runs on a strip-down version of process flow, meaning it has less masks, simpler backend and less types of transistors, it is not any easier to make, especially at an introduction of brand new technology. If it takes Intel 18 months from announcing SRAM in Mar. 2002 to launching real products in Sept 2003, we can only speculate AMD wouldnot have 0.09um until no earlier than end of 2004, just like what Juin said, simply because AMD does not have SRAM yet. Count on IBM? In my opinion, AMD is right on track to be a chip designer, less of a chip maker. Wanna bet?
BTW, why do you mention heat consumption of L2 Cache? I donot see any logic.
I will step down from here refraining from any more information adding, as I am not in any way studying semiconductors and therefore cannot go further than what I had read about in the past.
I mentioned the heat consumption because I thought you meant the cache's complexity would yeild high heat (assuming it works most of the time). Usually a complex area in the chip, if I am not mistaken, that works often with its complexity, would be yeilding a lot of heat. The 256K L2 added about 6W overall, so it is definitly not the biggest consuming part, and to me it sounds like it is definitely not the most complex piece.
Anyways, I'll stop here and keep my integrity intact by not walking into uncharted territory.
This post is brought to you by Eden, on a Via Eden, in the garden of Eden. :smile: