ToshiroOC

Before I start, keep in mind that what I am saying requires backgrounding. I have set the quote in context. I do not want to be denounced as a lying newbie, so I am giving the full story rather than a stupid one liner consisting totally of the thread title.

I was recently invited to go to a technology camp in Austin, Texas (www.theinternationalsummit.com). To keep a long story short, I went and I got the chance to go on the International SEMATECH site visit (www.sematech.org). Its basically a semi-conductor R&D non-profit company that is funded by over 50% of the semi-conductor market, including these companies:
AMD
Agere Systems
Hewlett-Packard
Hynix
Infineon Technologies
IBM
Intel
Motorola
Philips
STMicroelectronics
TSMC
Texas Instruments
It essentially works by having all of this companies chip in to afford silicon research that would be cost prohibitive otherwise, then all of the companies share the fruits of the research. I was being given a tour by someone who shall remain anonymous, but was involved in engineering. I learned that engineers from all companies and in some cases different countried worked cooperatively here to refine and pioneer silicon design and production processes. I saw a 248nm laser with associated silicon machinery. I asked and I was told that you take the laser wavelength and you divide by 2 to get the process size (so the 248nm laser was a 124nm process theoretically, and realistically you round that up to about 130, or a 0.13micron process), and he said that they had the world's first 157nm laser and had recieved shipment of a 193nm laser earlier that week. He told me that the 193 was going to me for 0.09micron processes, and that after a few 0.10 runs they should be able to get it down to the predicted 0.09. I said that there had been rumors that Intel had already managed to get prescott silicon running on a 0.09 process, and the response was "Oh yeah, Intel is way ahead of us on production for this one". I believe this is pretty interesting, because along with the 0.09 rumors about intel, this could mean that 0.09 tech is already feasible and ready for a release sometime relatively soon. Anyone have questions, comments, flames, concerns, etc.? I'd be interested in seeing what y'all think of this...


ToshiroOC

BTW, he also said that after 157nm lasers they were going to move over to 13.5nm lasers since they were easier to make than the ones between 157 and 13.5. That is considered extreme ultraviolet range, and is a whole new challenge. SEMATECH works on technology that is either in current use and state-of-the-art all the way out to 10 years ahead, which is around where the 13.5 tech is. I also got a chance to visit the AMD Fab25 in Austin, but my group was restricted to the PR rooms and auditorium. During their presentation they showed us pictures of their SOI process, showing that they had the silicon substrate down to three molecules... I didn't really understand how that worked, so I'm not going to say anything more. I also got to meet a very tight-lipped engineer who was helping Hammer Logical Design (vs. physical design/production) Research who was very cool...

__________________
[img][/img]
[color=White]Peace be with you, Joe.[/color]
Driverheaven Staff Member (Supermoderator)

shai

they have a working 0.03 so what

ToshiroOC

Who is "they"? SEMATECH? Where do you get this wonderful information? Could it possibly be from the bullshit fairy, who comes down from fairy land and fills forums with pointless one line pieces of utter crap? No one in the entire web has ever once mentioned a 0.03 process. Once we get there, chances are it will be referred to in nanometers (nm), not microns, as 0.03 is. 0.03 is extremely unlikely since the silicon imprinting lasers are all at special wavelengths, and different elements in the laser produce different wavelenghts. There are no known combinations of elements that could practically be used in a powerful laser to produce anything under 157nm right now. 13.5nm lasers I believe are so far only prototypes, if not merely theoretical, and a 0.03 process would require at least a 60nm laser, which does not exist. There is no existing 0.03 process. Do not do another stupid one line post for the sole purpose of raising your post score. If you go back and read all of my posts, I always says what I want to say in detail and I don't waste my time or other's time with idiotic bullshit one liners.

ToshiroOC

__________________
[img][/img]
[color=White]Peace be with you, Joe.[/color]
Driverheaven Staff Member (Supermoderator)

shai

they have one yes one chip 10 ghz one and also one bot that can fix cpus (takes days)
and its possible below 0.03 its not

ToshiroOC

Wait, sorry, I realized that you might mean that you are simply ignorant of what a 0.09 process is, and you simply hit a key six numbers away from the one you intended (a 3 instead of a 9). The Willamette cored P4s were on a .18 process and ran very hot and only sorta fast. The Northwood cored P4's were on a .13 process and scaled approximately 50% higher than the Willamettes. The upcoming Prescotts, if they follow this scale, should be able to reach 4.5ghz, since the Northwoods can hit around 3ghz, and another 50% increase in speed would be 4.5ghz. This is exciting for the hardware community because it helps us know that Intel, hated or not, has more fast hardware on our way and keeps in the spirit of the driverheaven.net forums, which so far has had a reasonably strong emphasis on upcoming products (just look at the front page, with all of the information regarding the r300, a so-far unreleased product, and look at all of the XPSP1 leaks, a so-far unreleased service pack, and you can see that a forward-looking community is strong here in driverheaven.net). This is to help feed the community information that it wants. This is interesting because it gives us a peek at what is coming up. So next time, don't say "so what", explain this to the total newbie who asks the exact same question.

ToshiroOC

__________________
[img][/img]
[color=White]Peace be with you, Joe.[/color]
Driverheaven Staff Member (Supermoderator)

shai

im the "Tweaking god" i know all

now sirusly a 10 ghz is at intel biggest factory again only 1
and the 0.09 micron process should be here by next year

ToshiroOC

Yes, they have one known ALU that is 10ghz that anandtech.com covered in their Intel coverage. The thing is, a 10ghz ALU does not mean 0.03 tech, it just means that Intel can scale their ALU's much much higher using current technology without other parts of the processor to keep the clock speed of the ALU to scale to 10ghz. For all we know, the cache is the limiting factor in P4's nowadays, and without any cache the P4's would be capable of a crippled 5ghz. "They" (probably meaning Intel from your context) do not have a bot than can fix CPU's. They do have engineers that can use a Focused Ion Beam system to change prototype silicon, but those are irrelevant to what I was talking about. Fixing a piece of prototype silicon with a "bot" or with a bunch of engineers is unrelated to the conversation. "and its possible below 0.03 its not".... I will take that to mean that its impossible below 0.03 but otherwise its possible... well, you are wrong again. 0.03 is possible, just not yet with today's technology. Lower than that is also possible, and you can continue until the transistors are smaller than the wavelength of an electron, in which case the transistors cease to transist (their semi-conducting properties stop), and at which point different technology from silicon will be needed to continue Moore's Law.

ToshiroOC

__________________
[img][/img]
[color=White]Peace be with you, Joe.[/color]
Driverheaven Staff Member (Supermoderator)

shai

i heard once that you cant make it below 0.03 i forgot why sorry

ToshiroOC

"im the "Tweaking god" i know all" wow, since you can edit your title that must mean you know more than me! Your post count in these forums quite obviously trumps any real knowledge carried outside of these threads, and I bow down to you. (wow, did I sound sarcastic there?). I am aware of the 10ghz ALU at Intel, but that us just an ALU, nothing more - it doesn't not have any other parts of a processor which add great complexity to the design and invariably make the peak clock speed lower because the more complex the design the more chance of error you get when running at overly high clock speeds. The 0.09 process might be here next year, its hard to say, but I am posting this information so people can know that the process probably exists right now and though that is not directly relating into a date or time for a 0.09 release of a processor, it is nice to know that the tech exists and is possibly ready for prodution.

ToshiroOC

__________________
[img][/img]
[color=White]Peace be with you, Joe.[/color]
Driverheaven Staff Member (Supermoderator)

shai

dont get mad
and i saw that bot it does fixes cpus with no help but takes days

DallasStar

Toshiro... I think its about time for you to realize that shai is just playing with your mind ....don't take things so seriously...

__________________
"A picture of my existence... would show a useless wooden stake covered in snow... stuck loosely at a slant in the ground in a ploughed field on the edge of a vast open plain on a dark winter night." --- Franz Kafka

ToshiroOC

Well, don't believe everything you hear. And if you just hear it once from a source that isn't very reliable (as it obviously wasn't), then you should question it and not state it as fact. For the past 50 years or so we have been saying that Moore's law will stop after 10-15 years, but we have always innovated and pushed that deadline back, but once the laws of physics intervene in our innovation and stop silicon from semi-conducting at the sizes we want, we will need to change. 0.03 is not the point where the laws of physics will prevent us from going further, but 0.03 might be the limit for today's technology, and will be pushed further with the next innovation EUV (extreme ultraviolet) Lithography, which will start to get into the extremely small and microscopic.

ToshiroOC

btw, im going to bed now. I'll be back at this thread for more debate tommorow 7/17/02

__________________
[img][/img]
[color=White]Peace be with you, Joe.[/color]
Driverheaven Staff Member (Supermoderator)

ToshiroOC

Sorry, just I left the [H] and the madonion forums to avoid people who do thoughtless one-line posts and then leave. Shai wasn't that bad in that he at least stuck around to continue debating his one-liner, but it still boils my blood seeing overly short posts, esp. when I percieve them as incorrect... my bad. I'll try and keep my cool next time...

ToshiroOC

__________________
[img][/img]
[color=White]Peace be with you, Joe.[/color]
Driverheaven Staff Member (Supermoderator)

ToshiroOC

Like I said above, sorry about getting mad, but its the one-liners that piss me off so much... About the "bot" though - It needs human help to do it's job, since its humans that are designing the processor, not any machine, and so humans need to tell the machine what changes to make to the silicon - the FIB machine can't find flaws in the silicon without human intervention and it cannot decide on how to fix something without human intervention. It can change the physical surface of the silicon according to the engineer's modifications without human intervention, as far as I can tell and know about it...

ToshiroOC

I'm actually leaving this time. Until tommorow.
EDIT: Spelling on tommorow was incorrect.....

__________________
[img][/img]
[color=White]Peace be with you, Joe.[/color]
Driverheaven Staff Member (Supermoderator)

DallasStar

A little off topic:

I remember once over at rage3d, I got in an argument with Cyborg007 about grammar. I eventually got very tired because it was getting late, so I kept telling him that it was getting late and that i was leaving, but I JUST COULDN'T GET AROUND TO IT.... I just wanted to argue argue argue (and ultimately win the argument)... Then, eventually, at around 4:00AM, I decided that the argument had gone on long enough. It was fun while it lasted though

__________________
"A picture of my existence... would show a useless wooden stake covered in snow... stuck loosely at a slant in the ground in a ploughed field on the edge of a vast open plain on a dark winter night." --- Franz Kafka

shai

what debate