Follow Slashdot blog updates by subscribing to our blog RSS feed

 



Forgot your password?
typodupeerror
Check out the new SourceForge HTML5 internet speed test! No Flash necessary and runs on all devices. ×
IBM

IBM Creates New Processor Production Method 70

Vandermar writes: "IBM's new production method uses an advanced insulation material -- a low-k dielectric -- that protects the millions of individual copper circuits on a chip." Apparently it works at sizes down to 0.13 micron and the insulation itself is primarily silk. IBM says it will be using this technology for its Power 4 processor, but with their technology sharing with Transmeta and AMD can we expect to see this enhancement in their chips?"
This discussion has been archived. No new comments can be posted.

IBM Creates New Processor Production Method

Comments Filter:
  • by Anonymous Coward
    This is excellent news. Now mobile laptop transceivers can benefit form this new IBM technology. I believe this new class of semiconductor cores can power object-oriented library classes for REALLY fast real-time parallel API computing.
  • by Anonymous Coward
    Uh, the RC delay is less of an issue, the more important problem is that with reduction in gate dielectric thickness, the gate-channel voltage of a MOSFET -- the basic device in Silicon digital ICs -- could create very high electric fields thru the dielectric (i.e. the insulator between gate and channel) and hence result in electrical breakdown of the dielectric, destroying the transistor.
  • by Anonymous Coward
    I couldn't help but notice and be annoyed how these people never fail to talk about how faster chips is good for making the internet faster. after all, talking about the internet is the cool thing to do.
  • by Ec|ipse ( 52 )
    I'd like to hear for sure that Tansmeta and AMD really get to use it rather then it just being a 'what if'.
  • Hmm?

    IBM make these fantastically powerful Power CPUs. What're *you* going to plug it into if you ever managed to get your hands on one?

    You can't buy Power or PowerPC motherboards for love nor money.

    Don't expect to see the technology in commodity AMD CPUs until IBM have fully exploited it in their high end systems.

  • > http://127.0.0.1:98

    apt-get --purge remove linuxconf
    :)
    #define X(x,y) x##y
  • Power4 != PowerPC. They have a common ancestor, though.

    Impact is limited to AS/400 and AIX boxen.

    One of these days, we may see Linux or NetBSD ported to Power2. 'Til then, you pays yer license fees.

  • I guess I'm going to have to change my sig to keep from having to constantly answer all the people who don't get it. Funny how the nastier they are about revealing their inability to understand it, the more likely they are to be Anonymous Cowards.
  • That's been bothering me too lately (in fact I was going to post that, but you beat me to it :-).

    Intel's TV commercials seem to claim the exact same thing. Hey, use an Intel PIII Processor and your Internet experience will be so much faster! Never mind the fact that you have a 56k modem!

    To get anywhere these days you have to use all the right buzzwords, and Internet happens to be the most important.
  • >>but with their technology sharing with Transmeta >>and AMD can we expect to see this enhancement in >>their chips?

    Does it really matter? I mean, heck, normal processors seem to work just fine for me. What advantage do these chips have?
  • Yeah. Apple should convince Motorola to allow desktop PowerPC's to keep up with the market rather than hog glory/profits from it's partners. I for one would LOVE to run Linux on one of those fast G4's IBM has made. But motorola won't allow them to hit the market because they can't get their OWN process to work as well as third parties (IBM/AMD) have.

    Darwin is Open Source. It won't be long until MacOS X runs on compelling alternatives to Motorola.
  • Repetitive Silk Injury
    Pro: The workplace is air-conditioned, with free all-you-can-eat vegitation. Get to fly to work.
    Slashdot effect increasing value of silkworm labor market.

    Con: Worms can leave their "work-elipse" or go for a walk outside. Cant dare have a smoke break because of what a lighter flame does to silk. Not much work variety. Impossible to change jobs once work is started. Only get to fly to work once.
    Rayon industry is taking biellions of jobs.
  • Read the article. This is a low-K dielectric with the trade name "SILK" (probably an acronym).

    Yes. Good advice. Read the article. It's SiLK, not SILK.

    Even better advice would be to read IBM's non-Yahooed version [ibm.com].
  • do you suppose this will help apple/motorola/ibm's massive problems meeting demand on the G4 chips?

    or is this process something that the chip has to be specifically designed to use? or something that will take a long time to be put into usage?

    how much collaboration is IBM still putting into the powerpc line? are the rumors [appleinsider.com] that they've begun helping produce G4s any more than rumors? IBM was manufacturing ppcs at some point (before Altivec came out), weren't they..?
    This POWER4 thing looks very, very impressive..
  • Silk has been used for insulation since the invention of electricity.
    Up to the invention of plastics silk was the most used insulation material for copperwires.
    The only problem is, that after more than 50 years it decays.

  • Yes [w3c.org].
  • Um... every chip fabbed at IBM will have the option of using this tech. You only make money off of licensing if you actually let other people use it... and the proposition of using should draw more chips to the IBM fabs. So... it's not only the POWER/PowerPC lines.

    Think about it.
  • I believe this is subsidiary work, in that IBM has patents on various copper technology for microprocessors. I'm sure they've already filed ...

  • Not entirely true, actually. Sink is the woven cloth of threads made of unspun silkworm cocoons. True, the chrysalis is burned alive, but no actual silkworm is contained in it.
  • AMD is part of the AIM Alliance? Since when? The A part is for Apple... not AMD.
    --
  • You forgot to use the words "Brave New World" and "Information Superhighway"

    -Ekapshi.
  • This is an artificial polymer with a low dielectric constant. Not the silk used for cloth.

    Thank goodness, a non-vegan processor would set a very bad precedent. Imagine if leather was standard for mouse pads, yuck! (Silk isn't vegan because it's worms boiled alive.)

    For more on veganism, see http://vegsource.com/ [vegsource.com].

  • actually I wrote this from my personal experience doing mostly bleeding edge std cell design over the past 8 odd years - we used to put up with really simple wire models - basicly lumped capacitance (often just extracted with simple math from the net length) tossed onto the gate caps - now we need fully extracted models that take into account real RC delays along the wires as well as gate loading effects.

    5 years ago this worked great - then we started to need really good models of long inter-block wires, now days we see RC effects starting to dominate on long or branching intra-block nets - the sorts of things that Synopsys models with fanout based wire loads - this sort of modelling used to work just great now it's become a liability forcing us to either synthesize smaller blocks or to over estimate the wire related delays in order to cover the slow tail on the bellcurve of routed net lengths - both of which leave performance on the table.

    These process improvements help - I think more so as we head into smaller features (again because of edge effects capacitance is not scaling with area like everything else). More important is probably going to be CAD software I think the combined route/synthesis sort tools are going to have to replace thing liks Synopsys pretty soon if we are going actually use smaller feature sized processes really in anger

  • Everyone else had this posted this morning. Same for the Russian astronauts. I've noticed this happening with a lot of posts lately and I'm curious why Slashdot is so slow to post articles these days. This is interesting because a year or two back Slashdot was famous for being the fastest on the draw. The number of readers and editors has gone up so you would think it should be even faster nowdays. Is it because so many junk articles are submitted? Is it because the editors are now actually confirming press releases with companies like IBM? Is it because this article would have been rejected but the sheer number of submissions made Hemos change his mind?
    --
    Be insightful. If you can't be insightful, be informative.
    If you can't be informative, use my name
  • Their stock quote is NYSE:DOW [yahoo.com]... if i had a good deal of money, i'd throw some in there. Keep an eye on this one.

    Mike Roberto (roberto@soul.apk.net [mailto]) - AOL IM: MicroBerto
  • This product has been dubbed as the condom of processors and computer circuits.

    Mike Roberto (roberto@soul.apk.net [mailto]) - AOL IM: MicroBerto
  • Actually, the low-K dielectrics make the electromigration worse, not better. EM effectively causes pressure inside the 'downstream' (positive) end of the wire, where aluminum atoms pile up and press against the surrounding dielectric. All of the proposed low-K dielectrics are physically weaker than silicon dioxide, and thus the EM limit is lower.

    Do you think that this will effect the overclockability of chips produced with this technology?

  • Close but no cigar. Chips are damaged by electromigration and electrostatic discharge but not electro-static migration. So..

    Electromigration is basically electrons slamming at incredible velocities into the edge of the chip wiring and actually eroding the wires.

    Electrostatic discharge is basically what happens when you zap chips with static electricity which destroys CMOS transistors which means all the Good Stuff (tm) inside most computer chips developed today. It's why almost all computer chips are packages in anti-static bags.
  • It would also impress everyone if you read the article before you posted.

    :)


    Be thankful you are not my student. You would not get a high grade for such a design :-)

  • IBM's website [ibm.com] has more juicy stuff on Power4... Yum.
  • The POWER4 has nothing to do with the G4 which is developped as a follow-up for the PowerPC architecture. There is absolutely no agreement between Apple, Motorola and IBM about sharing anything since the Power2 (1994).

    So, for about a year, IBM will keep this technology closed and license it after they made enough cash. They have six months to make all the cash they want after the POWER4 is on the market and another six months to sign agreements with AMD, Transmeta, etc.

    Unless IBM is already a major sponsor of Transmeta, someone knows?

  • Cheap, if you bought a single package of 1 sq. inch of silk. But, prices may ramp up upon mass production of these "special" chips. Plus, I bet its pretty hard to form silk (whats the correct term?) to the dimensions of a microprocessor, especialy around its circuty. But, this could also end up bieng cheaper then previous methods of packaging, since Im not aware of the current price of the materials they use now.

    Systems Administrator
    Servu Networks
    http://www.servuhome.net
  • sig nazis suck
  • While it is true that IBM licenses a lot of its technology to other companies, licensing may very well be a non-issue in this case. As the technologies mature (and certainly 0.13um CMOS is not what I mean by "mature"), the lower-k dielectric might very well be incorporated into all of IBM's lines. In this case, it will be available to any company that uses IBM's foundry services without a "licensing" fee. While the use of SiLK will affect device models used for design simulation, the technology to implement it in manufacturing (which is what patents would likely pertain to) will be transparent to the chip designers.
  • by Anonymous Coward
    what, didnt you do that elementary school experiment where you rub your hard rubber rod with a piece of silk?
  • note to all the sig nazis: I'm posting this without checking the box that prevents the extra point from being added so this post will show up as a +2 instead of my usual +1.
    note to taco: Why don't you make posting without a bonus point the default and give us a checkbox for adding it, since according to the moderator guidelines we're supposed to use it sparingly.

    Tower, hope you check your user page for replies (and not just to see if your kama has gone up or down :-). Most of the moderators are decent people but as with any other community there are some that are on either too much or too little medication and some that are just plain jerks.
    Apparently, there are also a discouraging number that don't understand the meaning of the word "redundant".

    Maybe we should post with the bonus all the time to draw the fire of the "attack" moderators to protect other posters (karma's for burning, right?).

  • (would have replied sooner but I wasn't able to get Slashdot to load for several hours, even though other web sites seemed okay)

    If you go back to the first comment [slashdot.org] and my reply, you'll see that someone has, over 48 hours after it was posted, moderated my reply down as offtopic, even though it was directly related to the content of the first post. I'm pretty sure now that it's an organized effort. Whoever they are, they seem to strike about once every 10 to 14 days. From some things I've read elsewhere I'm pretty sure several others are being targeted as well.
    The funny thing is, about a month ago, after my karma had been sitting at 12 for probably half a year, with me not particularly worried about it one way or another, only checked my user page frequently looking for replies, I got up to about 16 over the course of 3 or 4 days off of a few posts. Then within 2 days the "attack moderators" had me back down to 12, and I got annoyed enough to sink to their level and got busy, posting often, and, despite one temporary setback a couple of weeks ago, I'm now high enough up the karma scale to be able to spare plenty.
    I'm thinking of discontinuing playing the game and scaling back my efforts in order to qualify to moderate again.
    The trouble with getting older is that being juvenile becomes wearying, especially with problems in the real world that actually matter to deal with.

    note to all the sig nazis: I'm posting this without checking the box that prevents the extra point from being added so this post will show up as a +2 instead of my usual +1.

  • I gotta admit, I haven't read the article yet, but plain jane silk certainly isn't the most durable substance on earth.


    Read the article. This is a low-K dielectric with the trade name "SILK" (probably an acronym).


    But one has to wonder why the article says "silk" and not "SiLK®" or at least "SiLK[tm]". Even after reading the sentence "IBM has developed a proprietary technique to build chips using silk, a low-k dielectric material that is commercially available from the Dow Chemical Co.," I still wasn't positive that they weren't talking about the stuff from silkworms. Nowhere in the article do they actually say the "silk" they're talking about is not what people normally mean when they say "silk". I mean, just because it's "commercially available from the Dow Chemical Co." that doesn't mean it can't come from a worm's butt.

    That said, you're right that they don't mean stuff from a worm's butt [www.imec.be]. You can't conclusively determine that just from reading the article though, so nyah.
  • I gotta admit, I haven't read the article yet, but plain jane silk certainly isn't the most durable substance on earth.

    Read the article. This is a low-K dielectric with the trade name "SILK" (probably an acronym).
  • silk the sexy material was not mentioned in the press release. SiLK the dialectric was mentioned in the press release.

    Also, the new techique for applying SiLK works ostensibly at any manufacturing size, not just .13u. The reason the press release mentioned .13u was IBM happened to also happened to announce a .13u chip the same day. Not related (but equally cool).

    Here's the non-Yahoo link:
    http://www.chips.ibm.com/news/200 0/0403_lowk.html [ibm.com]
  • I don't remember hearing about any agreement between IBM and AMD. This technology is intellectual property of IBM and would not be available to anyone unless IBM decides more can be obtain from licensing the process. The rumored talks between Motorola and AMD wouldn't necessarily benefit from this development because, while the two are part of the AIM PPC alliance they don't share much in the way of manufacturing processes. Sounds like good news to the PPC camp though.

    -Jon
  • Silk. As in the threadlike material that comes from silkworm bottoms.
    I had absolutely no idea that silk had applications to electronics. Go figure.


    "If ignorance is bliss, may I never be happy.
  • "...IBM has developed a proprietary technique to build chips using silk, a low-k dielectric material that is commercially available from the Dow Chemical Co. (DOW.N)."

    Note to self: Read article in future before posting. *selfthwack*

    "If ignorance is bliss, may I never be happy.
  • Licensing patents on stuff like this is IBM's usual MO, so yeah, they probably will start licensing it to all comers.

    It wouldn't make sense for IBM not to license the technology, for we all know that in this fast-paced industry, holding on to patent rights rather than licensing them tends to not give you a competitve advantage for very long. Someone will come along and invent something similar, but not infringeing before long...

  • Science News had a really good article [sciencenews.org] about the pursuit of new insulators for use in semiconductors last week.
  • Tell that to the silkworm. He's still being exploited.
  • IBM says the Internet is important. We need faster machines 'cause the Internet is too slow. We just might license this technology to other vendors so that they can help speed up the Internet...'cause the Internet is important. We also have some new paper manufacturing techologies. This is important 'cause people need to print things they find on the Internet. Did we happen to mention that the Internet is important?

  • This is correct. As an IBM microelectronics employee I can say:

    I don't remember hearing about any agreement between IBM and AMD.

    AFAIK, this is true. We (IBM) have no manufacturing relationship with AMD.

    This technology is intellectual property of IBM and would not be available to anyone unless IBM decides more can be obtain from licensing the process.

    This is also correct. However, something important does need to be pointed out though. This process will (AFAICT) be the next generation technology for IBM's processor and ASIC designs. Again, IIRC, CMOS-8S is our latest and greatest, so this process will likely be called CMOS-9(S|SF).

    This means that any customers using IBM for their manufacturing (and back-end) can use this technology, not just IBM. Since Transmeta uses IBM manufacturing, they have the option of using this process (if they choose to, right now I think they're in CMOS-8SF). Lou Gerstner has declared that there will be no sacred cows in IBM, so everything we make as a product must be available for sale outside the company as well. [Shameless plug: All the more reason for companies to go with us, because our silicon roadmap kicks ass].

    Anyways, as is public knowledge, the Power4 is currently using a modified version of CMOS-8S, which provides low runtime electrical failure rates. CMOS-8S technology is basically 0.18 micron junction and trace widths, using copper metal over insulator (SOI). For the final release, they will probably use this process (CMOS-9S?).

    Disclaimer: All this is public knowledge, none of it is IBM confidential stuff.
  • Unfortunately, someone modded you down for what I consider to be an opinion issue.

    Not all information wants to be free. Especially when it comes to microprocessor design. Code is one thing, which, by and large, can be created in one's basement with little overhead. Microprocessors, on the other hand, are designed and built with one thing in mind, to make money. Producing microprocessors is outrageously expensive and therefore, any technique that will enhance the quality of those processors is worth quite a good deal of money to those producing microprocessors.

    Even if the information was free, it wouldn't do any of us Slashdotters any good. Unless of course one of us has a couple of million dollars to drop on turning our basement into a clean room full of silicon etching equipment. ;)

  • I believe this is just IBM's SOI (silicon on insulator) technology, though I could be wrong. This technology is already being used in the G3 chips being used in Powerbooks, if I remember correctly.

    Yes, IBM was, and still is manufacturing PowerPC chips. The PowerPC archetecture is owned by the AIM (Apple-IBM-Motorola) consordium and is based off of the older IBM POWER archetecture. There was a split over the inclusion of Altivec, which is Motorola's technology to help to DSP in embedded devices and personal computers, but IBM continues to make high-end server chips.

    There will probably never be a Macintosh that uses the POWER4 chip, though. The POWER4 is a 64-bit PowerPC implemenation intended for high-end IBM servers. It is not meant to go in a Mac. The 500 MHz memory bus and matching RAM alone would drive the price of a Mac far beyond consumer's reach. It will also not include Altivec, which Apple is betting heavily on. We will likely as not be stuck behind x86 chips in power for awhile unless Motorola can get off their duff and fix the PPC 7400 production problems or get a newer, better chip out the door.
  • The whole interconnect / gate speed thing has been a cycle, and probably will continue to be (though it's getting to be more and more interconnect lately). You architect a system - all is good, and you are usually held back by your logic, not your wires. You respin your chip a few times, ramping up the clock speed with new technology, until your wires get in the way... uh-oh! Time to fix a few things an re-release this one (think P-II/III)...

    The newer the design, the better you can take advantage of all of the new technologies, and really gain the most... you have total control over your wiring, and most of your paths. If you need to stay fully compatable with a previous chip, and want to retain as much of the logic as possible (finally got that pesky FP thing working right), then you are going to suffer in how far you can take things.

    There's tons of overhead in a full redesign (person-years, dev costs, etc) - look at AMD with the Athlon (maintaining compatability, but redesigning most of the internals), but look at the gains in clock speed (more complicated in many areas than the P-lines, but easier to ramp up).

    There's quicker cheaper fixes, but they don't buy you as much at one time - here we have the PPro/PII/P!!!/P!!! Coppermine...... re-optimize a little bit and you get a good MHz percentage increase... but you have to make these changes more often. (there are changes to be made with any chip that needs to ramp up the clock - even the Athlon, though as of now, a lot of that can be attributed to process rather than design - a good place to be).

    The Merced (aka Itanium, aka PA-Risc w/IA-64) is a good step at breaking out of a rapidly dying ISA (though Willamette shows some promise at extending the life of this storied line even further), but the delays there have really soured some people on it, and many are looking at the next generation Alphas and the next POWER chips from IBM (and PowerPC line, of course). There's a lot of good 64 bit solutions out there, and there's no doubt we should continue to push forward, but I'm doubting that Iced/Mertaium will get to market in time for its performance point...

    oh well...
  • Hehe - actually, I thought your .sig was pretty funny...

    I've stopped posting at 2, after three or four days where every post I sent at 2 was immediately marked Overrated... even if it was later modded up for other reasons... kind of a pain.

    I still see people complaining "Why did this crap get up to 2, it sucks!" and then several other people explaining the bonus... this community is getting less informed as more people jump in without a little reading... like another comment on this story - "Proposed New Slashdot slogan: Read First, Post Later" It might actually help...

    (now posting with a +1 to see if I'm still being targeted - and I didn't think I deserved any attention 8^D)
  • I always check for replies (my karma level is happy enough to be irrelevant most of the time, though I am hurt to see my posts at 0 8^D)...

    I've suggested the reverse option for the +1 bonus myself... not sure if things have fallen on deaf ears, or if somebody really has a good reason. I agree with most moderation (I don't find nearly as many items to mark 'unfair' in M2 as there used to be), but every once in a while, somebody gets targeted (I had a lot of AC flames that same week for karma whoring (what a surprise))... Started to feel like Sig, Enoch, et. al.... Probably just pissed somebody off with one of my comments 8^)

    Moderation is a Good Thing, though... it's been very helpful (haven't seen a ninja/natalie/ascii art/smut story post in a while (except when somebody mods them up as "funny", which is pretty unusual).

    I think that *is* the smell of burning karma ;-)
  • Has anyone ever noticed that Intel's commercials about how super fast their machines are really show slow systems? I don't know about you guys but when I watch these commercials I don't really see anything especially dazzling about their system performance! Who hasn't seen a video running on their computer, regardless of CPU? Who hasn't seen 3D graphics? Hell, the 3D stuff they show looks pretty much like any other computer I've ever seen, and we're talking about low-end systems. If they want to showcase real-world speed (instead of wild panoramic camera angles in space or whatever Intel does which has nothing to do with actual desktop system performance), they need to shoot stuff like old Apple commercials. Apple used to swing the camera around to a PowerMac monitor where all kinds of windows/video/content/gfx were flying around like a cyberflick on steroids. Oh, and you guys notice the @Home cable modem commercials on TV? Yes, the Internet sure is quick but when they show a video on the monitor in the commercial it looks like 5fps crap! Why do marketing guys let this stuff get by? Did they want it slow to make it look "computerized?" Is their marketing hardware supposed to look slow on TV or is that how sluggish the stuff is? Sorry if I sound arrogant; I'm on a G4/500 =)

    -----
    Linux user: if (nt == unstable) { switchTo.linux() }
  • Do you think that this will effect the overclockability of chips produced with this technology?

    EM is a lifetime-reliability issue. Overclocking basically affects EM in two ways:
    • More clocks: the EM life of the device gets used up faster. If the EM life was 10 years (a common target) then 25% overclocking will reduce it to 8 years
    • Temperature. EM is hideously nonlinear with temperature. Assuming that your overclocking strategy involves overcooling to keep the die temp unchanged, this won't matter./UL
  • Yet again we see the results of the IBM investment in basic research, especially in regards to copper for circuitry.

    What are the impacts? It's not so much the smaller processor or the faster speeds - it's the lower power consumption that will cause the most dramatic change. That plus the lower cost to manufacture.

  • 2000-04-03 13:47:25 IBM's plan to make 30% faster chips (articles,ibm) (declined)

    I posted this article with the link to the press release yesterday at 1 pm EST. WTF is wrong with /.'s sorting of story postings?

    Hemos, I'm glad you picked the story, but if I were you, I'd look to see who's sorting your submissions.
  • One of the real beauties in this is in production is existing chips. Testing of manufactured chips can be as much as 70% of their cost (high rejection rates due to fused wires, the aforementioned crosstalk, etc.) It would certainly be nice if this could lower the chip rejection (meaning more chips/batch, higher supply, lower prices) on older, but still useful, chips (DIMMs, M68K's, FPGA's).

    of course, the cynic in me says the companies would just absorb any savings as profit. :)

  • This method may help lengthen the life of microprocessors. As you know, CPU's are damaged by electro-static migration, and heat, maybe this will in some way prevent some of that damage. Lets just hope it doesnt raise the cost of the chips (isnt silk pretty expensive?

    Systems Administrator
    Servu Networks
    http://www.servuhome.net
  • It looks like most of the comments will be based upon the wording of the description and not the text of the article:

    silk, a low-k dielectric material that is commercially available from the Dow Chemical Co.

    Proposed New Slashdot slogan: Read First, Post Later

  • That's pretty much my impression of the Slashdot community. A good majority of you guys consistently fail to see the big picture and instead choose to find faults in the articles or what other people have said.

    As I scroll down the list of posts, I see the occasional post that actually discusses the issue at hand, but the majority of them are composed of arguments between two people or maybe a relevant discussion of how to pronouce "Linux." I suppose I could set my threshold level higher, but then I only can see what others think is worth reading.

    As an aside, I'm in the Computer Engineering program here at MSOE, and I see the same traits in my CD classmates. A good portion of them, while being intelligent people, can only focus on the small details at hand, rather than being concerned with the grand scheme of things, which is usually the more important thing to be concerned with.

    David Peters

  • by Christopher Thomas ( 11717 ) on Tuesday April 04, 2000 @02:00PM (#1151078)
    "...IBM has developed a proprietary technique to build chips using silk, a low-k dielectric material that is commercially available from the Dow Chemical Co. (DOW.N)."

    Probably an acronym of SIlicon, Low-K.
    The approaches I've read about to date mostly involve "foaming" whatever material is used, as vacuum or whatever gas ends up in the cavities doesn't have nearly as high a dielectric constant.

    Parasitic capacitance is directly related to the dielectric constant of the insulating material.
  • by vitaflo ( 20507 ) on Tuesday April 04, 2000 @02:04PM (#1151079) Homepage
    I think a big question here is, does this impact Apple? With Motorola holding up the release of faster G4 processers, etc, and with IBM just coming up with blazing technology such as this, where does Apple fit in? Obviously, if Motorola had it's act together, this IBM announcement would be awesome news for Apple, but now, it doesn't do a whole lot of good for Mac users.
  • by overshoot ( 39700 ) on Tuesday April 04, 2000 @02:29PM (#1151080)
    As you know, CPU's are damaged by electro-static migration, and heat

    That's either electrostatic discharge (zapping the part) or electromigration (wiring atoms getting knocked out of place by speeding electrons). Please don't combine the two.

    Actually, the low-K dielectrics make the electromigration worse, not better. EM effectively causes pressure inside the 'downstream' (positive) end of the wire, where aluminum atoms pile up and press against the surrounding dielectric. All of the proposed low-K dielectrics are physically weaker than silicon dioxide, and thus the EM limit is lower.
  • by overshoot ( 39700 ) on Tuesday April 04, 2000 @02:45PM (#1151081)
    One possible problem is poor heat conduction from die to package, but the tradeoff is probably worth it.

    Non issue. IC dice mount with the back of the die, not the active side, to the substrate. Even flip-chip (solder-bump) parts have to use a thermal contact on the other side for all but the lowest power dissipations because the total contact area on the metalized side must be a small percentage of the total.
  • by taniwha ( 70410 ) on Tuesday April 04, 2000 @02:08PM (#1151082) Homepage Journal
    Remember how copper was a big deal about a year back .... well this is really the same thing - wire delays in chips are rapidly becoming the dominating factor in how fast they can go.

    Basicly delay is proportional to R*C - the resistance of the wire times the capacitance of it. You can reduce the resistance by using a more conductive metal (like Cu rather than the more traditional Al). You can reduce the capacitance by:

    • reducing the wire's area (but 'edge effects' - proportional to the length of the wire don't scale in the same way as other features so this is becoming problematical)
    • moving the wire further away from other wires (but we're trying to make things smaller to fit more stuff in, not bigger)
    • use an insulator between the wires with a lower dieletric constant (what's happening here)
    5 years ago we designers didn't care much about wire delays - they weren't what made our logic slow - now they're killing us .... this sort of stuff is great news!!

    PS: Intel's 'coppermine' processers don't use copper wires - marketting is everything ....

  • If the technique is proven viable, IBM will surely get their patents in a row and start licensing the technology to all comers. Just because they have tech sharing with Transmeta and AMD doesn't mean that Intel, Motorola and Via are not going to be able to get their hands on this. IBM's revenues are not solely from hardware sales, and certainly not from chip sales alone. It is in IBM's best interest to license the patents out to whomever will pay for them.

    So, on that note, I do believe that Transmeta and AMD will end up with this technology, assuming it is useful.
  • by Christopher Thomas ( 11717 ) on Tuesday April 04, 2000 @02:19PM (#1151084)
    From http://www.dow.com/dow_news/co rporate/20000403a.html [dow.com]:

    The Dow Chemical Company is supplying IBM with SiLK* semiconductor dielectric resin

    This is an artificial polymer with a low dielectric constant. Not the silk used for cloth.

    The article also gives a moderately technical description of why a low-k dielectric is a Good Thing.
  • The industry's been working on low-k dielectrics
    for a long time.

    The real issue with low-k dielectrics is that they
    reduce the capacitance of the onchip wiring.

    This has several benefits:

    1) Wiring capacitances in general will be
    signficantly reduced.

    2) Wires running parallel to each other will
    have less crosstalk. I.e., if the voltage on
    one wire switches, it'll have less tendency to
    drag the voltage of wires close to it with it.
    This is a result of neighboring wires having
    mutual capacitance. --> EASIER DESIGN

    3) Since capacitances are reduced, less power
    will be needed to cause metal wires to switch
    voltage. --> COOLER CPUs, -->LOWER POWER

    4) Because of the reduced capacitances, it requires less current to switch wires.
    --> FASTER CPUs.

    This is a really significant advance by IBM,
    maybe more signficant than copper metallization.

    One possible problem is poor heat conduction
    from die to package, but the tradeoff is probably
    worth it.
  • by mcteague ( 160577 ) on Tuesday April 04, 2000 @02:07PM (#1151086)
    SiLK [dow.com] is a commercial product from Dow [dow.com]:
    "SiLK Semiconductor Dielectric resins developed by The Dow Chemical Company are new, low-k polymers for use as an interlayer dielectric (ILD) material for high-performance integrated circuits.

    SiLK Dielectric resins offer the lowest commercially available dielectric in the industry." (From Dow's SiLK website.)
    I read a much better, more technical, article somewhere yesterday, but I can't find it. Sorry.

Money is the root of all evil, and man needs roots.

Working...