Become a fan of Slashdot on Facebook

 


Forgot your password?
Close
typodupeerror
×
AMD

AMD's Next Generation Processor Technology 320

Posted by CmdrTaco from the chugging-the-numbers dept.
Esekla writes "AMD has released info about their upcoming processor technology. The press release claims that they're producing circuits that run 30% faster than any other published benchmarks using "Fully Depleted" Silicon-on-Insulator and AMD's metal gating technology and actually has a good bit of technical detail for a press release."
This discussion has been archived. No new comments can be posted.

AMD's Next Generation Processor Technology More

AMD's Next Generation Processor Technology

Comments Filter:

  • Excellent... (Score:2, Offtopic)

    by ceswiedler ( 165311 ) *
    If nothing else, this should pump up my AMD stock.

    Does anyone know if this is press-release hype or a real breakthrough? I'm not a semiconductor expert. But my suspicion is, real breakthroughs generally don't get announced in marketing press-releases on Yahoo Finance.
    • Does anyone know if this is press-release hype or a real breakthrough?

      Neither - it's incremental improvement. That's how most progress is made.

  • Metal gates? (Score:5, Interesting)

    by sleepingsquirrel ( 587025 ) * <{Greg.Buchholz} ... ingsquirrel.org}> on Thursday June 12, 2003 @01:38PM (#6183570) Homepage Journal
    I'm not a process guy, so could someone explain why they're claiming metal gates are better? I was under the impresson that metal gates were more compatable with high-k gate oxides, but I didn't see any mention of non-SiO2 dielectrics. And on that note, does anyone know if AMD is trying out any low-k dielectrics for the interconnect?
    I also noticed that one of the lines in the slide said something to the effect of, "Mesa isolation was used to keep things simple". Does this mean that they just did that for the one test wafer to keep things easy, but it'll be no problem once we get the process into production? Or are we talking about something that's still many years in the future?

    • Re:Metal gates? (Score:2, Insightful)

      by Aneurysm ( 680045 )
      I can imagine initially much more expensive chip,s because the chances of the chips being produced at existing plants using existing equipment is pretty slim, so new manufacturing plants will need to be built, or at a minimum modified
      • You mean kinda like then need to modify fabs every time there's a size change now? Wanna go to 90 nm, re-tool the fab, wanna go to 60 nm, re-tool the fab. It happens all the time now, and it's not gunna stop anytime soon.

    • Re:Metal gates? (Score:5, Informative)

      by FuzzyDaddy ( 584528 ) on Thursday June 12, 2003 @01:57PM (#6183771) Journal
      I'm no longer in the CMOS biz, but let me take a stab at it.

      Polysilicon has been the gate material of choice because it is much easier to process. However, metal would reduce the resistance of the gate. (The gate acts like a little capacitor, and the resistance of the gate affects the amount of time it takes to charge up and discharge, which affects the switching time.) I think the processing ease of Polysilicon is lost when you don't use Silicon dioxide as the gate material - for example, if you used a high-K dielectric. I don't know if metal is inherently more compatible with high-k materials, just that it's less compatible with SiO.

      They also mention the metal gives a "tunable work function" (probably by adjusting the silicon/nickel ratio), which I would guess would change the turn on voltage of the transistor. Tuning the turn on voltage could certainly tweak up the speed a bit.

      • metal work function (Score:4, Interesting)

        by sleepingsquirrel ( 587025 ) * <{Greg.Buchholz} ... ingsquirrel.org}> on Thursday June 12, 2003 @02:11PM (#6183929) Homepage Journal
        I probably need to crack my physics books for this, but I thought the work function of a metal was the amount of energy needed to free and electron from the metal (a la, the photo-electric effect). So I don't see how that could possibly have an affect on the transistor action. Any physics students out there?
        • As I recall, the work function measures the distance from the vacuum energy(the energy where electrons are freed from the metal) to the conduction band(the energy level where electrons conduct). Changing the location of the metal conduction band relative to the semiconductor conduction band changes the characteristics of the semiconductor.

        • Clear up some misconceptions (Score:5, Informative)

          by siskbc ( 598067 ) on Thursday June 12, 2003 @03:50PM (#6184787) Homepage

          You're quite right, you can't change the work function of a pure metal - but if you have a blend of materials, they will have to equilibrate, as the energies of the electrons in one material will have higher energies than the electrons in the other. Therefore, electrons will move from one material to the other like water flowing downhill, until the average energies of the electrons in the material are uniform between domains (or atoms) of the different materials. This yields a single Fermi level, which is described as the average energy of the electrons in the material. By varying the quantities of the materials (here, nickel and silicon), you can change the fermi level of the material, thereby changing the work function of the material. So, while you can't change the work function of a pure metal (you'd have to apply an impossibly obscene amount of charge to do so), you can make different blends.

      • Re:Metal gates? (Score:5, Informative)

        by sarpedon77 ( 159241 ) on Thursday June 12, 2003 @02:43PM (#6184235) Homepage
        Metal gates have 4 main advantages in advanced CMOS transistors:
        (1) The gate resistance is reduced. This lowers the switching delay in some cases. Remember that the delay is proportional to the product of the resistance and capacitance (the 'RC' product).

        (2) In polysilicon gates, the free carrier density is very high (1E20 carriers per cubic cm). Even so, under high electric fields that are needed to switch a transistor, there is a small depleted layer created right at the interface of the gate and the dielectric. This effectively acts as a capacitor in series with the dielectric and increases what is called the "effective oxide thickness". This is very bad, especially when process engineers are trying extremely hard to reduce the oxide thickness. At the scales we are at now, every Angstrom counts. In metal gates, the carrier density is 1000X higher. This makes it much harder to deplete and you regain the 4 angstroms. This means either higher performance with the same gate dielectric thickness, or you can get the same performance and increase the dielectric thickness by 4A, thereby reducing the gate tunneling leakage current (and hence power) by an order of magnitude. This is a big deal.

        (3) Some high dielectric constant materials (that are candidates to replace silicon dioxide) are not very compatible with polysilicon. This could mean either thermodynamic instability or interfacial charge created that "pins" the workfunction (and affects the switching threshold voltage of the transistor)

        (4) In fully-depleted silicon on insulator (FD-SOI, or "depleted substrate transistor" in Intel parlance) transistors, the threshold voltage comes out wrong when using doped polysilicon gates. It makes the transistor either too slow or too leaky. There is a desperate need for tuning the threshold voltage by using a different workfunction which can be found in some metal gates.

        Of course, metal gates aren't without their problems. (the predecessors of today's transistors had metal gates - hence the 'M' in CMOS - Complementary METAL Oxide Semiconductor - which were replaced by polysilicon gates for processing ease.) Inability to be easily patterned, withstand high processing temperatures, reliability issues are just a few of them.
    • It looks like the short answer is that the poly doesn't get as many dopant ions down close to the gate oxide, which results in an effective reduction of oxide thickness. Therefore, if the poly is replaced by SiN there will be metal all the way down to the oxide and the electric fields will be higher, which means a better transistor. Two good papers...

      Dopant profile and gate geometric effects on polysilicon gate [stanford.edu]
      Gate Length Dependent Polysilicon Depletion Effects [stanford.edu]

      Also EETimes [eetimes.com] has another interesting article with more information about AMD's presentation at the 2003 Symposium on VLSI Technology in Kyoto, Japan.

  • mmm... more speed... (Score:3, Funny)

    by blitzoid ( 618964 ) on Thursday June 12, 2003 @01:40PM (#6183598) Homepage
    Let's hope that these new chips are as inexpensive as current AMD processors.

    Or at least as heat efficient! (Badum-dum psshh)

  • If only... (Score:4, Insightful)

    by Fry-kun ( 619632 ) on Thursday June 12, 2003 @01:40PM (#6183599)
    if only they started *producing* those chips 30% faster...
    well, one can only hope...
    • Indeed. I fixed the computer of a woman who fixed/installed mainframes (IBM I believe). She mentioned to me that one of her last jobs was to install a mainframe at intel (sometime around '88) which they were using to debug the first prototypes of the Pentium. This was around the time their first prototypes were produced.

      That was a LONG time before the pentium was released. Don't hold your breath
  • Like, how long do we have to wait for the Silicon to deplete?

  • I/O Speed Please (Score:5, Insightful)

    by jabbadabbadoo ( 599681 ) on Thursday June 12, 2003 @01:42PM (#6183615)
    As a fellow /.'er has already indicated, processor speed improvements is very exiting. What I wanna see is a yearly increase of 30% on I/O speed. I'd rather have a super-fast bus and a new 50-ns-access-time storage technology than a 10 GHz processor.
    • "is very exiting"

      I meant, NOT very exiting...

    • Re:I/O Speed Please (Score:5, Insightful)

      by Jeffrey Baker ( 6191 ) on Thursday June 12, 2003 @01:51PM (#6183705)
      Opteron already has an excellent memory subsystem and fast paths to PCI-X peripherals. Aggregate I/O and memory bandwidth in 4-way Opterons is pretty sick, and although it won't compete with insane systems from IBM and SGI, it is a lot better than anything else you can get in an $8000 box. What were you hoping for?
    • I don't know who modded you insightful, but you are not. You state the obvious.

      Amd would really like to have what you describe, but they don't. I/O will only limp behind processing speed, as it has been since the 486. In the mean time, those cycles can be used for other things, such as datacompression of the data that goes to memory.

      However, latency is what is killing in the end, not the data transferrates. That is what we have cache for, but it doesn't yet have 100% hitrates yet. The fun comes when there

  • depleted silicon? (Score:5, Funny)

    by konichiwa ( 216809 ) on Thursday June 12, 2003 @01:43PM (#6183624)
    Wow, they must have circumvented UN Resolution 1441 in buying that depleted silicon from the depths of Niger's black market.

    WAR AGAINST AMD
  • this benchmark improvement thing rings [slashdot.org] a [slashdot.org] bell [slashdot.org].
    XD

  • Actual speed (Score:3, Funny)

    by Rosco P. Coltrane ( 209368 ) on Thursday June 12, 2003 @01:48PM (#6183679)
    they're producing circuits that run 30% faster

    Not to worry, the next generation of Windows will no doubt be that much slower.

    • Actual speed doesn't change when bloat happens. (Score:4, Interesting)

      by djh101010 ( 656795 ) on Thursday June 12, 2003 @02:11PM (#6183924) Homepage Journal
      I've got a Computer Shopper in front of me from 1993. On the cover is a reasonably high-end system, for 1500 bucks. Today, one can buy a reasonably high-end system for 1500 bucks.

      At the time, it took a couple of minutes for windows to boot, on a 486-33. Today, it takes a couple of minutes for Windows to boot, on say a 1.6 GHz P4. Yes, it's doing a lot more, but it's taking just as long as it did a decade ago.

      • Re:Actual speed doesn't change when bloat happens. (Score:5, Insightful)

        by rabtech ( 223758 ) on Thursday June 12, 2003 @02:22PM (#6184025) Homepage
        You must have a really crappy system then, because my WinXP workstation goes from power-on to logon in about 20 seconds total. That's a far cry from the 3 minute bootups of yesteryear.

        And FYI: you can build a reasonably fast system for less than $1,000, whereas a decently fast system in 1993 ran more like $1,500 - 2,000.

        You can build a more top of the line system for $2-4k these days, whereas a top of the line system in 1993 ran more like $3-6k.

        Computer people suffer from "The Good Old Days" syndrome just as much as everyone else.

        • Re:Actual speed doesn't change when bloat happens. (Score:4, Insightful)

          by mungtor ( 306258 ) on Thursday June 12, 2003 @02:55PM (#6184336)
          You must have a really crappy system then, because my WinXP workstation goes from power-on to logon in about 20 seconds total. That's a far cry from the 3 minute bootups of yesteryear.

          Yeah, but how long until it actually logs in? That's a typical MS gimmick. They only measure from power on to logon prompt appearing.

          It was incredibly obvious on NT 4.0 workstations. The logon box pops up, but the TCP/IP stack isn't even up yet. You get to type your login info 45 seconds after power on, but you still can't use the machine for another 90. Longer if you have to wait for all it's system tray stuff to load (chat clients, anti-virus, etc).

  • NetBSD (Score:2, Funny)

    by KillerHamster ( 645942 )
    Has NetBSD been ported to it yet?

  • Working together to defeat Intel (Score:5, Informative)

    by ikewillis ( 586793 ) on Thursday June 12, 2003 @01:52PM (#6183717) Homepage
    It's nice to see AMD, IBM, and Apple working together to defeat Intel.

    As you may or may not know, IBM originally developed Silicon-on-Insulator technology and licensed it to AMD. Here is the whitepaper: http://www-3.ibm.com/chips/bluelogic/showcase/soi/ soipaper.pdf [ibm.com]

    This is the same technology that was used to make the Power4 processor, and will also be used to make the upcoming PPC970: http://www-916.ibm.com/press/prnews.nsf/jan/06C1F2 11F9B1C24B85256ADF006163AF [ibm.com]

    AMD has recently built a new state-of-the-art fabrication facility in Dresden to produce the chips, known as "Fab 30": http://www.anandtech.com/cpu/showdoc.html?i=1773 [anandtech.com]

    I hope together IBM and AMD will continue to update their manufacturing process to keep on par or perhaps once again surpass Intel.

    • Yes, thank god someone if fighting the evil Intel. That evil corporation who have made supercoomputing power availible on the cheap. I don't know of anything predatory that Intel has done beside making good products.
    • I like AMD as much as the next guy (running an 1800 XP), but I'm not sure why Intel needs to be defeated... good company, good products.

      • I like AMD as much as the next guy (running an 1800 XP), but I'm not sure why Intel needs to be defeated... good company, good products.

        Intel doesn't need to be defeated, just "competed".

        Intel (and every other company) simply needs to be in competition, in a hotly-contested race to produce high quality products for the lowest price in a well-informed marketplace

        Absence of competition permits, even encourages companies to produce lower quality products because they can charge high prices for them [1[PD [ucl.ac.be]

      • Well, they are slightly evil... I think the Rambus Saga a while back showed that pretty well.

        They force expensive, unwanted, patented tech. on the public, that isn't any better than DDR, and through their lecinsing programs, they prevent any 3rd parties from doing so.

        I don't think Intel needs to be "defeated" per-se, but they could sure use some stronger competition, so they can't pull crap like that again, and screw over consumer.
        • If you actually look into that whole Rambus issue you'll see Intel was just as duped as everyone else by Rambus' underhanded submarine patent scheme. Just because they got duped (along with most of the rest of the semiconductor industry) doesn't make them evil.

          Please try again.

      • three things that contribute to some animosity to Intel. For every person that says something against Intel they usually subscribe to one of these reasons.

        1. The pentium floating point bug of yesteryear
        2. The processor ID hoohaa
        3. Competition is fun. No one really pushes MS but AMD pushes Intel and vice versa.

        I'm in the 3 camp these days.
    • It's nice to see AMD, IBM, and Apple working together to defeat Intel.

      Yes, lets all pray that Intel is defeated so that we have a different company that has a monopoly on mainstream microprocessors. Therefore, the existing competition that has driven down the cost of microprocessors, will disappear.

      Rip on them all you want, but overall, Intel has been good for the mainstream computer industry. They generally participate in standards groups and for the most part, have an open architecture. Otherwise
  • You donâ(TM)t have to worry about your current (pun intended) non-fully-depleted silicon oxide chips.

    So you donâ(TM)t need to go shopping for a lead ATX case.

    I think the full depletion increases insulation so the layer can be thinner.

  • Makes hacking tough, don't it? (Score:4, Funny)

    by derinax ( 93566 ) on Thursday June 12, 2003 @01:58PM (#6183780)
    As I read this I'm thinking the whole time of the enormous, nasty globs of dusty, cold solder that make up my 1978 Commodore Pet's motherboard.

    SOI, shmeSOI. I say we get back to centimeter processes-- much easier to hack.

  • All well and good, but... (Score:5, Interesting)

    by Aardpig ( 622459 ) on Thursday June 12, 2003 @01:59PM (#6183788)

    ...nowadays I think that the last component of a PC which needs speeding up is the CPU. Many other components act as a brake on the real-world efficiency of systems; one particularly close to my heart is the cache size. Most computational problems which I come across are too large to fit in less than 2 Mb; therefore, on processors which have a much lower clock speed than x86 offerings, but a much larger cache, I get much better results. The Sparc III series is a good example; the clock speed is around 500Mhz (maybe higher on more recent versions), but the 4 Mb instruction cache & 4 Mb data cache (IIRC) mean that the sort of numerical problems I solve can fly. Of course, it could be argued that this is due to the superiority of the SPARC architecture over x86, but you get my point.

    I'd be interested to try out one of the new Pentium M processors (as found on Centrino platforms); I understand they have 1 Mb caches, and this may give them quite a performance boost for numerically-intenstive stuff.

    • Try to watch one of those HD 1020i feeds on an athlon 1600+ and you will see why we need faster processors.
    • There are Xeon MP procs w/ 2 Meg Caches. I'm not positive but I believe that opteron also has a 1 Meg Cache. If your looking at doing something that high-end there ARE options out there. They just arn't cheap.
    • The problem is that larger caches (which are good for server/high-end computational work) generally mean higher latencies (or ridiculously expensive chips *cough* Sparc *cough*), which is bad for more 'normal' performance measurements (desktop/office/gaming etc).
    • If you want a 2MB cache, you could scout some PIII Xeons, I think they were availble up to 1.2 GHz speeds, I'm not sure.

      I'm pretty happy with my dual 500MHz Xeon. It's a touch slow on video encoding and games but I'm only using 512k cache CPU modules. My machine appears to have a dual memory bus, I think that improves performance too. What is in the system depends on who made it and such. Compaq even made their own chipset for their PIII Xeons and it's the most reliable x86 system I've ever owned.

      Even

    • Re:All well and good, but... (Score:4, Informative)

      by akuma(x86) ( 224898 ) on Thursday June 12, 2003 @07:44PM (#6186874)
      That large cache for your UltraSparc-III is off chip. This is important to note because the bandwidth you get from the off-chip caches is much lower than the bandwidth of an on-chip cache. In fact, if you read the Ultrasparc-III systems specs you'll find that the bandwidth to the cache is comparable to the bandwidth to DRAM on a relatively cheap PC (the new Intel Canterwood/Springdale chipsets have a peak DRAM bandwidth of 6.4GB/sec)

      If you need a 2MB cache you should consider a Xeon-MP which has just that. Couple this with a reasonably fast core and you should see some good performance for your application. Most x86 processors will have at least 1Mb of cache by the end of the year (Hammer, Prescott, Banias).

      As you might imagine, on-chip caches are expensive. As a rule of thumb, the closer the memory is to the processor core, the more expensive it will be.

      Your argument that SPARC is superior to x86 is weak. I've designed both kinds of processors and everything these days is basically RISC-like. The x86 code is translated into micro-ops that look like RISC. SPARC also has some stupid instructions and idioms. For example, register windows may seem like a good idea, but they really grow your register file and limit your frequency. Also, delayed branches are stupid and limit many things you can do. If I had to do another SPARC chip, I'd do some translation of my own into more efficient hardware-friendly micro-ops.

      SPARC systems are nowhere near as competive as x86 systems. Their last niche of superiority with server workloads will disappear with the proliferation of Opteron systems.
  • Last i heard was the the gates pull current during state changes, so faster transistors should result in the chip needing less power.(unless they up the clock, increasing the number of state changes)


    I'm i on the right track here?

  • So how does this translate into GHz? Intel is kicking butt in the marketing arena. AMD needs to ratchet up consumer perceived speed through high GHz to battle Intel. Great technology and metal gates are optional.
    • "AMD needs to ratchet up consumer perceived speed through high GHz to battle Intel."

      You obviously haven't seen their new chip, the Holyshiteron. It features a 12,000 stage pipeline which allows it to run at speeds in excess of 7 PetaHz.

  • Yeah, it sure sounds to me like they're not competing with Intel [slashdot.org] anymore.
  • Nice to see AMD working to be innovative.
    On the other hand, some of the terminology used sounds like they came straight from a bad breast implant procedure.

    "fully-depleted Silicon-on-Insulator"

    Another term for "Your artificial knockers have sprung a leak"

    "Strained-Silicon"

    Another term for "God, those are humongous Jugs!"

  • ...now it'll only take half a day!!!

    Disclaimer: I use Gentoo
  • Please correct me if I'm wrong, but AMD signed an agreement to develop these sorts of things with IBM. If that's the case, I wonder if these techniques will translate over to PowerPC processors ...

    Here's [hardwareanalysis.com] a small article describing their relationship.
  • The press release claims that they're producing circuits that run 30% faster than any other published benchmarks

    But the question is: are these real %% or is it 30+ performance marking?

    Robert
  • A little question about processors etc. We all know that a system is generally no faster than the sum of its components. So if you have a crappy bus then having a 2.6Ghz Athlon isn't going to get you much better than a 1.8Ghz... but how about things like Video?

    I can go out and buy an AGP card, decent on-board RAM, outdated GPU. If current video cards run at, say 500Mhz, and this card is antiquated etc etc...
    Why will a 1.4Ghz /w a slick video card (hardware rendering) machine kick ass over a 2.x Ghz mach
    • shouldn't I be able to go and sink lots of money into a fast-fast- fast processor, and then skimp on the video card and let the CPU take up the slack?

      if the graphics card says it can do it why would the cpu expect that it can't? If you want good snappy video performance, then that's where you should sink your mad-money, what good is it for the cpu to spend 10 cycles waiting on the video card instead of 5 cycles?

  • Intel vs AMD (Score:5, Insightful)

    by nepheles ( 642829 ) on Thursday June 12, 2003 @03:09PM (#6184444) Homepage
    It's about time that AMD got some recognition for their work, and, more specifically, their R&D. 3DNow! was miles ahead of MMX, and the Athlon was vastly superior to the P3. The AthlonXP in turn beats the P4, Mhz for Mhz. The widespread opinion is that AMD processors are the poor-man's Intel. "Good, but not as good". Hopefully the new Opertron (it will be amazing if the Itanium does nearly as well in the 64-bit marked) and announcements like this will help redress the balance. And show that marketing budget isn't a measure of CPU quality.

  • Phaeton Sez (Score:2, Insightful)

    by Anonymous Coward
    Many mention that processor speeds are irrelevant these days, because there are so many other bottlenecks in the system. I will agree that we should leave processors alone for now and work on the other issues to see any real gains.

    Unfortunately, the other industries are market driven, and there are too many people who stroke off to Overclocker Weekly centerfolds of the Latest Greatest Processor(tm).

    What we *really* need, is to completely pitch the entire x86 platform and start over from scratch. You all
    • Right. We're all dupes. The small percentage of the market that would adopt your mythical "well-thought out 64-bit design" would not be dupes. They'd be in a completely different boat than say... those that bought Be Boxes. Sure.

  • 30% faster circuits, "hmmm..." says marketing (Score:3, Insightful)

    by J. Patrick Graves ( 84348 ) on Thursday June 12, 2003 @03:18PM (#6184520)
    "hmmm..., if the Athlon XP 3200+ actually operates at 2.2Ghz, then, assuming the new chips start at 2.2 Ghz, we can market them as 3200 * 130% or 4160. Heck, just round it up to 4200+ "

Slashdot Top Deals

Kleeneness is next to Godelness.

Close