Please create an account to participate in the Slashdot moderation system


Forgot your password?

IBM To Add Silicon-On-Insulator (SOI) To PowerPC 115

hypos writes: "According to this ZDNet article, IBM is going to add an insulating layer of oxide between the transistor and its silicon bed, which IBM claims can increase a processor's performance by 20 to 30%. Best of all, it's supposed to come to new Macs soon. "
This discussion has been archived. No new comments can be posted.

IBM to Add Silicon-on-Insulator(SOI) to PowerPC

Comments Filter:
  • Obviosly flame-bait. However IBM has released an open MB spec for PPC. There are a few box makers trying to build one. The Mac II (1987) was VERY easy to open. But the new G4 is even easier. It is easier to get to the MB in my iMac than in my generic x86 CrapBox. Glad you like the Commodore 64 so much. I hear they are very popular with children.
  • IBM are among the top two or three vendors in every market they compete in -- except for PCs.

    They put a lot of effort/money into claing back some market share (they did after all invent the PC and were rather peeved about the Wintel alliance high jacking the whole thing).

    However recently they seem to have realised two things.

    1. There is not much money in commodity products.

    2. Thier core business is high value, high end servers, is completely at odds with providing cheap commodity hardware.

    They are very happy to leave this to people like AMD (to whom they readily license technoligy) and to the Apple/Motorola alliance.

    When IBM talks about competing with Intel. They are talking about what IBM calls the "mid range server" market and Intel calls the "high end" server market.

    If Intel had actually produced the IA64 on time then they would have made a serious dent in that market. Instead Itel have had to squeeze yet more performance out of the ageing x86 architecture, where (until recently!) there was no viable server Operating System to effectively utilise all that CPU power.

    The intersting competion in this field is from Sun, who for the first time in several years do not have the fastest chip/biggest server to offer thier customers, as both IBM and HP have leapfrogged them.

    It will be very intersting to see if they can regain there lead with thier next generation Sparc - do out sometime soon.

    Isn't competition wonderful!

  • I distinctly remember the iMac/G3 starting at ~300Mhz 2 years ago.

    The first iMac was 233Mhz.


  • If I don't want a Macintosh, it's flamebait?

    I think I miss the logic in that.

    Actually, the C64 was a pretty awesome computer. I'm not the biggest fan of the x86 architecture, but at least you can upgrade it, buy stuff from more than one vendor, and assemble it yourself.

    I don't want to buy an overpriced, pre-built system. I just want a motherboard and CPU.

    Slashdot really sucks sometimes. I don't know why I bother posting. If you say anything controversial, it's obviously flamebait.
  • Give me a break. I'd like to have a PowerPC motherboard and CPU. Do you think I work for Microsoft? What the hell is wrong with you Apple people, anyways?
  • Anyone who refers to Mac as an acronym (MAC) should not be taken seriously. It shows their total lack of insight in the industry. Unless your are refering to an ethernet MAC address....

    Oops, I meant "Mac", not "MAC". Am I insightful again?

  • I looked all over IBM's as/400e site and couldn't find anything that matched your as/400e 20 box...

    the 20 was the problem, model numbers just don't match that exactly, so I couldn't figure out exactly what you have. I'm sure that a linux is available that'll run on it, just not sure what since I can't find out about your box.

  • in fall 1997, Apple released 233 and 266 MHz G3s. Since then, they've gone up by less than 300MHz. I think the first 300MHz G3s were released in March 1998.
  • Wait a minute... doesn't Microsoft own a stake in Apple now, anyways?

    I don't know why I'm still posting about this. I'm obviously dealing with morons.

    Go ahead, mark this as flamebait, too. I don't give a shit.
  • get SUPERIOR font and color handling...stuff Windows/Linux can not touch

    Sarcasm aside, I'd like to know what MAC-only applications you are referring to in this statement.

    On the color handling front, I imagine he's talking about ColorSync []. This sort of color management is the flaw in the "GIMP is just as good as Photoshop" argument you hear around here. It's true -- for what people use the GIMP for: making graphics for on-screen display. Print graphics is a whole other issue and ColorSync provides a tremendous capability that other platforms lack.

    I know nothing about Windows font handling, but if you're comparing MacOS to Linux, it's less an issue of anything on the Mac being great as it is of font handling in X being nightmarish.

    OT: There's certainly a lot of discussion being generated in response to a particularly dimwitted troll. I don't just get those people. I mean, I don't like using a Palm Pilot and would never buy one. So I turn off Palm stories in my prefs! And I certainly don't go wading into every Palm article shouting, "Palms suck! I don't need one so why would anyone ever buy one?"
  • wait lemme guess, you're on a dual-celeron box, right?
    i mean, c'mon x86 is about as outdated as a you can get but that certainly hasn't stopped Intel from taking over the world, has it?
    chip architecture has very very little to do with sales in the real world, except as a tool for marketing people.
  • From the ZDNet article:

    SOI can be used in one of two ways. It can be used to create low-power chips. In this case, by keeping clock speed the same, SOI would reduce the power consumption of a chip by two to three times.

    Apple already claims >5 hour battery life for Powerbooks. With a 2 to 3 times power consumption decrease for the processor, how much more battery life can they get?


  • The article states that IBM believes it is 2 years ahead of its competators. Well, I think they better act on their belief and prove it.

    I can't figure out why a Copper PowerPC has not been released yet. It has always been in the wings for 2 years.

    Also, I remember way back when in the days of the optimism over the open PowerPC Platform, there was some information about a PowerPC 615 processor with built-in x86 emulation. I don't remember if the chip was designed by Motorola or IBM, but that should have been great technology released to the public.

    So, I am skeptical about IBM immediately acting upon their advantages. If they wait 2 years to do it, there is no point.

    It is really a shame that the PowerPC G4 is currently only at 500 MHz. RISC processors should easily have a MHz advantage because they are simpler in design. Man, imagine a 1 GHz PowerPC G4 available today. People would definitely notice.

    So, IBM, prove your new advantage.


    P.S. AMD has access to Motorola's Copper CPU patents. I bet they will use them here soon.

    "...we are moving toward a Web-centric stage and our dear PC will be one of
  • That 1 GHz PPC was essentially an overfunded science project to get a paper into ISSCC and had very little to do with a productizable MPU. It didn't implement the whole instruction set and it had tiny caches. Moreover, it didn't even run at 1 GHz unless you avoided load and store instructions! To get data on and off that chip it had to be run below 900 MHz.

    But I guess it had the intended effect. Two years later and some people think IBM has a 1 GHz chip.
  • I don't mean this to sound like flamebait,

    Well it is.

    but it always seems that the PowerPC is playing second or third fiddle.

    Everyone can't have a monopoly. Unless you run Windows 98, you're second fiddle or less. But you know what? An orchestra composed of only one fiddle sounds pretty dull.

    It gets knocked aside by x86 users on the desktop due to lack of applications.

    Which of your needs is so esoteric that it can't be addressed by the 25,000 available Macintosh applications? Most of the best-selling PC applications have Mac versions, and vice-versa.

    Yeah, sure, it's nifty to do graphics/desktop publishing, but if you're serious about that, you're running an Amiga or BeOS anyway.

    This statement is superficially true, but you're way off in degree. Macs RULE graphics and desktop publishing. Their share of those markets isn't what it once was, but I believe it's still over 65 percent.

    I recall reading once that a Mac makes a nice webserver because it's too dumb to really break into or do any damage.

    You are mistaken. Dumbness has nothing to do with it. Macs make secure web servers because they don't provide unnecessary, insecure services on commonly attacked ports by default like Unix and NT servers do. Macs make nice web servers because you can pack a lot of power into a small, stable, efficient little box.

    So why do people insist on using the Power PC?

    Most of the time, they don't. They insist on using a Macintosh, or an IBM RS/6000, or an AS/400, or something like that. They don't care what the chip is, they like the platform.

    In the cases where people do insist on PowerPC (typically in embedded controllers), it's because it packs a lot of punch in an energy-efficient package, and it's cheap.

    Why do companies like IBM spend development dollars trying to push an outdated chip architecture, when they could be pushing next generation technologies.

    Quite simply, they don't. You must have been asleep at the wheel if you didn't notice that PowerPC is one of the newest architectures on the market. The only outdated architecture people are spending lots of money and effort keeping alive is IA32 (X86), and that's just because it's entrenched.

    Look at how hard Intel has tried over the years to get people off of X86 and onto modern architectures like i860 or IA64. They're victims of their own success, and while it's making them rich, Intel is keenly aware of how IA32 is holding them back technologically.

    The PowerPC will still make a good "beginner's" pc, but I honestly can't imagine anyone who has been computing for more than three years using one.

    Expert answer: Not quite. The Mac is a good beginner's platform, no doubt. The PowerPC itself is an excellent choice for the true hardware snob: It has kick-ass CPI (Clocks Per Instruction), low power consumption, and (in the G4) good support for multiprocessing and an awesome built-in vector floating point unit. An IA32 chip has to have 33% higher MHz to equal a PPC, and dissipate something like 3x to 10x the power.

    Novice answer: Your brain appears to be disengaged. Can you imagine anyone who has used a computer for more than three years switching platforms? They'd have to throw away $hundreds or $thousands in software and start over, unless they started out with a freeware platform like Linux. And we all know once you get hooked on Linux, you never go back. :-)

  • by MrEd ( 60684 )
    What the hell was this moderator doing? The bloody post is three sentences long and doesn't express any opinion besides "Maybe Intel will make better chips". How many posts in how many stories has that little tidbit been said in? Eugh. The post doesn't even have capitals for chrissake! I guess if your post is #3, it gets noticed first.
  • It's all done in software by MacOS, and then done better by the commercial software SpeedDoubler. The current version of MacOS can still run 68K applications via emulation. The PPC architecture itself never had anything to do with it.
  • >is no longer avaiable, sorry) The G4 processor >doesn't need to match the AMD and un-Intel >processors in Mhz. The only numbers you can >really can compare are the MIPS rating on the >chips; which, by the way, is for one reason or >other, hard to get. When things rely on the Actually, comparing PowerPC MIPS to x86 MIPS is absolutely, completely, 1000% meaningless. Comparing MIPS only really means something if you are comparing two chips with the same instruction set, such as an athlon to a pentium3, or as a PPC G3 to a PPC G4. The only real comparison between PPC and x86 is the speed of real-life apps, such as photoshop processing something, lightwave rendering a scene, quake FPS, etc. that measure what the end-user actually sees the computer doing in real-life usage situations.
  • Following on from the previous posts, as to why this has taken so long to develop, there are 2 basic reasons: 1. It is not easy to make SOI wafers! In the SIMOX process, (see previous post), it took years to figure out how to reduce the defect density to acceptable levels. 2. SOI devices are fundamentally different to 'normal' devices in that they may partially or fully deplete the top silicon layer of charge carriers. This does not occur when the silicon device layer is the thickness of an entire wafer! Practically, this means that the device modellers had to go back to square one and produce entirely new models for how the devices really worked, and it took IBM some years to learn how to design transistors on SOI. So it's really quite an achievement, and not just a small incremental change. - r.
  • It's all Bill Walker's fault. Bill Walker (the Motorola GOON who ordered all PPC-based machines tossed out, and replaced with Dell's containing Intel CPUs, running Windows - hows that Love Bug coming along Bill?) has ordered Motorola's engineers to not release 500Mhz or faster chips because that would threaten Intel, and cause his heavily weighted in INTC stock 401k to plummet. Or maybe Andy Grove is giving him blow jobs. Who knows what is going on in Bill Walker's mind. All I know is that he is a servant of evil. I'm not sure how he engineered the IBM split over Altivec, (thus preventing Apple from telling Moto to go jump in a lake, and sourcing solely from IBM), but he has to have his fingers in there somewhere. Evil bastard! Give me my 1GHz AltiVec G4 dammit!~

    I just remembered this old Metallica song. . .
  • another correction:

    Your statement that IBM has PPCs running at faster clock speeds than MOT was a *rumor*. I'm not saying it's true. It's been officially denied, which probably lends it more credibility, but if it's true, the situation is totally wack, and is probably Bill Walker's fault.

    I just remembered this old Metallica song. . .
  • I think this is actually more news than the SOI stuff.

    Anyway, PPC in digital cameras, internet appliances etc etc etc ...

    The finalised design hit the webservers - today so expect a couple of months for the independants to get production running. BTW - anyone can build boards using this design.

    IBM site:

    POP PPC motherboard vendors?:

    Prophecy systems:

    Silicon Fruit:

    POP computers:

  • The info is in another post.

    Looks like there are a few vendors planning POP based motherboards. I'd expect it to take a couple of months to get the boards into production.
  • Now I know it's as popular amoung apple partisans to bash Motorola as it is for Slashdot people in general to bash apple, but don't believe everything you hear. The bit about how G4's would be at 800 Mhz by now except for that Nasty Motorola Keeping IBM From Supplying Them is probably nothing more than a typical Apple-management started rumor, or so I've heard from former Motorola employees with no real love of Motorola. Yah, I know I have no real proof, but noone else here does, and at least I bothered to check.

    I think this is part of the whole overarching problem in that Apple takes a lot of heat for things they don't really do, but they've done enough stupid or venal things that an aura of doubt has attached to them.

  • hey, don't forget that the g4's will also be going multi-processor soon...

    so a 2 Ghz G4... yummy
  • This would prove that IBM is interested in making technology for use in its embedded markets, and is willing to throw Apple and Motorola a bone every once in a while to justify their position in the AIM alliance. Nothing more.
  • by um... Lucas ( 13147 ) on Tuesday May 23, 2000 @06:02AM (#1054030) Journal
    I don't mean this to sound like flamebait, but it always seems that the PowerPC is playing second or third fiddle.

    I think this is flamebait...

    It gets knocked aside by x86 users on the desktop due to lack of applications.

    Same thing always happens to Linux around here, but everyone seems to defend that OS.

    Yeah, sure, it's nifty to do graphics/desktop publishing, but if you're serious about that, you're running an Amiga or BeOS anyway.

    The Mac decimates both the Amiga and BeOS in the graphics and desktop publishing arena's. The Amiga used to be (and still is somewhat) a wonderous machine to work with video with, but it's been floundering the past several years due to not really having an owner that's been willing to pour money into it. The BeOS, yes, has a more elegant architecture, but alas, it lacks color management, postscript font support, and applications from Quark, Adobe, and Macromedia. Until it gets more apps, the BeOS will remain an oddity to everyone except it's core users.

    I recall reading once that a Mac makes a nice webserver because it's too dumb to really break into or do any damage.

    Well, that's true, but if you're buying a mac specifically to serve web pages, in most cases I'd say that you just wasted a pile of money... Or else you're limiting your sites functionality serverly. The mac is missing a lot of support i the server arena. Doubtless, that will change when OS X arrives, but until then.

    So why do people insist on using the Power PC?

    Because 99% of the computers that use PowerPC's run the Mac OS. And some people prefer that OS to those available from Microsoft, IBM, Redhat, Be, or any other. They even like it enough to spend a few more dollars on the hardware i nwhich to run it.

    Why do companies like IBM spend development dollars trying to push an outdated chip architecture, when they could be pushing next generation technologies.

    With Apple shipping nearly a million iMacs and G3/G4's per quarter and with Power PC chips selling for (a complete stab in the dark) $250 a piece, that translates to a BILLION dollars a year of business for IBM and Motorolla. If you ran a company, would you turn down that much money?

    The PowerPC chip isn't geared towards "PC's" as it's name implies... At least in my world, i equate PC with "x86 compatible". People shouldn't be buying Power PC based computers unless (for now) they want to run the Mac OS, or for the small percentage of folks (Linux PPC users) they value to superior hardware designs enough that a few more dollars doesn't hurt. Asd for your 3 year time limit, I don't quite get it. Are you suggesting that Mac users should abandon the platform they chose and switch to Windows or Linux after 3 years for no reason? I've been computing for 15 or so years and the Mac is still my favorite platform for getting work done on and I'm sure plenty of other /. readers will agree..
  • I'd rather use a Commodore 64 than a Macintosh. Do they even let you open the case on those things yet?

    This is a perfect example of what I'm talking about. Apple has excellent cases on the G4's, that open very easily. There's a lot wrong with Apple; there's no need to make up outlandish lies if you want to say something bad about them. But the Apple people do the same about Motorola, so it balances out.

  • IBM has been working on, and has actually built and tested, a 1GHZ PPC chip. I supported the UNIX boxen involved when I worked there.

    I'm not sure why it's not on the market, since the live chip test was well over a year ago. Maybe they're soaking as much cash as they can out of the lower-speed chips.

  • Motorola has demonstrated G4 processors with 7 stage pipelines and integrated 256KB 256bit cache. This will scale much better than the present G4 and it is beleived IBM is focusing its attention on these G4's. Also the G4 is still being made on a .22um process. Moving to .18 or smaller will increase MHz even on the present version of the G4. If you reply and want links I can get some.
  • Remember when Apple stuffed the MacOS licensing genie back into the bottle?

    Motorola had something like $100M worth of "StarMax PowerPC Computers" built and ready to go when Apple did that. I don't think Motorola is too happy about them using PowerPC. I don't think anyone would want to supply anything to Apple after getting backstabbed like that.
    computers://use.urls. People use Networds.

  • I'd rather use a Commodore 64 than a Macintosh. Do they even let you open the case on those things yet?

    No, sir! We're not allowed to open the cases. That would let the radioactive steam out.

    I wish I owned a Wintel PC. They don't even let us have screwdrivers.

    --------------------------------------- ----------------------------

  • by tak amalak ( 55584 ) on Tuesday May 23, 2000 @06:08AM (#1054036)
    Intel has no chips that use copper.


    Motorola/IBM own the copper technology, and Intel is severly lagging in developing it.

    Not really. No one owns copper technology. IBM and Motorola both have patents on manufacturing processors with copper but any company that wants to can develop their own process of making chips with copper interconnects.

    AMD Licenses it from IBM, rather that re-invent the wheel.

    Actually, AMD licenses it from Motorola.


  • T051400000000
  • Why is it that I rise to the bait so easily? why "x86" is yesterday's chip. Is the statement based on instruction set? No? Perhaps you meant archiecture? Hmm.

    Actually, one could argue quite reasonably that the instruction set is outdated (It's CISC, for crying out loud. It's the design for a calculator that they've bolted things onto!) and that only the poorly informed would seperate the instruction set from the architecture, since, in the design field, the two are almost synonymous. The acronym is ISA, if you don't believe me.

    Sure they use the x86 instruction set, but they are deftly ahead of any PowerPC architecture.

    Sure. Links to numbers would be appreciated. Back your statements, rather than spouting opinion, please. N.b. Quake is not a benchmark.

    Sarcasm aside, I'd like to know what MAC-only applications you are referring to in this statement.

    First, if I could reiterate an AC with a good point: either use bold if you want emphasis, or tell me why you're using MAC as an acronym rather than a diminutive of Macintosh?

    Second, the MacOS would be a good example. From font installation and matching, to integrated color picking and handling, developer color matching pachages, and gamma based on print rather than TV, the MacOS is built around publishing and blows Windows right out of the water.

    Ushers will eat latecomers.

  • by TuRRIcaNEd ( 115141 ) on Tuesday May 23, 2000 @07:09AM (#1054039)
    AMD's Athlon and Intel's Wilamette are excellent x86 designs, which are severely cutting-edge. I'm not sure about the Willamette architecture, as I haven't done much reading into it, but pretty much everything from the Pentium (P6 in particular) onwards has strayed from being pure x86 in favour of a heavily pipelined architecture. In the case of the Athlon, effectively the CISC x86 instructions are emulated [] by splitting them down into what AMD terms MacroOps, and letting the RISC core deal with the rest.

    Where the PPC scores here is that it is fundametally RISC by nature, although IIRC the PPC has a basic 68k emulation frontend for legacy purposes. However most modern apps on the Mac are tailored for the PPC, and as such, can use some of the funky RISC features to gain a speed advantage. For example, some Photoshop filters will render quicker on a G4@400MHz than on an equivalent, or higher spec x86 box, simply because the architecture is less cluttered, and the compilers don't have to take the legacy baggage into consideration. The same applies to the Alpha, in that by using MHz as a speed comparison between architectures, you are doing the more modern chips a major disservice, because they don't need to be clocked as high to gain comparable application performance.

    As for the colour and font handling, he's right. The Macintosh's ease of use made it a very strong contender in the DTP arena early on, and as a result, programs like ATM and ColorSync allow a far greater degree of output control than the Windows (or Linux, sorry guys) equivalent. This is just by the nature of having existed on the platform longer, however.

  • I can't figure out why a Copper PowerPC has not been released yet. It has always been in the wings for 2 years.

    The copper PPC has been out for a long time. I believe G3s faster than 400 MHz and all G4s are on copper. Where's the copper X86?

    Also, I remember way back when in the days of the optimism over the open PowerPC Platform, there was some information about a PowerPC 615 processor with built-in x86 emulation. I don't remember if the chip was designed by Motorola or IBM, but that should have been great technology released to the public.

    IBM. Lots of potentially great technology gets scuttled for technical or business reasons. Since IBM never officially admitted the chip existed, we'll never know why it was pulled. Maybe Intel threatened IBM's supply of Pentiums and the emulation wasn't good enough for IBM to tell Intel to take a hike. Or maybe they didn't think it would sell well. Or it didn't work at all. Or maybe it was only a rumor.

    It is really a shame that the PowerPC G4 is currently only at 500 MHz. RISC processors should easily have a MHz advantage because they are simpler in design. Man, imagine a 1 GHz PowerPC G4 available today. People would definitely notice.

    IBM demonstrated a 1.1 GHz PowerPC a year ago at (IIRC) HotChips '99. Current rumor is that IBM is ready, able and willing to ship G4s at 780 MHz but has business reasons for not doing so.

    Don't believe for a second that the super-duper high clock speeds you see in the PC world is more than an Intel/AMD pissing contest. Neither company is capable of shipping large quantities of their highest-rated chips. In the absence of the PIII/Athlon rivalry, PCs today would be topping out at about 700 MHz, and they'd be comparable to today's top PowerPCs.

    So, IBM, prove your new advantage.

    IBM clearly likes to hold its cards close to the vest. They don't hype, they just quietly deliver.

    Does this mean I'm happy with the current state of affairs? Of course not! I want a quad-1GHz-G4 PowerBook NOW, dang it! :-)

  • The original macs were a little difficult to get into without a 'Mac-cracker' tool, since they had a CRT inside. From the II onward, it was usually one screw to get into one. Macs are very easy to get into and very servicer-friendly.

    The Commie-64 is a different story. The voltage regulators used to go on those things at regular intervals, and normally took the fuse in the external 'brick' power supply with it. This could be cut open and the fuse could be replaced, until the second case style came out. After that, they started filling the 'brick' with epoxy resin to prevent any repairs.

    I used to hate Commodore for that reason alone.

  • They are using it in chips for cellular phones. (if I remember correctly) Why we don't see this stuff in computer processors is beyond me but I don't remember them ever mentioning plans to do so. Perhaps it is difficult to work into such a complex design or would require an entire redesign of the chip. That's just a guess though.
  • by madGenius ( 124267 ) <> on Tuesday May 23, 2000 @08:22AM (#1054043) Homepage
    As this is getting into a fest of Mac Bashing I though I'd try and do something on-topic. So here is a little explaination of how IBM's method works (note this is not a new method as the Silicon on Saphire method has been around for years .. however IBM seems to have altered the idea by replacing the expensive saphire with cheap silica)
    A normal NMOSFET transistor would to something like this (side view)

    Drain Source
    (M Metal track) (# Doped silicon)
    (O Silica) (S Intrinsic silicon)

    When the transistor is in operation it has a charged region (depletion zone) around it as such...
    Drain Source

    This charged region (boundary shown with @ signs) acts as capacitor which is bad (slows down the operation of the transistor). What the process does is to place a layer of oxide below the chip as shown

    This stops some of the charged layer from forming as the charge cannot move (easily) in the oxide. Which speeds up the transistor.
    'Tis a nice idea doing it with silica though which should make the chips very affordable (ie. only 2 or 3 extra process stages)
    (btw. this is my on words no trade secrets - not that /. would allow that ;) ) -mG
  • I can't figure out why a Copper PowerPC has not been released yet.
    Hmmm... Maybe you should talk to some Mac users who already have Copper G3s. If you go to XLR8 Your Mac [] you'll note some situtuations where Copper G3s behave differently than their aluminum counterparts.

    I'm not sure if IBM has released any G4s to Apple yet (there was some hassle over getting any IBM G4s, but I think that was worked out a couple of months back, so Copper G4s should be forthcoming soon.

  • Because IBM has been able to do wonders to this chip that Motorola hasn't. Being that there is still oh what? how many months until the end of the year? The G4 in it's current state, *should* be capable of 800Mhz solid, easy, but motorola just has been unable to fab them at this speed.Tan extra tweak of development (such as this) and better fabbing techniques, and business trends are where I got my number.
  • Santa Cruz, CA - A rift in the space-time continuum was created today when overclocker Jamie Aperman ran a 750 MHz Coppermine Pentium III at 1.6 GHz.

    Overclocking has long been blamed for causing global warming, but this is the first occasion that the fabric of space-time has been damaged.

    MIT Professor George Greznowski said, "It appears that the CPU was operating so fast that it began to execute instructions before they arrived. This execution of future instructions created a small tear in the fabric of space-time itself through which part of the motherboard passed into a parallel universe."

    No one was injured in the accident, but a computer motherboard was partially damaged. Mr. Aperman better known as SpeedPhreeek said, "I'm pissed. I lost a brand new Alpha Cooler and Coppermine to a parallel universe. I called my insurance company and they don't cover losses to rifts in the space-time continuum."

    Intel researchers have long warned of such damage to the space-time continuum, and added clock multiplier locks to their CPUs before they were required by Congress. A bill is now in the US Senate which would require a three day waiting period for purchasers of Alpha Cooling Fans and Peltier cooling devices. The bill would also require clock multiplier locks on all new processors.

    Overclocking advocate Horace Spencer said, "This bill before Congress won't prevent overclocking. They'll just create a black market for non-locked processors. Most of the top overclockers already get their goods from Taiwan."

    Article stolen from here [].
  • Actually, SOI is far from new as a concept either. What has happened is IBM has managed to make it mass producable. The big problem with SOI in industry in the past has been that of yield. Getting a consistant thickness in either the buried oxide or the SOI silicon layer is difficult, let alone getting both of them at the same time!

    The wafers I am working with have a nice uniform SiO2 layer, but the SOI layer varies in thickness from 2-3 um across the 5" wafer! That variation would kill any big chip you tried to produce in mass quantities. Since IBM is using SIMOX wafers, this means that they have one heck of a good ion-implanter with amazing straggle control.

    Judging from the cross-section SEM images [] at IBM's website, their SOI layer is of similar thickness, so there is still a bit of silicon between the source/drain regions and the oxide. This distance is what allows them to use the SIMOX wafers, handling the straggle from the ion implantation.
  • I had a Starmax, it was crap. Actually *most* of the clones had watered down bus speeds, and were overall weird - even though Apple "approved" them apple was lax in what they said could happen, but then again Gil was driving the company and he's not exactally the brightest boy in the world.

    On a different note, since the upturn of Apple Motorola has made a ton of money off just the sales of the chips, it would definately be in their best business plan to fab the chip correctly.

  • Thats nothing compared to the Duron layer! IBM can't possibly produce a chip with as much sensitivity as the AMD!
  • Why do companies like IBM spend Development dollars on the PowerPC?

    That's easy. Becuase there is so much more to computing than the desktop. The first implementation of the SOI chips will be in the new 8xx models of AS/400's. RS/6000 soon to follow. The AS/400 announcement includes a kick-ass 24-way processor with memory and storage capacity that surpasses IBM's mainframes.

    All of this on the _Sixth_ generation of RISC based 64 bit PowerPC chip (Remind me, how much longer until Intel releases a 64 bit chip?).

    Outdated? Beginners? Not in the least.

  • the more competition to intel, the better. look what happened with AMD...i know powerpc is not x86, but it does run linux. maybe intel will get some more needed kicks in the pants.
  • IIRC Western Digital drives are technically IBM HDs.
  • I don't believe so, now that VIA own Cyrix I think they do all the FABing themselves.

    My Webcam []
  • Motorola has been trying to get a G4 over 600Mhz to no avail, while IBM has them well beyond that. This move will yet again prove that IBM is more interested in Apple then longtime Motorola. I would imagine that if Apple went solely w/ IBM that we would see a 1 - 1.2Ghz G4 by the end of the year.

  • Didn't people say that about IBM like two years ago?
  • The G3 and G4 are the consumer PowerPC chips. IBM are pushing it - they have the POP motherboards, which are slowly starting to appear now, and with a SOI

    Well, the POP board have just yet been completed, and there will still be time before they go into production. so we will still have to wait some time before you can really buy some not to say at an affordable price (well they cost around $1000 for 1000 pieces, if you decide to make them yourself)

  • This, of course, means that Macs will be coming with some faster chips.

    Why is it that the PowerPC chips have slowed down the rate at which new, faster chips have been released? I distinctly remember the iMac/G3 starting at ~300Mhz 2 years ago. Why is it that they have only climbed 200Mhz or so?

    Perhaps AMD and intel have me brainwashed into thinking that every month new speeds should be coming out..

    Rami James
    Pixel Pusher
  • It's not only processors. IBM is also leading in some other fields, most notably HDD technology.

  • Will IBM license the technology to other companies ?
    AMD used copper before Intel, could they now use SOI before Intel to gain an even bigger advantage ?

    My Webcam []
  • Hrm.. I am typing this on an IBM Copper PowerPC (750L(LoneStar)) atm.

    By the way, Motorola is the cause of IBM not beeing able to release 500MHz+ G4's atm as Motorola cannot do it themselves, they do not let IBM do it either, as Motorola owns the AltiVec rights, they can (and have) done that.

  • if they're inseperable, why is there a K7 and P6? same instruction set, different architecture.

    Properly, that's "same ISA, different design." Picky, I suppose, but the design should be immaterial except for how it performs. There could be little smilie faces in the silicon for all you care. It just wants to be fast.

    There's an AC who's dubious of, and I thought I'd reiterate that and amplify it by saying that the most recent Motorola chip bench is the 604, which is like bringing a PII into the discussion. Hardly fair.

    Ushers will eat latecomers.

  • Yes but wiil this new SOI PPC be as XML compliant as the chip for the AS/400?

    Don't be fooled by non-XML compliant processors!

  • by Ed Avis ( 5917 )
    Do IBM not make x86 chips any more? They used to have their 'Blue Lightning' clock-tripled 486 clone, and later they manufactured Cyrix's stuff. But nowadays, they don't seem to be making 686es. That's a shame, since all the cool stuff they're doing for other architectures (copper, SOI) might benefit the PC world too.
  • I don't think so, it's true IBM often come up with good research and technology, but they always seem to fail to take full advantage of it and someone else does.
    Of course this is probably all part of IBMs plan to make a killing by thinking clever stuff up and then licensing it to anyone who can afford it.

    My Webcam []
  • Is it just me or does the consant flow of patent applications and new processor/storage technologies coming out of IBM remind you of Bell Labs and Xerox PARC in their heyday. It seems like the next generation of high-performance processors will need to include technology licensed from IBM in order to remain competitive. Are we seeing the rebirth of IBM?
  • From the article:

    IBM officials said they believe that its copper interconnect technology, combined with SOI, give it a two-year lead over competitors, such as Intel Corp.

    If they want to compete with Intel, why don't they announce plans to bring this to x86 anytime soon? What they're doing currently is not competing with Intel. They're just supporting Apple in its competition with any servers that run on x86, such as NT, Linux, and OS/2. Oops...scratch the last one :) Does anyone actually see this as competition? Do you think any will have their buying strategy based first on the processor and then decide on an OS after it picks the processor? Heck no.

  • Wait, let me think for a moment...

    500Mhz is the current speed of the G4. Add 20-30% = 600-650Mhz.

    Where the bloody hell do you get 1-1.2GHz?

    Check your math before you post.

    One other thing before I go, if IBM was the only supplier of chips, you would see a HUGE increase in the price of a G4. Competition is a good thing for prices. Don't knock it.

    Rami James
  • Correction:

    The 20-30% improvement is an improvement in performance, not necessarily clock speed as you assume.

  • "The PowerPC will still make a good "beginner's" pc, but I honestly can't imagine anyone who has been computing for more than three years using one."

    I manage an small NT network(dozen workstations & 3 printers) at a medium-sized company.

    At home, I have an iMac and powerbook sharing a cable modem with a cute 5 port hub. I just put LinuxPPC on the iMac. Joy!

    I do so because over the years I have learned to really hate NT and most Microsoft products. And I now resent having to work with it.

    Having all Mac at home reminds me how fun it is to use computers and deters me from ever having to take work home.
  • Many fast processors on the market = incentive for processor designers to push the envelope to be competitive.

    Everyone, not just Mac and PPC users, should be happy at this announcement. Even if it makes Macs so crazy-fast they go back in time, can you imagine how good the eventual competitor from the x86 community is going to be?

    Speed good. Choice good.

  • 3)PPC is a next gen technology. x86 is yesterday's chip.

    Please define "next gen" and why "x86" is yesterday's chip. Is the statement based on instruction set? No? Perhaps you meant archiecture? Hmm. That couldn't be your intent because of both AMD's Athlon and Intel's Wilamette are excellent x86 designs, which are severely cutting-edge. Sure they use the x86 instruction set, but they are deftly ahead of any PowerPC architecture. Of course you could point to the Altivac vector processor as an improvement, but it's not: it's a retrofit -- a coprocessor on die.

    get SUPERIOR font and color handling...stuff Windows/Linux can not touch

    Sarcasm aside, I'd like to know what MAC-only applications you are referring to in this statement.

  • Today, /. announced a new way of improving their memory by adding a layer of caffeine to each synapse in their brains.
    This is supposed to increase memory efficiency by 20 to 30%, making it possible to remember stories that have been posted one page further down.
    This should eliminate having the same story, or nearly the same one, posted twice in the future ;)
  • by JMZorko ( 150414 ) on Tuesday May 23, 2000 @06:26AM (#1054073) Homepage
    ... and to this I submit that the usefullness of a computer is not necessarily attached to the age of it's technology. In other words, people _use_ them, even if they're not the fastest anymore, or the biggest, or even the prettiest. I've got 2 Mac 8500s at home; one running MacOS 8.6 with a G3/366 card for audio stuff (i'm big into ambient music and sound experiments), the other running LinuxPPC on a 200MHz 604e for everything non-audio. Both of them continue to do everything I ask of them; the audio Mac still runs all of the latest Mac-based audio software well enough.

    A lot of us are more interested in _doing things_ with the computer, even if it's 3 years old or more.

  • IBM has big fabs, and I believe they fab for AMD on some occasions. Can't recall if it's for AMD processor, AMD memory, or networking applications, but I'm fairly certain they do contract work for them.
  • I always buy IBM drives for file servers. I have NEVER had one fail on me, and I have had all too many failures from other suppliers. The price difference is not extreme, and the hassle reduction is huge.
  • Motorola had something like $100M worth of "StarMax PowerPC Computers" built and ready to go when Apple did that.

    The funny thing is that all of the StarMaxes around Motorola are leased. Why would any company lease a product that they built themselves. That's like installing a soda machine in your house and thinking "Man, the more soda I drink, the more money I make."

    I don't think Motorola is too happy about them using PowerPC.

    They're not, that's most of the reason they are switching to NT machines and getting rid of all the Macs.

    I don't think anyone would want to supply anything to Apple after getting backstabbed like that.

    Well, just because you don't like what somebody does doesn't mean you stop making a bunch of money off of them.

    MacSlash: News for Mac Geeks []

  • by loki7 ( 11496 ) on Tuesday May 23, 2000 @07:26AM (#1054077) Homepage
    Despite what the poster said, this probably won't be appearing in Macs 'soon'. SOI requires redesigning the entire chip -- it's not just a different manufacturing process. The Register [] (everyone's favourite source of fun rumours!) has a much more techinically detailed article.

    It's a shame that /. relies so heavily on zdnet for stories. They're always technically weak. The Register isn't always accurate, but at least they're not writing for PHBs.


  • What they seem to be implying is that they are increasing the depth of the field oxide, seperating the poly from the substrate. This would hardly take five years to come up with....

    This misses the point. The field oxide only provides lateral isolation between tubs.

    In a typical CMOS process, the devices are constructed in n-type and p-type tubs which are implanted into the substrate. The field oxide helps with the sidewall, but you still have parasitic capacitance and leakage currents at the bottom of the tub.

    Silicon on Insulator tries to reduce the parasitic capacitance and eliminate some nasty problems with parasitic bipolar transistors formed by the implant regions. Originally it was done by growing a thin layer of crystalline (epitaxial) silicon on a sapphire wafer, but this is expensive and has problems caused by mismatch between the lattice spacings and by unequal thermal expansion.

    The problem with an all-silicon SOI process is that the silicon dioxide that is used as an insulator is amorphous, so it is easy to grow polycrystalline silicon (poly) on it, but difficult to grow an epitaxial layer. The approach that IBM is using is called SIMOX, for Separation by IMplantation of OXygen. This works by implanting a high dose of oxygen ions into a silicon wafer and then annealing it, forming a well-defined buried oxide layer beneath the existing single-crystal silicon.

  • SOI is a silicon process that's been around for several years now. I'm pretty sure that Intel neither invented, nor patented, it. My old high-energy physics lab used the SOI process in some of the front-end detector chips for ATLAS at the LHC at CERN.
  • The BeOS, yes, has a more elegant architecture, but alas, it lacks color management, postscript font support...

    Minor quibble: BeOS supports Type 1 fonts.
  • There are two variations of the PowerPC architecture that are being marketed to two distinctly different markets.

    The first variety, the IBM Power4, is for servers. This is the area of that should be of most interest to the Linux market at the moment. With two of these 64 bit processors on one chip and a 500 MHz bus between them (See this ZDnet article [].) they will make killer e-commerce servers that will beat any x86 chip this year.

    The second variety, the G4 from Motorola sports the 128 bit AltiVec vector processor. This caters to the graphics market that is Apple's bread & butter. It wipes out the P3 in applications that are optimized for AltiVec - even at a lowly 500 MHz clockspeed.

    This processor is also of interest to the academic research/defense industry. Look at what makes a Cray, a Cray - you guessed it, vector processing. While I'm not comparing the G4 to a Cray, it's 128 bit vector processor does make a cluster of them a much cheaper alternative for algorithms/code designed for a vector processor.

    The Linux Angle
    I've asked myslef over the past couple months why all the common x86 distros were introducing PPC versions? Surely the Linux on Mac market isn't large enough to make it worth trying to compete with the established players like LinuxPPC and Yellow Dog.

    It is now that I realize what may be their underlying motive. -- I think we're going to see IBM pouring money into efforts to see to it that Linux/Apache/etc are optimized for their processor. Everyone wants to be a contender for potential IBM funding to sharpen Linux and other OSS for their fast hardware that isn't hamstrung by backwards compatability... This can give one of the lesser known distros an opportunity to steal some of RedHat's mindshare. -- Just a thought, but who knows?
  • SiGe is available []. However, leveraging its advantages would indeed require substantial redesign of a processor (using bipolar transistors, no less).

    See my post [] from the other SOI article.

  • Info from IBM (including a white paper) []

    Since this isn't breaking news (other sources had articles about it on Monday), it would be nice if /. invested a bit of time investigating these stories first. Sorry if I sound a bit bitter, but it's been a long time since /. was the first place I saw something really cool. Other sites tend to beat them to the punch, and provide better coverage to boot. That's unfortunate, because I really like the /. format.


  • Calm down, there... I was being facetious.

    The only thing I don't like about my 6100's case is that the Sonnet Crescendo upgrade card works fine, but I got suckered into buying an SCA hard drive, and it's slightly too long for the bay. The sled fits okay, but it won't lock into place, and the front bezel won't fit back on, making for a loose case fit.

    ------------------------------------------------ -------------------

  • I think what this move proves is that IBM is more interested in IBM. If Apple went solely with IBM we'd be using G3's as the liscensing for the G4 and atlivec prevents IBM from selling directly to Apple.
    Maybe we'll see a superfast Powerbook or iMac 6 months to a year from now but don't hold your breath for an SOI G4.
    The whole AIM alliance is just way too crazy.
  • As a Mac user and a person who makes a living off of Photoshop, I'm not so much interested in the Mhz rating as I am in the Velocity Engine, and Motorola's in control of that. I just wish IBM and Motorola would get over their pesky little bickering and just make better chips.
  • Um... IBM has about a 50% marketshare for 2.5" laptop drives... nobody does component miniturzation better than IBM STD.

  • 1) IBM is rolling these new technologies (copper interconnects, SOI, etc.) into their own, proprietary, platforms (RS/6000, AS/400, S/390) first for what should be obvious reasons. They are competing with the Intel server chips by producing non-intel servers. Why just make a profit selling the chip when you can really rake in the dough selling the chip, system, software, and services? You can be sure that it would take a $hitload of Athlons to match the profit produced by a single S/390 or big AS/400.

    2) IBM also directly competes with Intel on several other fronts, namely communications chips. Intel makes more than x86 chips, even though those are Intel's cash cow. IBM is more than willing to license these technologies to Intel's competitors in the semi market on the theory that "the enemy of my enemy is my friend."

  • Um... IBM already has machines with copper interconnects, and now SOI, shipping right now. The chips that IBM is shoving out the door are POWER 4 architecture chips (closely related, but not identical to the PowerPC), (used in the RS/6000 and AS/400 not PowerPC chips. IBM was shipping copper (about a year now) even before anyone else had demo units available. Now the SOI is shipping, and again, no one else is even close. Just because it doesn't exist on PC's doesn't mean it doesn't exist.

  • Not to nitpick, but BeOS does definitely have PostScript font support.

  • IIRC IBM has filed more patents per year than any other company for at least the last decade. Further to that they make at least $1Bn per year from licensing that Intellectual Property. Not a bad earner for a company that does R&D very well.

    Secondly, for the last 8 or so years since Lou Gerstner got in, things seem to have turned around. Look at the success of Lotus Notes since they bought Lotus. Look at Copper Chips, speech reco, RS/6000 posing a real threat to E10k's, S/390's are still used by all the banks, as is CICS. They have been very active in providing Linux solutions, they are onboard Monterey, AIX is still going strong. HDDs are amongst the best out there, they are making innovative PCs (see the NetVista) and very good flat screens. CATIA is widely used, their business PC's are rock solid (but pricey) and Thinkpads are the best laptops I've used.

    Doesn't sound like making a mess of it to me.

  • It's not flamebait to say you don't want a Mac. It is flamebait to denegrate Apple products through feigned (giving you some credit here) ignorance. Unless you have been under a rock you know that Apple has made the easy open case a selling point of their G3/G4 line (the iMac, iBook and PB G3 are also very easy to open). If easy to open were the most important criteria then the Mac leads x86 and even the mighty Commodore by a long way (423% faster in the specINT(opencase) benchmark). You are spouting FUD that would make M$ envious. I don't care if you don't want a Mac, just don't use off base inferences as reasoning. -PSM
  • by ravenwing_np ( 22379 ) on Tuesday May 23, 2000 @05:15AM (#1054094)
    Faster chips are nice, but what I really want is a faster front side bus. I've been working on a long running project that deals with processing lots of large multimedia data. Even though our CPUs have been getting faster, our processing isn't taking less time because we are being IO limited.

    I'd much rather have a 400 Mhz cpu with a 400 Mhz bus then a 1.5 Ghz box sitting over a 133 Mhz bus.

    A starved CPU does no math.
  • Well, IBM reckon that they would have 700MHz G4's next week if Motorola would let them make the damn things, so add 30% onto that and you have 910MHz G4s. Couple that with Motorolas G4+ core, which will be released in around 4 months time at 800MHz to 1GHz and you could easily have 1GHz+ G4s by the end of the year...

    The G3 and G4 are the consumer PowerPC chips. IBM are pushing it - they have the POP motherboards, which are slowly starting to appear now, and with a SOI G4+ processor and AGP4x, and ATA-100 (or Serial ATA) support you would have a really good motherboard that really kicked ass. Shame the motherboards that will appear will be half-functional, but hopefully some server-level motherboards will appear for multi-G4 setups. That would be sweet, much better than a dual PIII Xeon or whatever. A G4 is around as powerful as an equivalent speed Xeon (give or take a little), and they cost a lot less than a Xeon, so people shouldn't give the processor so much stick. A PIII 700MHz Xeon with 1Mb cache can set you back $1000.

    Posted with Mozilla 2000052120. Damned great now, faster than ever, looking good (the buttons are in the correct place!).

  • IBM's PowerPC processors for thier rs6000 and as/400s are COMPLETELY diffrent from the g4 design - they are compatable on a binary level, but then again, so are the 386 and the pentium II... IBM uses very specialized processors in thier high-end systems - three integer pipelines, 2 fpus, 1000+ pins on thier interface (HIGH bandwith bus!!), tons of cache... each integer pipeline is Long, like 12+ stages, allowing for high scalablity in the Mhz of the processor...
    The g4 has too few stages in its pipeline, leading to it being Very hard to scale to high Mhz... copper/soi g4s _might_ get to 600Mhz - but the yeilds would be just as low as the current 500Mhz ones. Remember, the g4 is Tiny it was designed to be used in embeded apps as well as desktops - and thus was designed with as few transistors as possible. AMD and motorola have been working on a 'new' g4 that is basicly a redesign with a few more stages in the pipelines, allowing for much higher speeds.
  • Ooo, we are cranky, aren't we? Three things for you to read, then maybe you can follow your own advice and "think before you post":

    1. IBM already has PPCs running at faster clock speeds than the MOT chips that are being used in Macs. It's a contractual/political problem that keeps these from being used by Apple.

    2. The 20-30% improvement is an improvement in performance, not lock speed as you assume. So the math is irrelevant.

    3. The advantages of "competition" within the AIM alliance are often cancelled out by the alliance itself. In PPC development, no partner wants to release a technology to the others when it could use it for a non-PPC-related benefit to itself. Also, if MOT can keep faster IBM PPCs from being used by Apple, how can competition be functioning correctly?

  • It is true that IBM is a great innovator in HDD tech, but they really don't have much market share. How often do you come across an IBM drive in the field? They seem to price themselves out of the market.

    Their SCSI LVD prives seemed quite competitave last time I bought a drive, it was only $10 more then the lowest price when I bought it (according to pricewatch). I gladly payed the extra $10 because of all the drive failures we have at work, IBMs are the most rare (HP was the most common, untill they left the biz), and they also start giving R/W soft failure errors hours or days before they actually go. I have never seen a hard failure on an IBM SCSI drive without soft failures before it. I want the chance to say "where is my DAT drive?" before the disk goes Tango Uniform.

    Their IDE drives arn't as competitave, that might say more about the quality of their competiters drives then any lack of desire of IBMs to "own" the market. Maybe. It's not like I understand the PC hard disk market.

    As far as CPUs go, I think your thery pays out better. RS/6000s have a mammoth profit margin. Far more then the PowerPC. Then again, maybe the same is true in the SCSI vs. IDE thing too.

  • Everyone dogs MacOS - perhaps it is not the newest OS in the world and unquestionably it has it's roots for being one of the first GUI OSes out there. But let's be honest - the only idea behind MacOS to begin with was to create a perfectly functional operating system. The concept of resource forking was unthought of prior to this, and still is for the most part.

    Honestly we need to give MacOS more credit, and Apple for that matter. How many people have ever been able to make a user friendly unix? or frighteningly enough a cooperative multitasking operating system work so well?

    I am a Linux bigot and as much as the next - but we should give credit where due. Most of us come from the PC background where hardware needs to be kicked to work. From Apples camp the hardware is usually good enough to make do, so this concept is nearly oblivious to us.

  • I don't know about the Starmax, but Power Computing came out with some damn fine boxen in its time (particularly the PowerTower Pro).

    - Jeff A. Campbell
    - VelociNews ( [])
  • Back in the mid nineties when Apple was migrating to PPC and the Taligent venture with IBM was begun (to be orphaned and starved to death), in the days of Pink and Copland, there was talk of IBM buying Apple. IBM had never quite gotten the consumer space and Apple had never quite gotten into the business space; they were considered a beautiful match. The OS's Apple was working on were more geared towards a server and client relationship. With Macs becoming ROMless and Darwin booting accross platforms it would seem no big deal to bring Darwin to IBMs server series (the PPC based ones as well). Though there are some major differences between the G4 Apple uses and the chips IBM, if Darwin can run on x86 I am sure it can run IBM's PPC. Perhaps this (the opening of Darwin) could lead to the begining of the fruits of IBM's relations with Apple from the 90's. Darwin as an alternative *nix-like os for IBM servers? Drive-less iMacs net booting from Big Blue's Big Iron? SOI for everyone?! -PSM
  • It is true that IBM is a great innovator in HDD tech, but they really don't have much market share. How often do you come across an IBM drive in the field? They seem to price themselves out of the market. It wouldn't surprise me if they used similar thinking about their processors. Sure, people are willing to pay big money for the big steel, but are they willing to pay big money for a better G4? To my mind the dual G4 sounds far more promising.

Who goeth a-borrowing goeth a-sorrowing. -- Thomas Tusser