Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!


Forgot your password?

AMD Demos 1Gigahertz cooled K7 152

An anonymous reader wrote in to say that "At the AMD shareholder's meeting today, AMD and KyroTech demonstrated a K7 system running at a cool 1 GHz! " Update: 04/29 10:26 by J : An article at news.com discusses AMD's plans for the chip, including pricing and initial speeds.
This discussion has been archived. No new comments can be posted.

AMD Demos 1Gigahertz cooled K7

Comments Filter:
  • by Anonymous Coward
    It's nice to know that this world is
    populated by a bunch of people willing
    to have such a strong opinion on what
    other people have fed them. I think I
    counted about 20 "engineers" who know
    all the in's and out's of every
    processor and why their's is best.
    (Hey, such and such website says it
    rocks. It must be true.)
  • by Anonymous Coward
    Screw the hype-machine. Let me see some benchmarks. Like a 2.2.5 kernel compile time.
    K7 has gotta by the most over-hyped, under
    specified chip I've seen yet.
    Alpha's are gonna be cheaper than K-63's
    by mid-summer. (250$ range) And quite possibly
    faster than the k7..if it ever get's released.
    Also they will be .18 micron by then as well.
    I'm still smarting from AMD's moronic
    move in which they gave compiler technology
    to *one* vendor..Morons! They coulda helped
    gcc or egcs but nooo... They force all AMD
    owners to buy from metrowerks _AND_ run wintel
    to produce 3dnow optimized code....*spit*.
    Their mystical attitude about the K7 isn't
    giving me much confidence about their ability
    to produce something that's gonna be useful
    to me.

  • by Anonymous Coward
    and resisting the urge to visit their offices at 3:00 AM in black tights and a glass cutter.
  • by Anonymous Coward
    Quake will not run much faster unless you cool your fav 3D card as well or unless you use only software renderer (absent in Quake3Arena).
    Hence, your example of "benchmark" is not relevant.
  • by Anonymous Coward
    The judgement of a cpu's ability to compute is not
    just based on MHz. It is based on many other
    factors including cache and pipeline size. The
    K-7 is an infinitely superior product to the
    celeron. A 500Mhz K-7 will easily outperform
    an overclocked celeron. No comparison.
  • by Anonymous Coward
    The "Janitor" approach isn't nearly as effective nor as fun as the "I've got a gun" approach.
  • by Anonymous Coward
    Well, AMD made the announcement for their Ghz 2000 goal in Maximum PC, and it looks like they're going to deliver!!
  • Intel keeps the market satisfied by charging several grand for processors. AMD's become what it is today for having nice, lower cost CPUs. Naturally, when a cheaper product comes into a market monopolized by an expensive product, they have much more demand. It's nice to see that AMD isn't succombing to this and keeps their new CPUs at $2-300.

    By the way, I caught on AMD's web page that they introduced a 475 Mhz K6-2 (why it wasn't a K6-3, I don't know).
  • > Honestly, I think I'll be happy on the
    > day the Alpha comes out on top of said
    > "chart". Unfortunately it's still priced
    > out of the range of something I would
    > purchase for personal use.

    The Alpha will unfortunately never be at the top of the price/performace list until it becomes a commodity like the x86 processors have become - mostly due to the Wintel duopoly (and some help from IBM in the beginning). Once the masses break away from Wintel, architectures that are technologically superior to the x86 will take over. I can see AMD positioning itself to be making Alpha clones in the event the bottom drops out of the x86 market.
  • Posted by The Masked Miscreant >:):

    Actually, the flash on the MB is more likely to have been produced by AMD. They have a larger market share in flash memory than Intel, due largely to the fact that they sell it for less AND it has a longer life cycle (measured in write/erase cycles).
  • I seriously doubt that your lawnmower gets 100 mpg. Even a walk-behind push mower, without driven wheels doesn't get that much. (Consider how many days you would have to walk to get 100 miles, consider that you use up a typical 1-2 gallon lawmower tank in an afternoon.) It can only get worse when you talk about a self-propelled mower. is your mower electric by any chance?

    Or maybe your 100 mpg meant "meters per gallon"?

  • Intel gave us a demosntration that amounted to a publicity stunt. AMD and Kryotech demoed a real product that will be on the market. I'll bet it could be done with an Intel chip also, but I'll also bet your warranty would be void.

  • ..they've only offset the emmissions, not eliminated them...

    At the car's site they have a suggestion which addresses this concern. To make liquid nitrogen we liquify air. If the liquefaction takes power plant exhaust as its input we can get liquid CO, CO2 etc as a byproduct of making the liquid nitrogen. Then you take these and put them in the empty coal mine shafts. Viola! Its almost totally emmision free. Its a very cool idea. :^)


  • In it's place leave a 486 amd cpu :)

  • 15 hp, actually, and they are aware the engine is inefficient (it was made in the 40's!).
  • I think the more important story here is that AMD
    is much more friendly to overclocking than Intel
    is. As I recall, Intel won't let Kryotech sell
    systems with overclocked Intel chips. It's nice
    to see AMD being more friendly to the overclocking
    community. Now what I'd REALLY like to see is
    something along the lines of "This chip is
    verified to run at speed X MHz at temperature A
    and at speed Y MHz at temperature B...", etc.
    Kevin Doherty
  • Alpha's are gonna be cheaper than K-63's by mid-summer.

    they already are!!

    I just a bought a Samsung 21164A 533MHz for £149 from a Samsung dealer. The cheapest i can get a K6-3 for is about £250!!.. 600MHz Alpha's are about £480, comparable to a P11 500 or a P111.

    The 533MHz way outperforms the K6, and a relatively recent 533MHz will clock to 600 no problem.

    Alpha - you'd be nuts to get anything else!

    (PS: $/£ = ~1.5/1)
  • but there's not much software being written for them.. when ms ports to these chips a huge userbase will follow. until then the alternatives u suggest wont be adopted regardless (sadly) of their technical superiority.

  • If they are still using a MOS transistor design, the current gain decreases with increasing temperature. While this temperature/voltage gain ratio is good for preventing thermal runnaway (that bipolar junction transistors can exhibit,) this may pose a limit on high frequency operation at high temperatures.

    When frequencies increase, the higher gate switching speed generates tremendous heat.

    I would imagine that cooling it down may allow for useful operation of the transistor gates as they still have sufficient gain to operate usefully. I suspect that my celeron 300a did not like 504MHz longer than 10 minutes due to the gate temperatures decreasing the gain required to switch logic reliably.

    That's my guess.
  • I have been reading recently that the latest 3d cards are so fast that the pentium II cpus can't feed 'em data (triangles?) fast enough.

    Since 1GHz is in the microwave domain, I can imagine microstrips and waveguides feeding the video card. Imagine the framerate of Quake several orders of magnitude higher. Can your brain keep up?

    Seems like 1GHz leaves many more opportunities for the way electrical propogate. Rather than using a wired bus, it could be possible to beam out the processed information directly from the silicon into our world. My computer makes FM radio and TV useless, why not just take advantage of the emissions and make them useful? Telepathy, anyone?
  • yeah I've got one on now :) who the fuck cares. I'm probobly running faster than you and I can shut it off re-nob!
  • ARGGGG you fucking AMD guys don't get the point do you? the K-7 isn't competiting with the celeron; but instead with the PIII and other high end intel chips. The point he was making about a High speed celeron was that any moron can overclock it's the processor that makes the difference and that Intel makes better CPU's no questions about it. comparing a K-7 to a celeron is like comparing a BMW to a GEO you just don't do it.
  • Hmm, I think at around a gig, I may just need
    to stick a slightly larger heatsink, or just go
    way out and put a fan on a PowerPC CPU..
  • Does anyone know when the K7 will hit the market? Tom's is not more specific than 1H99, and that is in an article from October. Ars Technica (as far as I can tell) has no articles on the K7 at all. So- does anyone know anything about a release date? Speculation and wild rumours welcome...
  • Kryotech sells refigeration units for CPUs. Basically, it's a little refigeration motor about the size of a shoebox that sits under your PC, connected to an appropriately sized little cover for the CPU.

    It's news just because it's cool to have a chip running at 1G in a commercially viable environment (kryotech already sells systems with K6's and alphas).

  • Shurely a warm 1Ghz ;)
  • hert is not the singular form of hertz.
  • My UPS weighs 120lbs.. hehe :)
  • I can't really tell what they are selling here. Have they integrated cooling onto the CPU, or are they selling some kind of box with active cooling built in?

    If this is just Kryotech getting a stamp of approval from AMD, I can't really see it as news.

    But if they start selling chilled-down K7 cpus on the Chip Merchant, that would be pretty spiffy.
  • Flame me if i'm wrong, but KryoTech's assertion that CMOS chips naturally run faster when they are cooled sounds a little fishy. The clock oscillator that provides timing to the chip is on the motherboard, not in the CPU core, where it would be cooled. The clock chip runs at room temperature, and runs at its normal speed, therefore the clock pulses go to the CPU at the same rate. Unless the K7 adjust its multiplier depending on temperature, all cooling the chip does is *allow* it to run at high speeds (on a fast MoBo, with all the clock jumpers maxed out). This is a little different from my definition of "natually faster" I may be splitting hairs here, but they ought to let people know you can't just slap an active cooler on a pentium in a tx motherboard and expect it to run at 1.0 GHz...
  • I read an interview with an AMD employee who was talking about the SMP capabilities that the K7 would have, they said that since it is running on the ev6 bus, it'll be able to support 16(I think that was how many it was) CPUs.
    Sorry, I don't remember where it was at. I'll look at a couple hardware sites I frequent and see if I can stumble onto that... if I do, I'll post a reply with the URL.
  • OK. Cool and I want one.

    How would you get Linux on it though? It's got
    an Intel compatable chip and an Alpha motherboard.

    Anyone know? Is there a project working on this?
    Am I just waaay out in left field?

    Superstition is a word the ignorant use to describe their ignorance. -Sifu
  • ...assuming that your audience had at least the IQ of a pencil.

    ``... 1 GHz (1,000 MHz or one billion cycles per second) ...''

    I've always wondered about that. Thanks for clearing that up.

    ``Al Quick, Chairman and CEO of KryoTech. "Working together, we have produced the Super-G, a true next-generation computer system. With performance measured in gigahertz instead of megahertz, ..."''

    Yikes! Where do I begin?! ``Performance measured in gigahertz instead of megahertz,'' OK. I ``measured'' the performance of my old Columbia Data Product XT-clone in megahertz. Guess I can drag it un from the basement and re-``measure'' the performance by saying the CPU clock is running at 0.00477 Gigahertz (BTW, it's not ``gigahertz''). When was the last time someone actually spouted crap about performance based on clock speeds? About 1985? Thought we knew better than that by now. Of course, the CEO was talking at a shareholders meeting so I guess he had to dumb it down just a tad, eh?

    All the technical usage errors aside, I want one of these! Of course, that'll be after I install a screen room at home to run this computer in so I can keep the FCC at bay.

  • That's *Alpha*, not "altha"...

  • One coal fired electrical plant causes a lot less pollution than the thousands of internal combustion engines it could replace. The coal plant operates much more efficiently -- it's always running at a constant rate picked for maximum efficiency, rather than accelerating and decelerating constantly.

  • by DP ( 11614 )
    The cool thing is that this is a .25 micron chip oc'd to 1GHz. This pretty much makes it clear that when AMD switches to .18 + copper they'll clear 1GHz easily. Quad .18micron 1.5GHz K7 w/8Mb full speed L2....mmmm...drool.
  • Since 1 GHz is in the microwave domain, I can imagine it being mind blowingly incredibly expensive. And I don't want my computer to sterlize me either. (makes me nervous to think of holding a laptop radiating those frequencies)

    There's nothing magical about operating in the microwave frequency range. In fact, your cell phone already does. 0.18 micron chips will be able to do it fairly easily, and will cost no more than any other chips made with the same process technologies.

    Re. microwave emission, I wouldn't worry. Designers go to great lengths to minimize emissions (microwave or radio) from chips and bus traces because that causes cross-talk between lines. Modern motherboards put a lot of ground wires and plates around active wiring to shield them and reduce capacitive coupling between signal wires (at the expense of capacitive coupling to ground, but that's a tolerable tradeoff). If communication was done using microwave waveguides instead of wires, then there would still be little or no leakage - because leakage would again mean cross-talk, and waveguides are by nature very well shielded.

    Even if an incredibly poorly designed motherboard did manage to radiate microwaves in quantity, you'd end up with an output power no greater than your input power. A few tens of watts (the amount of power that the eletronics consumes, as opposed to the drive motors and monitor (if you plug that into your power supply). It takes a couple of minutes at over a _kilowatt_, confined, to nuke a hot dog. I wouldn't worry about being near an unconfined (radiating in all directions) source in the range of a few tens of watts.

    Re. sterilization, I've been hearing mixed information about whether or not that actually happens and under what conditions. It's generally associated with people stepping into microwave relay beams, which are *far* more powerful than anything your motherboard would produce. More information on this, with hard references, would be appreciated, though.

  • So- does anyone know anything about a release date? Speculation and wild rumours welcome...

    I remember hearing early June at one point. AMD's page might have a better answer, if they've kept it up to date.

  • If the liquefaction takes power plant exhaust as its input we can get liquid CO, CO2 etc as a byproduct of making the liquid nitrogen. Then you take these and put them in the empty coal mine shafts. Viola! Its almost totally emmision free.

    Firstly, CO2 doesn't have a liquid phase at atmospheric pressure - you'd get dry ice "snow" instead of a liquid. I don't know about the CO.

    However, more importantly, you would get gaseous CO2 and _CO_ boiling out of the mine shaft as soon as it warmed up a bit. This is not a practical disposal method.

    I suppose that you could get rid of CO2 by binding it chemically to form carbonate rocks, but that's probably more effort than it's worth. Plant more trees instead :).

    CO you get rid of by building more efficient furnaces. It's the product of incomplete combustion, and so will form CO2 if fully oxidized. SO2 is more of a problem. That is usually extracted by chemical reaction to form sulphate minerals (this is what "scrubbers" do).

    IMO, the most convenient solution to the fluel problem is to run vehicles on methanol. You don't have to worry about CO2 buildup, because the plants you grow to produce the methanol take in as much CO2 as you get out by bruning the methanol and/or the plants themselves. The resource will last as long as the sun will, and methanol has a very high energy density (the energy density of liquid nitrogen is miserable IIRC).

  • not that im an Intel lover or anything like that, but I wish I a dollor for everytime I thought Intel should be worried...
  • The K7 bus system is not just the same one as the newest alpha uses, but it's been designed by the same person who was involve in the alpha design (whatever it's called - I keep forgetting).

    If AMD would make their chips pin compatible with the alphas, there would be a very strong case for buying K7/alpha based system (or would there?). In fact it could easily be AMD's chance to steel one up on Intel. Especially given the K7 is aimed at high end of the industry (who might use alphas anyway).

    And personally I'm putting off my next computer upgrade until the K7 comes out. The P-120 is fine for everything except mp3 encoding.
  • AMD has make specs form 3DNow available--nothings stopping you from writing support for it into egcs yourself.

    Yes, it would have been nice if they'd paid someone to work on this (not that SIMD optimization is exactly a solved problem :) but I sometimes think we need reminding that the linux kernel doesn't exist because of corporate support. Same goes for the GNU utilities, the Gimp, and the a lot of the rest of free software.

    Don't complain that AMD didn't give us a compiler, complain that there is no compiler.
  • take a look at the picture of the machine on www.kryotech.com, its a monster of a box.
  • "enchanced p2 with mmx". The PII had mmx. The PIII had KNI and mmx.
  • >K7 has gotta by the most over-hyped, under
    >specified chip I've seen yet.

    No.... that would be the Transmeta whatever-the-hell-it-is/will be.

    >Alpha's are gonna be cheaper than K-63's
    >by mid-summer. (250$ range)

    Don't get me wrong, I am impressed by the Alphas too... but there's also the cost of a mainboard-- a K6-III will run in some older boards; and Socket 7 is easier to find and probably cheaper than Alpha socket [does it have a name?] I wonder which is cheaper with both factors figured. Not to mention the price of an Alpha Linux CD to replace the x86 one I have already...
  • At lower temperatures electrons flow alot easier. So when you cool a chip it can run faster with less resistance. And when you run a chip fast it heats up. Anyone who's overclocked knows this...but if you cool the chip ALOT you can overclock it alot. Along with a smaller transistor size means you have a much faster chip.
  • They are sending us subconscious messages.
  • No, a cool 1GHZ.

    And don't call me shirley.

  • ...that we'll all be driving mailtrucks in the future?

    BTW, low emmissions don't mean bullpucky if the thing making your energy to make your fuel is a coal-fired electric plant...

    "i feel like a quote out of context."
  • True. But my point was that in creating this propulsion system, they've only offset the emmissions, not eliminated them. Centralizing power production usually makes more sense... except for nuclear!

    "i feel like a quote out of context."
  • How would you get Linux on it though? It's got an Intel compatable chip and an Alpha motherboard.

    I'm running a K6-2 right now with Linux, and have been for about a year now. I don't have a single Intel chip in the whole machine. Runs great.

    The K7 won't be much different. You'll probably see a mainboard from FIC when the K7 comes out. FIC typically has mobos released right about when AMD has a new chip. Now if they could only fix their AGP implementation...


  • But I still want to see the SPECfp numbers. (not bloody likely since you need an expensive license)

    I have a feeling that the Alpha 21264 will _still_ toast the K7 on SPECfp at half the MHz. Unfortunately, the 21264 is sort of out of my price range -- unless someone sells it for a lot less than Compaq does at around $7000 for NT and $10000 for Digital^H^H^H^H^H^H^HTru64 Unix.

    Of course, on the 21264, just about all my Windows 95 games are useless. Looks like the K7 wins...

  • Well i'd hope the alphas would wipe the table with just about any other processor around, I am biased for them and all that...and i do love them to death. The best thing coming out of the K7, besides some nice competition for microsoft is hopefully cheaper and more availible boards for the alpha processors, and hopefully cheaper alphas sometime in the near future.
  • I hope this is the start of K7 news from AMD, nearing release :)
  • Duh, I know that... I meant is Kryotec going to build SMP systems that are OCed to 1GHz? Probably not, but I'll give them the benefit of the doubt.

    All I said was that I wouldn't want one of these supercooled systems unless it was an SMP rig.

    ya jerk!

    - - - - - - - - - - - - - - - - -
    I run BeOS. The rules don't apply.
  • The spec for the K7 is awesome (200Mhz bus, with rambus ram, altha techonlogy bus, designed by the very person you designed the altha chips, multiple cpu support, ohh and almost forgot COPPER instead of aluminium ). All that sounds very tasty and i will get it for sure. I think AMD are slightly over confident of their chips but then again if you know anything about marketing then you would know that calling your chip slower then your competitiors is not gonna help sales. Intel have too much of my money already why not give some to AMD.
  • I don't foresee any difficulty installing linux on one of these babies. K7 will either work as a x86 or altha chip. I even heard somewhere that dual motherboards will be able to take both the K7 and Altha chip.

    probably not but a nice thought If there will be a difficulty it will probably be sorted out in a matter of hours by distributions and everybody will be reporting there compile time to the newsgroups.

    I'll be getting a dual or quad board for sure but i will wait for the next gen of K7 ( the first ones won't be in copper) the second bunch will be 0.18 and in copper, RAAAA!!! Imagine HALF-LIFE run on that with a TNT2 or even better a POWERVR SG - dream machine.

    - debian rules - :> (couldn't help myself)

  • really?
    if I say Alpha do you know what I mean?
  • Are you sure IBM demoed a 64bit chip at 1GHz? I could have swore it was a modified 604, being as how IBM still doesn't have a single chip 64bit implementation other than the 620...
  • x86 will never be the "high performance" platform. AMD and Intel are going to be making some very fast chips but it will still be easier to make an Alpha or a PowerPC that runs quicker. x86's thing is price/performance and legacy software. If you want pure performance there are much better alternatives. (just about everything is better...)
  • I'll disclaim that I'm not an ECE but the way I read their statement is that the actual transistors on the chip operate in a more electrically efficent manner at lower temps. As a result, when the higher clock speed sends more pulses into the chip, the more efficient transistors are able to cope where at a higher, less efficient temperature they would malfunction. In this way the chip core is "naturally faster" at lower speeds, since it can raise its theoretical top speed.
  • First of all, performance won't be terribly stellar for applications that thrash the K7's cache. Main memory isn't cooled, and still has a _latency_ in the 6-10ns range (bus speed notwithstanding).

    OTOH, things like Quake that fit within the cache will run more quickly.

    Just remember that the K7 has the ability (with the extra [tag?]-RAM added) to use up to 8 MB of cache. Now, granted, things can still thrash 8 MB, but it's a lot less likely than with Xeon's 2 MB maximum.

    my 0.4c (hey, Rands don't go as far these days :) )

  • You obviously don't own a K6, nor bothered to look at any data aside from MHz.

    I've owned three. And received excellent performance from all of them. Integer performance has always been a strong point of the K6, and that held true. Certainly I am able to perceive a difference in performance between the K6/200 and the MMX/200, in the AMD's favor.

    My friend bought one for that reason. It couldn't compile the kernel, thanks to various errors AMD eventually got rid of.

    The Signal 11 / Segfault problem was resolved in steppings prior to the debut of the 233. I believe it existed only in steppings prior to revision C, all of which were 166 and 200 MHz parts. My early K6/200, at 225 on a 75MHz bus, is quite good at compiling kernels, or X, or what have you.

    if you try something FPU intensive like playing mp3s or quake

    Need I even bother trotting out the old "AMD FPU is faster, it's just not pipelined enough" argument? I get the feeling this is going to be one of those perennial Hatfield & McCoy blood feuds.

    Be careful buying AMD. They make good chips, but they've never released a chip which could compete with its intel equivilant in all categories.

    I would say the same things about Intel chips. The C300A cranked up to 464MHz, roughly the equivalent of the K6/3-450, lacks 3DNow and the AMD part's higher-performance cache, for example.

    I'm a firm believer in price/performance. That's why I sprung for the C300A with my most recent CPU. $60 for a 464MHz part that outstrips a P2 is, hands down, the bargain of the year. Until the stock of 300 MHz Celerons dries up completely, Intel is at the top of my chart.

    Honestly, I think I'll be happy on the day the Alpha comes out on top of said "chart". Unfortunately it's still priced out of the range of something I would purchase for personal use.
  • First AMD said about the same thing about the K6 among other chips...

    And shortly after ramping up production, the K6 was the fastest x86 chip in production inclusive of Intel's offerings for several months. Try not to omit that little tidbit.

  • Check out www.pricewatch.com [pricewatch.com], It's not just one retailer, it's everyone.

    AMD's yield's are just too low, and they have to mark chips they do produce with almost no safety margin on the MHz rating. (unlike Intel's chips, the majority of which you can overclock to 1.5x to 2x their rating.)
  • Then I'll get a K6 III and start
    watching the price of K7's come down.

    I like to stay a gen or 2 behind.
  • all right, we're way off topic here, but you don't have an email address posted, so I can't take it off line.

    You collect the dry ice, and put it (as gaseous CO2) into depleted oil wells. Not a mine shaft. Hopefully, it stays down there.

    Yes, LN2 has crappy energy density. The negative emissions, though, is a bonus that no other fuel technology has. Non-flammability is another bonus. And (like methanol or H2 fuel cells) it beats the hell out of the currently hyped battery powered vehicles which take forever to refuel and need replacement of pollution-causing batteries every few years.

    Anyway, I just like the fantasy of LN2 (or actually, liquified air - it's cheaper if you don't separate the oxygen) cars. Imagine LA's freeways naturally air-conditioned with headily fresh-smelling CO2-depleted air... and, as I said, as a side benefit the computer in your car could pack in the gigahertz.
  • IBM joins 1,000 MHZ Club [news.com]
    Here's to:
    1GHZ processors
    1GB plus RAM configurations (QUIMMS?)
    TByte storage
    cheap big HD displays
    ... that I can afford
  • it's more or less a x86 processor like any other so it should work no problem, but I don't think anyone will be able to test that for awhile still...
  • haven't seen how people banter about how AMD isn't as fast as Intel in processors just because of a 50 Mhz clock speed difference have you?

    People think Mhz mean something unless they actually know what they are doing (which can be rare very rare).
  • From what I've heard K7 SMP setups will probably have to wait on none-AMD designed chipsets. They are quite possible, but I don't think AMD plans on having their reference chipset design do SMP or we would have seen an extra processor slot on the pictures of the motherboard they've been using (the nice looking red one). Hopefully Ali & Via make chipsets that can do SMP for the K7. They could have for socket 7, but didn't so I guess we will just have to hope they think SMP is important this time.
  • The design rules change as the frequency increases. Time is measured in millimeters of copper trace on the motherboard, so just because the silicon won't vaporize at a higher frequency does not mean the circuitry will work. Parasitic factors like capacitance and inductance also rear their ugly head as frequency goes up. So much changes that it's just not realistic to expect any of this to trickle down to a consumer market. At least not anytime soon.
  • You produce one vendor that is selling the K3 at way too high a price, call your local wholesalers, you can can it 100-200 cheaper the a PIII.
    ________________________________________________ ________
    Can We trust the future - Flesh99
  • AKAIK Intel voids the warranty on any chip that
    has been overclocked. This looks like AMD/Kryotech are demoing a product that they will be selling and providing support on...could be wrong. Of course, they coudl also be showing that AMD is catching up with Intel. 2 months is a HELL of a lot closer than the 6 mo behind they've been historically.
  • Um two months ago Intel showed of a Gigahertz chip. Not to mention that they are actually SHIPPING the chip they overclocked to 1000Mhz.
    And the fact that Intel can PRODUCE enough chips to keep the market satisfied unlike some other chip manufacturer.
  • Just the CASE and fridge unit weigh 77pounds!

  • Nope, they are going to ship a .18um version. Intel CAN do it with TODAY'S technology if they are using Kryotech just like AMD is planning on doing.

    Hell AMD can't even make enough K6-3's so they are still pumping out K6-2's. Hmmmm... wonder why AMD's K6-3's are MORE expensive than even the Pentium III's at the same clock speeds? Gee what am I going to do, get a whole new motherboard, and CPU or pop out my existing CPU and put a new one in? Which is going to be better for the consumer?

  • And my Sony 24" monitor weighs in at 88 lbs. What is your point?
  • Cheaper? Explain to me why the K6-3's are currently more expensive than the similarly clocked Pentium III's? Don't believe it? Go check out CPU Prices at http://www.sharkyextreme.com

    Why? Because they cannot even make enough of the K6-3 450's to go around. They are almost IMPOSSIBLE to get a hold of because AMD cannot produce em.
  • I read something to that effect here [hardocp.com]. They're using TEC elements and anti-freeze ro run it @-57C or so.
  • While the K7 has multiple processor support, Kryotech's releases indicate that they are making cooling devices for uniprocessor systems only.

    Kryotech clearly has the ability to cool a multi-processor system, in an older document of theirs they claimed to be able to cool processors up to 1KW. This looks like a newer cooling system - one which might not have quite as much capacity, but I would certainly expect them to be able to do 4 processors without much difficulty.

    Only time will tell. I certainly hope they do come out with a multi-processor solution, even if I can't justify spending that much money on a computer for myself.
  • The Celeron is NOT to underestimated. I've got a PPGA 300A running 504mhz no sweat with only a TEC and decent heatsink to cool it. Another system runs SMP Celerons@464mhz each. I measure performance by the bang for the buck ruler - Celerons clearly win that battle right now! None of my CPUs cost me much more than $60 plus heatsink and MSI adapter. That's dirt cheap computing and I'm going to be very interested to see how my NT server responds to an overclocked PPGA 366. Don't worry - the SMP system is RH5.2 right now...

    As for the K7 - I'm waiting for it to arrive with bated breath. I've got contacts in the hardware world telling me that the performance numbers quoted by so many hardware sites are crap - the K7 configs seen by those sites were far from shipping systems. Whether or not this is true I don't know but I REALLY want the K7 to succeeed. My biggest concern is cost - will this puppy be so expensive that no one can afford it? I'm hearing that the new PIIIs are overheating so maybe the competition is faltering and AMD will have a good chance - we'll see. AMD is supposed to release this chip in July at something like 550mhz, meanwhile Intel readies it's 600mhz PIII to counter the faster AMD. If both chips benchmark closely (FPU on the K7 promises to be stellar tho') then it'll be interesting to see who decides to switch. Overall cost of switching (mb etc.) may be what stops many from doing it.

    Next couple of months are going to be VERY interesting. I'm rooting for AMD myself and have bought some stock as well. If nothing they'll give Intel a pretty good run for their money. Don't use their previous CPUs as a benchmark for the K7 - I don't believe those had the benefits of Alpha technology. The yield they can produce from such a big die will make a difference, K6-3 are supposed to be hard to get now (sigh). When they goto .18 it's supposed to halve the wafer real-estate needed and increase the number of chips produced by better than half....

    P.S. Since AMD has said they won't be locking these chips it's overclocking ability will be fun to explore - someone needs to mass produce a cooling system like HOCP used, I'd buy one!
  • IBM showed off a 64 bit PowerPC 1 Ghz chip back in feburary of 1998. I think that IBM and their server/mac chips are way ahead of Intels' chip designers.

    IBM has created Sillicon on Insulator(+33% performance), changed the wiring to copper(+33% performance), and have been keeping the chips small, less power comsuming than previous models, cooler, and still standard with the ZIF socket.

    AMD chips run cooler, and sometimes depending on configs faster than Intels. AMD's K7 specs look a hell of a lot better in technology than anything Intel claims to be making in the next few years.

    All Intel has done is extend their old chip set in a poor manner, their chips are proietary with the slot 1, excessively large, way to hot in laptops even with cooling features.

    I think that IBM and AMD look like the leader in chip technology for the presonal computer. Intel is no longer doing anything worthwhile.
  • Intel beat em to it- they demonstrated a 1Ghz processor a month ago.
  • I have been reading recently that the latest 3d cards are so fast that the pentium II cpus can't feed 'em data (triangles?) fast enough. So a processor 2 or 3 times as fast would undoubtedly increase the speed of 3d accelleration.
  • And because each cpu has it's own bus instead of sharing one like Intel's P-II's and P-III's... the K7 will actually scale above 4 cpu's without bus contention.

    Intel ought to be very worried

  • Go to Tom's Hardware [tomshardware.com] read. -Be enlightened.

    Its a clamshell refrigerator with a compressor which encloses the CPU... -And yes you might say that it is overclocked. But then again, if AMD doesn't "call" it overclocking... then it ain't.

  • Sort of that, but the main thing was not that they couldn't put out the chips fast enough. They made money on the processors anyway, though they could have made more if they had better production. It was the fallout in the prices of RAM (AMD makes RAM too) and the Asian crisis (they like AMD stuff, aparently) which made them actually lose money.
  • Well, x86 performance has always been measured in Mhz (of course, everything has been measured like that; it's just with x86 that's about all people care about). I remember seeing on a gaming page one of the maintainers saying something to the effect of, "SGI came out with a new processor today. But it's only 300mhz! Hah! I'll keep my Celeron!" Gamers and business people don't understand the more arcane measures of performance. At best they know that a K6 has worse FP performance than a comparable Intel. But I doubt they could quantify that. So, don't blame just AMD. It's the whole x86 market. Why else would Cyrix and IDT (I think the two of them) be puting out PR-200mhz for something that actually runs at 180mhz (just an example; I don't remember the numbers) but performs like 200mhz (but not really, after all)? Because mhz are all people undrestand. I still cringe at hearing a salesman try to sell my friend a computer and after he said mhz (which he said a whole lot) he would say "millions of cycles per second" really fast, to make him look extra smart.
  • From Kryotech's Super-G FAQ [kryotech.com]:
    Q13: Will the Super-G be used as a workstation, or as a small server, or as a very high performance personal computer?

    A13: Yes(!) The Super-G is a very high-performance uniprocessor computer system. KryoTech expects it will be used in many computing environments where a very fast uniprocessor system is needed.

    Looks like not even Kryotech can get a hold of a dual K7 (and you know that if they could make a dual GHz machine, they would -- wouldn't you?). At least if they had a demo of a cooled SMP K7, we'd know if such a thing even exists. I personally can't wait to get a K7, and I'll probably get one the day they come out. But if AMD will have dual CPU setups coming out later, I'll have to wait. Does anyone have any info about a the dual K7 rumors? I haven't been able to find out anything.

    Also, you might want to get Kryotech into the next century and tell them to ship with Linux pre-installed. I already mailed them about it and you can to. Kathy Hemby [mailto] is the person you want to talk to. Maybe if we /. them, they'll wake up.


  • If it started out as a 500MHz chip, and they O/C it to one GHz, then it's no different than what Intel did, except it will be for sale. There have also been reports on Ars Technica that someone has gotten a Celeron 300a to 633 MHz.

    Personally, I would only buy one of these if they had SMP machines like this. mmmm 2GHz. drooooool

    - - - - - - - - - - - - - - - - -
    I run BeOS. The rules don't apply.
  • Me too, I wouldn't underestimate Intel anytime soon. Nobody in the chip business has their resources and they could probably deliver a string of flops before they got in to real trouble (arguably the x86 line has been a string of flops... but I meant in terms of marketplace performance)

    They also have their hands in many other aspects of the PC business. The main reason PCs have the price performance edge they do is because Intel controls the chipsets, a lot of the motherboard market, controllers, etc.. We have yet to see K7 prices, K7 systems could end up costing more like alphas because they are using a lot of alpha parts. Not to build them up but Intel is a very formidable foe and AMD hasn't been cutting profits.

  • Okay I hear a lot of people really excited here about the K7. Put a few things into perspective. First AMD said about the same thing about the K6 among other chips... We all know they didnt come through as strong as they wanted or they would not have reported such huge losses. Now I hear all this talk about the K7 and SMP and 2,000 MHZ processors. Im more weary than excited about the K7. And possibly a little fearful that the only real competition INTEL has seen on the PC market is going to die out. And what about the SMP all anyone talks about or ever says to the press are these obfuscated riddles. Nothing concrete has yet to come from AMD. Yet people want the chips even if they are not as great as said? AMD has showed stockholders a overclocked chip thats all. Ugh I can see this being just like the K6-2

    Jeremy Allen
  • the K7 because of the bus system used (the same one in fact as the newest alphas) can support up to 16 processors on one board if someone felt like making a motherboard like that. A little more power & I think Kyrotech might be able to cool them all even....

    Just think a 16 Ghz system (for only ~14,000 probably), that would be fun...
  • Since 1 GHz is in the microwave domain, I can imagine it being mind blowingly incredibly expensive. And I don't want my computer to sterlize me either. (makes me nervous to think of holding a laptop radiating those frequencies)

    Somehow I think wider busses and more parallelism is more realistic.
  • by Christopher Thomas ( 11717 ) on Thursday April 29, 1999 @04:48PM (#1910150)
    Quake will not run much faster unless you cool your fav 3D card as well or unless you use only software renderer (absent in Quake3Arena).
    Hence, your example of "benchmark" is not relevant.

    Geometry acceleration is still done by the processor for the time being. AI and physics for your game will always be. Go to Tom's Hardware Guide and check out figures for the same game on the same card using different processors.

    Any game that performs better on a PII-400 than on a K6-2-400 is CPU-bound.

    Any game that performs better on a PII-450 than on a Celeron "450A" is cache-bound.

    Any game that performs equally well on most processors, differing only with the video card, is bus-limited or fill-rate limited.

  • by Christopher Thomas ( 11717 ) on Thursday April 29, 1999 @01:20PM (#1910151)
    First of all, performance won't be terribly stellar for applications that thrash the K7's cache. Main memory isn't cooled, and still has a _latency_ in the 6-10ns range (bus speed notwithstanding).

    OTOH, things like Quake that fit within the cache will run more quickly.

    Secondly, 0.18 micron fabrication should start some time this summer, and should have decent yields for high clock speeds by the end of the year. You should be able to pick up a chip in the 800 MHz - 1 GHz range *without* cooling around then. Drool over what will come out of AMD/Kryotech then, as opposed to now (at the tail end of 0.25 micron) :).

    Anything bought now will depreciate rapidly in value over the next few months, as 0.18 micron fabs are almost due to come online.

  • by Grandpa_Spaz ( 29498 ) on Thursday April 29, 1999 @01:47PM (#1910152)
    According to the Dirk Meyer's presentation [tomshardware.com] given at the Microprocessor Forum last October (and this is the only extensive dicussion about the K7 AMD has given that I can find), the K7 does use Alpha's EV6 bus technology to run the bus at 200 mhz, but the chip itself utilizes many x86 technologies, including 3 parallel x86 instruction decoders and the support for up to 8 MB of L2 cache (not mention SMP). All of this will utilize a x86 structure as far as the software is concern; Win98 (just an example) and Linux should run fine out of the box; you'll just have to worry about the hardware. I seriously suggest you read the about link; it gives a good bit of information about the chip itself (although, remember this was in Oct. of 1998, and it could be different, hopefully better).


Thufir's a Harkonnen now.