Catch up on stories from the past week (and beyond) at the Slashdot story archive


Forgot your password?

Petition Intel Not to Disable SMP Celerons 125

McKing writes "Cpu Review has a petition online to let Intel know that technical users do not want the SMP ability of the Celeron CPU's to be disabled. Several sites have stated that Intel will disable the AN15 pin on Socket 370 motherboards to discourage Socket 370 SMP systems. "
This discussion has been archived. No new comments can be posted.

Petition Intel Not to Disable SMP Celerons

Comments Filter:
  • [] sells OC'd celery chips that they've burned in and tested. Haven't bought from them myself, I just saw it on their site.
  • Ugh...
    K7s(Athlons) will not need 200mhz ram, they currently use pc100 ram...
    And K7 MB's should be pretty cheap by the time coppermines come out...
    However, you are right to question AMD's .18 abilities. The MHZ race is where this battle will be fought.
    Keep in mind that the k7 should still be faster then coppermines per MHZ, but less they are to the p2/p3's. The K7 is still a superior architecture, just the memory subsystem in the coppermines will be catching up, the rest of the core is still slower..
    Learn first, post second...
  • as for the celerons; they were designed to be a "stripped down" version of p2 (and with the new one, p3). thus the celeron shouldn't support things like smp. and imho, shouldn't support the simd extensions. by all means, use the .18 fab because it allows higher clock speeds and lower heat, but adding features that should only belong to the QUOTE High Performance /QUOTE processors.

    IBM invented "functional pricing" in the 50s. At that time they had two similar models of printers that differed in the number of pages per minute printed. The faster model printed twice as many and leased for twice as much per month. If you wanted to upgrade from the slower to that faster model, IBM would send out a tech to do a "field upgrade". In this case, the field upgrade consisted of replacing one of the wheels in the drive train with one half as large and replacing the plate with the model number. They tried to do the same thing with Lexmark laser printers some years ago. The only thing different between the high performance and low performance models of one printer was that one had a ROM that wasted half of the processor's cycles in idle loops.

    when a monopolist can protect their margins and price structure with type of predatory pricing, it's usually called "unfair competition". In the case of the Celeron, the chip is functionally capable of SMP operations. Intel only disables it to protect the higher prices on the "High Performance" line.
  • Where did you hear this?

    FWIW, I believe I heard this on NPR... Someone was talking about approachable technologies, or something. I forget the gist of it, other than the fact that car companies for some reason didn't like the home mechanic. As I recall, it was more about the fact that technologies are supposed to be further out of the control of everyday people.

    The guy could have been full of it. He made it sound like every technology today was beyond the grasp of anyone except the professional, although he did allow that maybe computers might be an exception. He seemed unaware of Linux, homebrew computers (both of the 70's build-it-chip-by-chip variety, and of the overclockers buy-a-board-and-tweak variety) etc.

    I guess I chould check NPR's web site. I think it was a month or two ago.

  • 1)Coppermine = Xeon with 1/2 the cache. K7 is faster than a Xeon already - how will coppermine beat it?

    Xeons are not based on the coppermine core. There are PIII's based on them, Intel is saying that they will start making coppermine based Celerons in the future. See for the story.
  • So car companies should make it so you can't put
    a radio or air conditioner in their car if you buy the model that comes without them...

    hmm, ok. =)

  • Two things. First Coppermine dispite its name doesn't use copper interconnects. Intel isn't scheduled to use copper interconnects until the first proc they design for .13 (Deerfield if memory servers) sometime in 2001. Second Ars Technica has an article [] that covers why it would be difficult for intel to disable SMP. Basically the SMP pin has to have voltage to work, so not only does intel have to cut the pin, they then have to run 1.5V power to it inside the chip casing or on-die.
  • You can even hack the firmware in the fuel system, if you really want to. It's strongly discouraged, of course. There've been numerous articles about people who put "hot chips" into their cars to boost performance.

    Better plug the stock module back in when it comes time to pass the emissions test, of course. . .
  • I suppose that one bonus side-effect of the increasing complexity of cars is that dealerships are making more revenue on repairs and maintenance that "normal people" can't do at home anymore. I doubt that was intentional, and is probably offset a bit by the longer warranties new cars have nowadays.

    Interesting proposition, though. He's probably right about some of it - that technology is getting more and more "black box" to the end consumer. "I just push these buttons, and my burrito comes out of this box heated up!" Probably not a very good thing in the long run - it takes control and choice away from the person using the technology and leaves them at the mercy of the company developing the technology.

    For what it's worth, the "professionals" who work on cars today are mostly just parts replacers, who swap out parts based on error codes the car gives them. They're not the same breed of mechanic as 30 years ago, when the guy who worked on your car actually knew how it worked...

  • I realize that price discrimination doesn't happen when there is enough competition. Ideally, there would be more competition in this market, and that would drive down the prices further and spark more advancements as the various producers compete for marketshare.

    But that's not how it is today. We've got Intel, AMD, and maybe a couple other smaller companies. That's not going to change in the immediate future, so for now we can just consider what would work best given the current market.

    Option 1: Intel leaves SMP enabled the Celeron. This creates a relatively narrow difference between the "low-end" Celeron chips and the "high-end" PIII/Xeon. Many buyers (primarily companies) realize that an SMP Celeron setup is sufficient for their needs. Sales of Celeron go up. Sales of PIII/Xeon go down. In order to maintain maximum profitability, Intel must make price of Celeron relatively higher. (The price may still go down as costs decrease, but it would be higher than it would be with less demand.)

    Option 2: Intel makes SMP an important difference between Celeron and PIII/Xeon by disabling it in Celeron. Celeron is now aimed at the low-end market (where it was originally intended for). It is no longer competing with Intel's high-end chips, and therefore can be priced lower without hurting profits. Less demand for the chip (since it is aimed at a smaller portion of the market) will also help keep prices down.

    Either way, competition is still the major force keeping the prices down. The more competition, the better. But with the low number of competitors, it is better to have multiple differentiated products at different price levels than essentially one product at one price.
  • I want to make an SMP box using K-6/2 or 3, but I haven't seen a dual processor board. I assume the Athalon will be the first AMD CPU to have SMP.

    I just started looking into SMP so spare the flames. I'm sure I haven't heard of _every_ mobo mfgr.

  • Yeah and you KNOW this when you buy it.... no fraud.
  • Sounds like you've bought a lot of Intel's marketing BS. Intel doesn't want to sell quality and all that idealistic crap - they want to make money out of us. Don't try and make them out to be some valiant company doing things for our benefit: this is business and Intel are just as nasty as the next company (e.g. Microsoft).

    How do people frying their CPUs screw things up for everyone else? It seems to me that if people have money to throw around, let them, however foolish this may seem to you. Besides, it gives the chip manufacturers more money - if that matters to you.

    Anyway, who said anything about over-clocking? It sounds like you have a "chip" on your shoulder about something irrelevant. Also, an Abit board, or any other board, is typically cheaper or better value than its Intel equivalent - competition is good. I have an AOpen board... with an Intel chipset!

    Two Pentium II CPUs will set you back considerably more than two celeries. I bought my motherboard and a celeron for less than the cost of a P2 at the same speed.
  • Why the hell am I going to waste my money buying an overclocked 300A when I can just get the 466 to begin with

    Its two celerons at 1.8 * 450 = 810Mhz
    thats why
  • what? that is so the bullshit from a
    guy whose parents buy him everything
    and doesn't have to look around for
    cash to uppgrade his systems.

    And even if you do buy your own stuff,
    get a life - realise there is more to
    life than just the pooter, and use your
    money elsewhere !!
  • Everyone is dumping any division that has anything to do with the Alpha. Its days are numbered.

    AMD can't live long without becoming profitable, and Intel is going to do everything they can see AMD die.

    Cyrix was sold.

    Transmeta or Intel are the only hope.
  • Why wouldnt you use the celeron as a server platform? Personally, i think a dual celeron makes an excellent server! Look at price/performance, you just cant beat that with a PIII or Xeon chip.. Just because a chip is inexpensive, doesnt mean it is not a viable solution for a 'powerful server'...

    Its so sad nowdays... people get quad xeons just to share word documents... :)

    Perfect server: Dual Celeron - 128 meg dimm - Linux - cheap case - 3com 3c905b - (2) 20 gig ide drives, mirrored..... cheap and VERY functional
  • The cache isn't faster, but it's got one less level, so although it thrashes more often, its thrashes are much lower-latency. Depending on the code, this can be anywhere between much faster and much slower.
    "'Is not a quine' is not a quine" is a quine.
  • I read somewhere (bxboards, I think) that the 400 is more oc'able than the 466 (and besides, it's cheaper ;) )

    FWIW, I have a dual celeron 400 box at home and I can run seti@home on one processor, sift through a dataset in 16 hours, 40 minutes, and simultaneously run anything (or several things) on the other processor (and my box isn't even oc'ed)

    good luck with your new dual box and enjoy the sheer power it will give you :)

    Who am I?
    Why am here?
    Where is the chocolate?
  • There is always one of you Beowolf people on these things!
  • Ha ha ha, that's awesome, you're doing (and planning to do) EXACTLY same things I am. Except I have a shitty-looking case. I can't wait for the Athlon. I just wish ABIT was making an Athlon motherboard, I like ABIT. :( I'd love to vote for AMD with my dollars, but the Athlon is a little rich for my Free Market Points(tm). I've already gone the single-Celeron 300a + BH6 jammed up to some ungodly speed, which was far less (including the snazzy new case I got) altogether than a comparably fast PII CPU by itself. However, this is a tide-me-over until sometime next year. I'm waiting 'til AMD does the process shrink, after which we will be able to buy >600MHz chips with fat (2Mb -> 6Mb!!) L2 cache and the new motherboards that have >2-way SMP and support for NUMA. Of course, this is all rumor-mill...
  • Let's be realistic folks. We've seen how Intel operates. There's a bunch of ex-Intel employee web sites that tell tales of how they recruit kids out of college and work em to death. They've bullied various companies to not make, sell, other x86 CPUs, chipsets, etc.

    They've got money, they'll do what will get them more, such as making you pay many hundreds of dollars per CPU to have an SMP system. Many people here like/run Linux, and there's bunches of other architectures to do it on. Alphas aren't all that much more expensive. Athlons will be around enough to buy, perhaps with some SMP boards, early next spring. There's all those Mac ports (do G3/4's do SMP?). Aside from an old Asus T2P4 motherboard, I've lived a long time now without Intel. I can compile kernels in six minutes, play a variety of games, etc. They'll go where the money is, and what you pay for your Celerons doesn't bring in all that much, so they'll only allow P2's and up to do SMP.

    The point is, move on already.
  • First the crap with the 486DX/SX/487 being the exact same chip, then some other stuff [],and now this. Intel is just like microsoft, except their stuff works most of the time. Their processors are bloat, their business tactics are, at times, very dishonorable. I'm running a dual celeron 300a system at 450MHz. And it works great because it's a fast core that's sold cheap to compete with AMD. I spent a lot of money on a dual slot motherboard, and I'm sure intel got a nice chunk of that. A lot of the people who work at intel are really nice, but if you go high up enough, some of those people are complete dicks. I hope the athlon makes them a little more humble.
  • [Somewhat OT, sorry. But I had to get this off my chest.]

    Two reasons: Production costs and Federal mandate.

    "Production costs" because you can build a car cheaper if you use microelectronics (fuel injection) over a complicated mechanical device with all sorts of finely-machined parts (a carburetor). "Federal mandate" because you must satisfy emissions restrictions imposed by the Environmental Protection Agency, and warrant that your vehicle will continue to pass smog tests for a very long time.

    The two issues are intertwined, of course. You can get a car to run cleaner cheaper if you use EFI.

    The whole conspiriacy theory about complicated cars smacks of neo-Luddism to me.
  • Apparently...

    The pin that controls SMP operations cannot just be cut. (Remember that this is basically a PPro/PII die, and they never planned to make a SMP disabled version). If the pin is cut, then the processor is deactivated from the SMP cluster (in the case of the uniprocessor situation, this reduces the CPU count from 1 to 0 -- which isn't very useful :-).

    In short, intel has to do some redesigning of the die, or add an extra bit to the package to supply an erroneous signal, so that SMP celerons would just get confused (and hence not work). I think that Intel's main aim however, is to get rid of dual S370 boards (which can't even be upgraded to PIII's)

    In short, it could go either way

  • is just doing this to get people buying dual motherboards to use PII and PIII chips instead of Celerons. They also are thinking ahead, when the Coppermine comes out it's going to use the Socket 370 chipset which is also what the Celeron uses. There will most likely be dual 370 motherboaords for use with the Coppermine but Intel wants you to buy a Coppermine from them rather than just use older Celerons or having Dell make dual Celeron systems rather than dual PIII or Xeon systems. I'm a little mad about them dropping Slot 1 but I understand it's for performance reasons which is a good thing.
  • I don't have enough money to go SMP, and I don't use Intel processors anyway.

    My next x86 chip will be an Athlon. Maybe if I have the money I might get 2 of them... but seeing that I just got laid off, not likely.

  • 1. Learn how to type.

    2. Did you ever think about sending your broken K6-2 back to AMD for warranty replacement?

    3. Do you have a good heatsink/fan combination on the K6-2, with a nice layer of thermal-conductive paste between it and the processor?

    4. Your probable answer to one or both of those questions is: No.

    5. My K6-2 350 has been happily overclocked for almost a year with no adverse effects.

  • No actually those systems do make sense for people who want to try SMP, but don't have $1200 to shell out for p3s, or $6000+ for xeons. And where would you recommend we spend our money? Maybe we do, and maybe we don't...
  • I doubt that was intentional, and is probably offset a bit by the longer warranties new cars have nowadays.

    Nope that was the whole point. That and Patent restrictions on OEM parts are the only way to get post-sale revenue.
  • The value that they add (manager speak) is removing the risk of buying celery chips for overclocking.

    When you buy a chip, you are garanteed that it will work at the rated speed. Anything above that is a crap shoot. If you really want a 450Mhz cpu and you buy a 300A, there is a good chance you will get a chip that won't do it. If you get a bad chip, then you didn't get as much value as you might have. This company pretests the chips for you. You know that you are getting a 450 cpu. In essence, the company is taking the risk for you.

    It is really no different than intell selling the same chip for different ammounts of money with different #s written on the box.
  • They've always done this. Remember the 486SX/DX? No one really needed the 387 FPU, and when Intel put the FPU into the 486 chip it raised the cost somewhat. Then came cheap alternatives, and BLAMMO! "lets take a fully functioning chip, disable the FPU, and sell it cheap to undercut competitors, even though it still costs the same to make!" sounds a lot like "lets take a fully functioning Celeron, disable the SMP pin from the core to teh package, and force people who want SMP systems to get dual PIII's!"
  • From that, it follows that there is less customization that the car owner can do - it's much easier to adjust a carburettor butterfly than to reprogram a fuel-injector control system...

    Reprogramming a fuel injector system doesn't have to be hard. You can buy a injection computer with little potmeters you can turn, or a rs-232 connection to your laptop pc. A good pc program or a manual on those pots and it is no worse than adjusting old-fashioned carburettor and timing.

    The car industry don't put units like that in normal cars though. The usual injection computers is about as adjustable as a carburettor with the lid welded shut. :-(
  • What are you gaining with dual celeron CPU's? nothing.. intel sells the very same warrantied, tested and supported setup with Dual Pentium II CPU's. So why spend the time and money you wasted on an Abit board when you could get a full powered Dual CPU PII board and 2 CPU's for the same cost.

    The fact you like overclocking may be your reason you want a dual celeron. But, Intel wants to sell quality and since it re-assures its quality its a limited few out there screwing it up for others that end up driving costs and losses on cpu's. So while you may be able to have fun with your celery sticks, its not supported, and intel will do whatever it can to get that point across.

    CPU's are like cars, screw with the chasis and your engine blows.. your simply fudged..
    Intel wants to stop them few fudged up people from draining at its market. from fried cpu's to loosing money in its only profitable market (high end cpu's). Thats what its trying to prevent.
  • Perfect server: Dual Celeron - 128 meg dimm - Linux - cheap case - 3com 3c905b - (2) 20 gig ide drives, mirrored..... cheap and VERY functional

    Wow. You just described the machine I am building, except that I use scsi drives.
  • I fail to see your reasoning.
  • What the guy is trying to say is very fucking obvious to me.

    A Celeron 466 x 2 + Abit BP6 MB is $330.
    ComputerNerd are selling OVERCLOCKED Celeron 300A's with an Abit BP6 MB for $365.

    Therefore, you're getting ripped off.
    You're getting LESS processing power for MORE money. I don't see how this is not obviously clear.
  • Intel's Dirty Secret is a PC Computing magazine article title in the new September1999 issue, page14. The article points out that Intel messed up while trying to beat Cyrix and AMD on the low end, Intel ended up beating itself! According to the article, the price/preformance of a Celeron chip is 1.5 times the power per dollar as a PII and 3.5 times the power per dollar as a PIII. I'm not a hacker, just a computer user. I'd be interested in slashdot comments on this article.
  • The point is that Intel does not want the
    Celeron to further cannabilize its PII (dead
    anyway) or PIII sales. Most persons who do a price performance comparison would obviously buy an SMP Celeron system over the much more expensive
    PIII systems. Intel is still trying to limit
    the damaged they caused themselves on PII and
    PIII sales when they rushed out cheap powerful
    Celeron to stem the tide of K6's from AMD. Its
    all ecomonic, Intel needs a very low priced
    processor to keep AMD on the edge of bankrupcy,
    but its a two edged sword because its so close
    to the power of a PII/III it cuts into their high
    margin sales (or rip-offs).
  • I can never go back to single processor!

  • doesn't matter anyway... other system are out there, more capable smp systems exist. And well. If you want to hack your machine, then you will just have to find a hackable cpu.. Otherwise just use something thats supported or get yourself an athlon.
  • Intel probably doesn't care much about overclockers and others who will run dual (or quad... or...) Celerons. They are out to protect their lucrative server market. The possibility of cut-rate dual Celeron servers would probably scare them enough to disable the Celeron's SMP ability.

    It's kind of sad. I heard the same thing happened in the car industry. All through the 50's, 60's, and 70's people would tinker and tune their cars, fiddles with the carburators, etc. Meanwhile, the car industry strived to make "tinker-proof" cars, since they saw the home mechanic as some sort of threat (not sure why).

    Now, we have the same thing going on in PC's (well, there were always PC tinkerers... now, though, it's gone big time). People tinker with clock speed, improve cooling, and push their systems. Meanwhile, Intel tries it's best to throw up roadblocks.

    It's a shame, really. Intel has little to fear from overclockers and people who will use Celerons in SMP systems. I doubt that there are businesses out there that would trust overclocked hardware, or hardware not offically rated for SMP use in their mission-critical servers. It would be nice if they just relaxed and let people do what they want with their products. Or, perhaps, they are just making offical noises so that if you smoke your new PIII while trying to overclock it, you won't have them to run crying to. It seems they haven't done as much as they could to prevent Celeron overclocking.

    Personally, I'm ramping up to purchase that dual celeron board and a pair of Celeron 466's.
  • It's also worth bearing in mind that sometimes the Celeron can be faster than an equivalent MHz PIII (I think it's something to do with having a faster cache speed). Check out this article [] over at Ars Technica []. Disabling Celeron SMP would effectively reduce the chances of Intel having the fastest x86 based system even more (especially now that AMD have managed to overtake Intel FPU performance).
  • by tallpaul ( 1010 )
    I can't think of a better way to let Intel know exactly *how* much money they are losing, and stand to lose by not disabling it.

    "Uuuh boss.. I've got this petition here.. all these people want us to let them cost us $XYZ"

  • It will probably cost Intel more by disabling it since people will just either buy less, or switch to another company. By having it enabled, at least Intel is getting some money.
  • Regardless if it's true or not, I signed the petition, and I encourage others to do so.

    I personally don't own a celeron processor, nor do I plan to buy one. However, I respect the people who choose the celeron over the Pentium II or III, as well as their reasoning. I don't think it should be a problem to allow users to run Celeron-based SMP systems. (Intel would be hurting their own market. They think that if they disable it on the celeron, they're forced to buy a PII or PIII, when in reality, they could just opt for an AMD based system.)

    -- Give him Head? Be a Beacon?

  • Sounds like just as much "control-freakism" as revenue loss, but what can you expect from a fundamentally closed system that apparently missed the clue train [] by a country mile? "No! That's OURS! You can only do what WE say with it!"

    Bring on the open hardware alternatives, as well as increasing numbers of companies who remember customer goodwill and think in the real long-term, instead of this short-sighted, attention-deficit, post-modern "non-thinking" that gets taught to everybody in this day and age.

  • Can't they just not bond out that signal from the die? Or is there some other signal also fed out through that pad on the die?

    (I wish a URL was given, if indeed there is an article on Ars Technica that explains this)
  • I don't know why they would want to change the whole ppga setup... we should all be grateful for the 300a, although it would be nice to know Intel is listening...

    Have FreeBSD questions?
  • Intel is even now redesigning the Celeron line to take advantage of the new die size (.18?) and copper technology. I saw a story about how they are redesigning the leads from the chip to "shorten the electrical signal path" and so on. Since they are already futzing with those parts of the chip, it'll be simple for them to not connect the pin that allows SMP.

    Which probably also means that this petition is moot, one way or the other. The 600 Mhz copper-based Celerons are supposedly due in a month or two. The deed is done.

    SMP is probably safe in the older chips, though. It would cost them to break the current production runs to disable SMP. THey probably aren't going to go that far out of their way to disable it, but they probably will snip it, if they can do so during the regular course of chip production.
  • The initial posting on this here slashdot says disabling the motherboards. Shouldn't it say disabling the chips themselve.. Intel doesn't make all the Mobo's out there.....
  • Let Intel run its business they way it wants. Go out and buy new SMP Atheron (sp?) or PPC machines instead the day they hit the street. Free markets work.
  • No geek I know would recommend a SMP Celeron setup over a full fledged PII/PIII/Xeon for a production box.

    And if it's not a production box, how many of us would honestly go the Intel-non-celeron multiprocessing route? For the cost of two full-fledged PII CPUs, I can buy 2 celerons and a couple other neato toys.

    Intel won't lose money on me.
  • First off most buisnesses out there care about money first, quality second. They wouldn't stay in buisness otherwise. "Go Amd, Go Cyrix" ? That's strange since I've yet to see a single amd/cyrix SMP systems ;) Yes it would be a shame if Intel chose to disable SMP for their celeron processors, but please don't believe that AMD or Cyrix well cater to your SMP needs either.
  • I can see a market of people who would buy a dual celeron system instead of opting for one of the expensive high-end chips of any manufacturer. I myself, if I had a few dollars to spend would either buy a dual celeron, or not buy anything. It doesn't make much sense to disable something that allows users to add a second Intel chip. People would buy 2 instead of one, so how much money would intel really loose? If instead the person would go and buy a single faster chip, what makes them not go and buy an AMD?
  • isn't pretending that their Celerons are PII's, so what fraud is involved? As far as 300/450A's vs. 466's, if you really can clock a 466 to 700 MHz, then yes, it will beat the 300/450, but unless you can do that, the 450 will win anything that goes significantly out of cache (just about everything except a synthetic computation kernel). Why? Because Celerons are overclocked by boosting the FSB frequency from 66 MHz to something else (100 for the 300/450, for example). This reduces the memory bottleneck substantially.
  • When you buy CPU's by the gross, the price difference between a Celeron and a PIII really adds up!

    BTW, it's "Beowulf", not "Beowolf"
  • if users opt for an amd based system for smp, then they would end up paying the pentiumIII equivalent prices anyway, since in order to get smp capable smp, they need to buy athlon. of course, i'd rather have the athlon system since it seems to be beating the pants off p3.

    as for the celerons; they were designed to be a "stripped down" version of p2 (and with the new one, p3). thus the celeron shouldn't support things like smp. and imho, shouldn't support the simd extensions. by all means, use the .18 fab because it allows higher clock speeds and lower heat, but adding features that should only belong to the QUOTE High Performance /QUOTE processors.
  • Remember, the people building SMP systems out of Celerons are generally hackers trying to build the maximum computer on a shoestring. This enables Intel to sell them two processors instead of one, which undoubtedly offsets the other considerations. I'm considering building a dual-Celeron system myself. However, if they disable SMP, I'll just build a single Celeron machine. They certainly won't be getting any extra revenue from me by killing SMP!
  • AMD? I'm sorry, but I can't go out Right Now and get a dual AMD computer. One of AMD's biggest f*ckups on the K6-2&3 line was the inability to SMP.

    While Athlon should SMP nicely, I hear that the current chipset that AMD is using only allows Dual, and nothing above. Besides which, I know I'm not seeing any Athlons in stores for a while *sigh*
  • True Dat! I myself have just days ago purchased 2 celeron 366 chips which I plan on overclocking to 550+! Buying the 2 chips was a more cost effective option for myself (a student on a limited budget) than buying a second pentiumII processor, by about $50!!!
    In any case what did Intel expect!?! The reason they built the celeron (correct me if I'm wrong)was to compete with AMD. When the K6 (K6-II?) came out, it mashed the PII and and gave the upcoming PIII a run for it's money, so Intel had to build something faster and cost effective. What I like to call a "Ghetto Pentium". They did, the public loves em, and now they complain. Sheesh!
  • Although it doesn't seem to have flown in the x86 world, AMD and Cyrix (and others) created an OpenPIC standard for SMP. Do some searching for it, just not on their websites. :)
  • If Intel shafts us many of us will go to AMD if not there already. Does Intel think we are stupid? Why do they do such foolish things? The people who are into dual processors are most likely knowlegeable about computer architecture issues. They are most likely not newbies who randomly pick processors because of name. I assure you if Intel does this, the sale of AMD processors will increase.
  • by cr0sh ( 43134 )
    I have heard of this - where can I find more info?
  • So my Honda can never ever have a turbocharger bolted on. So my TV can never ever connect to a bigger audio system. So my Playstation can't use a large memory card. So if I fly somewhere with my family I have to upgrade everyone to first class. Do you fucking get it? If buy something I want to use it the way I want - shit even AT&T got the message years ago when they stopped forcing customers to buy phones from them and limiting how many they could connect. It's within Intel's parvue to withhold support from SMP Celerons but to forbid it, to prevent it? Who fucking handed them a flaming sword?

    Tell you what - I'll sell you my pontoon boat but you can never ever add a second outboard. If you want more power you have to put in a racing I/O.
  • ... Are they talking about severing the pin on the CHIP or on the MOTHERBOARD?

    I ask, because of this:

    Several sites have stated that Intel will disable the AN15 pin on Socket 370 motherboards to discourage Socket 370 SMP systems

    should s/motherboards/CPUs/g ?

    Either way, I gots mine...
  • I currently run Celeron 370's, and the population is rising rapidly.

    If the folk at Intel decide to demonstrate their forward-looking approach to CPU design by crippling future Celerons, I will have nothing more to do with them. End of story.

    If that's the degree of interest they show in catering for the low-end SMP market, they'll get exactly the same amount of interest back.
  • I would think your duty as a manger would be to do the best thing for the business, which is not necessarily the cheapest thing.

    I'm not saying that the Celeron was not the best thing for you, but if that's your attitude as a manager it won't be long before you get burned by some of your decisions.
  • Ignoring over clocking, the AMDK6-3 has wiped the floor with the first PIII's that came out in many of the benchmarks I saw. I think there was even a /. story on this a while back, I'll have to check.
    Compare PIII prices with K6-3 or even K7 for that matter.
  • I'm not sure if I agree with all of Intel's business tactics, but they certainly make a better price/performance ratio when it comes to celeron/dual celeron systems. AMD and Cyrix are both bottom of the barrel as far as I'm concerned.

    Well, Cyrix I agree on, but should take another look at AMD, they have come a long way from the K5 and I consider them to be vastly superior to Intel. In fact, I heard somewhere that they were outselling Intel in the retail market. Either way, they are fast, don't have PSN, and they are not badly overpriced.

  • I heard that the pin they would have to disable is actually required for operating the chip in addition to SMP. They would have to run another trace along the chip to substitute for the pin's required functionality which would require a completely seperate manufacturing process. It was cheaper just to advertize no SMP support. Maybe the Coppermine Celerons have an architecture that allows SMP to be disabled but they've already ended socket 370 for that purpose.
  • I seriously doubt many people are goign to be trying to use Celerons (even dual) as their servers. why not go out and get a single PIII or better yet, a Xeon...i work with Alphas and Intels, and we see that usually, 2 Xeons do the work of 1 alpha chip.. so i'm sure that the celeron would require at least a dual SMP setup to compete with a Xeon...why bother??? if intel thinks that people will use SMP celeron servers, they must be smokin somethin wacky...

    as far as cars go, the carburators of 50-70 gave way to much better technology (fuel injection). i don't think this was done to spite the home mechanics, but to improve perfomance, fuel effiency, reduce pollution,'s just i'm not sure how well this analogy holds.

    --carburators do celerons :-)
  • Celerons are not a "stripped down" version of the p2, they use EXACTLY the same CPU core for both p2 and Celeron. The only difference is cache and packaging, and the only reason that the first celerons were not SMP capable (easily) was that the package did not connect the pin to the outside world. The pin was still there, and knowledgable people found that out, and told the world about the "drill and solder" technique for making SMP celerys. As the slotckets and dual celery mobos came out, Intel said "hrm... we're going to .18 micron anyway, so why not rearrange the package a little, and while we're at it we'll cut ALL traces to the AN15 pin. externally and internally!", thus completely disabling any hopes of SMP capability.

    It is the same thing as with the overclocking protection. Why the hell should Intel care if we run "their" CPUs out of spec. If I fry a CPU, well, then I won't complain to Intel about it. But at least give me the option!!
  • No, I'm talking about the newer Celerons, which have two levels of cache, vs. the P3, which has 3 levels of cache.
    "'Is not a quine' is not a quine" is a quine.
  • (or was this macro? damn, maybe I should have studied more too...)

    Price discrimination isn't the result of a happy, healthy, ideal marketplace. It's something that monopolies can do to turn most of that annoying "consumer surplus" straight into profit for themselves. In an ideal market, with lots of producers, anyone who tried to pull a stunt like this would simply be undercut and beaten down by their competitors who didn't disable cheap SMP chips.

    We'll see how soon Athlon SMP motherboards come out. It's not quite in the Celeron price range, but between a bus designed for point to point SMP, and a cheap set of K7-500s with that sweet, sweet FPU, dual PII systems just aren't going to cut it.
  • Where did you get this information that the P3 has three levels of cache, could you provide a URL? AFAIK The P3 only has two, a 32KB L1 cache and a 512KB 1/2 speed L2 cache, while the celeron has a 32KB L1 cache and a 128KB full speed L2 cache, and neither proc has a L3 cache, if they miss in L2 they go to main memory.
  • Like I said, whatever is going to happen with the .18 is already determined. But do you know for sure that the .18 die does, in fact, connect up the pin needed for SMP?

    I'm not sure what the turnaround time is on redesigns, and I'm not sure when the public caught wind of the SMP ability of Celerons. So, maybe Intel didn't have enough time to change the .18 dies if they were only made aware of the "problem" by harware hackers setting up SMP Celerons. There's still the possibility that they figured it out internally, and decided to kill the SMP ability without making any sort of announcement.

    Time will tell.
  • The only petition that intel will hear is the sound of cash flowing to AMD and others. SMP users can already opt for an AMD system, or Compaq alpha, or MIPS, or Power PC, or Sun...

    I repeat, this petition and "I respect the people who choose the celeron..." sort of stuff is just so much masturbation and other forms of self gratification for whiners who simply have to do *something* no matter how meaningless it is just to feel good about themselves.
  • I ordered it on Friday from []. Ok, so maby Intel should be mad at me for getting them gaurunteed overclocked to a total of 1012Mhz... I can GAURUNTEE I'll by 2 550 Celeron SIMDs if they became available and had SMP support. I figger that's going to run me about $300-400. I also know I can't afford or want a dual PIII box. I'm building a workstation for Quakeing/Mapping, thusly I don't need the extra expense of SCSI. It is hard to find a good dual slot-1 mobo that has other decent features and that is inexpensive. (like the Abit BP6).
  • I heard the same thing happened in the car industry. All through the 50's, 60's, and 70's people would tinker and tune their cars, fiddles with the carburators, etc. Meanwhile, the car industry strived to make "tinker-proof" cars, since they saw the home mechanic as some sort of threat (not sure why).

    Where did you hear this? Automakers aren't afraid of the home mechanic. The reason cars from that earlier era were more "hackable" than cars now isn't because the manufacturers decided to make them "tinker-proof" - it's because stricter and stricter government regulations and tighter competition from other companies forced them to computerize the hell out of every system in the car to reduce emissions and increase gas mileage. From that, it follows that there is less customization that the car owner can do - it's much easier to adjust a carburettor butterfly than to reprogram a fuel-injector control system...

    Intel, on the other hand, is just trying to boost sales of their more expensive lines of processors. There's no government regulation forcing them to disable SMP on the Celeron. Unfortunately for them, I suspect it will backfire. The kind of people who run SMP Cels are a) the kind of people who won't buy a dual PIII anyway, and b) the kind of people whose friends ask them for computer-buying advice. The badmouthing Intel will take from this is surely going to send a few more folks AMD's way...

  • I'm not a big Intel fan, and I hope that they do continue to allow SMP on the Celeron. However, it would seem perfectly reasonable for them not to, and could actually (possibly) be a good thing.

    It's been a few years since my economics courses, but most of it is common sense anyway. A business can maximize their profit by selling everything at the highest cost a consumer is willing to pay. Of course, this cost is different for each consumer, making this goal very hard to achieve. The airlines have come about the closest. Everybody else generally has to aim for finding the price that will attract a sufficient number of consumers with a significant enough profit margin.

    One way to help increase the number of consumers willing to buy your product without lowering profit margins too far is to offer multiple similar products with different feature levels. For example, when buying a car, there are usually a few variations on the same model (e.g. Honda Accord DX, LX, EX) at different prices. This allows the company to sell their product at three different prices, hoping that the additional features of the more expensive variations will encourage people to buy them, while the cheaper versions will attract those consumers who do not want to spend a lot of money on their car, and are willing to sacrifice some features.

    Intel is doing the same thing. They have several chips, all with the basic function of executing x86 instructions. They know that some people are willing to spend several hundred dollars for a chip that will execute the instructions, while other people are only willing to spend less than $100. If they sold all of their chips at the sub-$100 price, they couldn't afford to develop or include some of the features of the more advanced chips. But if they sold all of their chips at the several hundred dollar range, they would lose a lot of business from customers who are not willing to spend that much.

    Thus, Intel offers several chips with different features. Now their goal is to balance price with features to get as many customers buying their chips for as much as possible. The introduction of the Celeron was aimed to help do this, by providing a good performing processor for a relatively low cost. Intel's competitors were having a significant impact in this area of the customer base. However, Intel did not want the introduction of the Celeron to impact sales in the server market, where they had little competition.

    Well, the Celeron turned out to be a bit too powerful. Intel found that a lot of people were content with the Celeron for higher-end machines, at a much cheaper price (and therefore, less money for Intel). Disabling SMP is a way that Intel can continue to offer this cheaper processor while providing incentive for customers to use the more expensive PII/PIII/Xeon chips for their server machines.

    This is not entirely bad for the consumer. It (potentially) allows Intel to keep the Celeron cheaper in order to compete without drawing business away from their other chips. And it may even (potentially) allow Intel to keep the higher-end chips cheaper because of increased demand for those processors. Whether these potential savings become real probably depends a lot on AMD, since competition will be the major driving force in keeping prices low.

    Anyway, that's my 2 cents. I don't care a whole lot, since my next machine will probably be an AMD.

"If it's not loud, it doesn't work!" -- Blank Reg, from "Max Headroom"