Submitted by On a Windows Vista or Vindows 7 disk, all versions of the operating system are present, from Starter to Ultimate, and everything in between. So, if you want too upgrade to a more capable version of Windows down the road, all you need to do is pop the Windows disk in, let Windows Anytime Upgrade do its thing, and you’re done. It seems like Intel is experimenting with a similar technology… For its processors.
Yes, Intel is currently testing the water with processors that have certain features disables or limited, so that users can buy upgrade cards later on and improve the performance of their processors. This allows resellers and system builders to increase their margins – says Intel.
“Intel Upgrade Service enables down-the-wire upgrades of PC platform capabilities after the initial hardware shipment, offering unprecedented flexibility to resellers. Now your customers have more options with an easy upgrade path for additional performance or features when they are needed,” Intel claims, “In 2010 we are rolling out a small pilot program offering performance upgrades on Intel Pentium G6951 Processors.”
The pilot program is currently available to a limited number of customers in the United States, Canada, the Netherlands (OHAI), and Spain. The idea is that users buy a computer which has some processor features disabled, after which they can buy an upgrade card ($50) which downloads some software and unlocks the disabled features – double the L3 cache, enables four-way HyperThreading instead of just two-way.
An intriguing concept, but also incredibly confusing, and, dare I say it, misleading to customers, to whom buying a computer is already unpleasant enough an experience as it is. Also, want to bet how long it’s going to take for someone to crack this system?
AMD could use the added market share
I don’t see that happening.
Hack like this allows intel to be more flexible with their prices – possibly making cheap CPUs cheaper, since the can expect $50 extra soon for all sales (and as such better compete with AMD on low end).
I also think it should be quite possible to do this in secure fashion – there is impenetrable hardware involved.
They’ve bought McAffe to make it difficult to toolkits trying to crack this upcoming system ?
You don’t need mcafee to make it difficult. All you need is basic cryptography.
You just need to know basic cryptography to know why this is a bad idea. The receiver is not supposed to be the same person as the attacker.
Which is why DRM is always cracked eventually.
The only place I can see this taking off is in prebuilts with people who don’t know any better. Most of the computer enthusiast market will probably have negative reactions to some degree.
Computer enthusiast market has been irrelevant for years already, apart from the few guys buying nvidia & ati high end gpus.
Sure, because the people that write the code have nothing to do with computing.
Sure…
Well, I can only speak for myself, but… I do CS research and so I obviously do write code… and I have a 3 years old notebook at home, which is, by definition prebuilt. It runs everything I can throw at it (or write for it), and it has been almost a decade since I’ve felt the need for tinkering. And I am not alone in this; most of my IT friends have the same opinion.
In other words, I think vivainio’s point stands, at least on the consumer market. Company/server is a different beast, though.
Yes, computer enthusiast market is irrelevant, but gamer market is still very much relevant. That market has the highest margins of them all.
Yes, because no one ever bypasses digital security measures.
This can be done 100% securely unless the master cryptographic key is leaked – intel already has all the components in place (cryptographic verification has been done for years for microcode updates, we’ve had unique per-CPU IDs with the Pentium3).
keep drinking the koolaid, i’m sure it will come true one day.
The quickest way this happens is if Apple dumps Intel as a primary CPU provider for AMD with the bulldozer line coming up and makes Intel only for custom build orders.
The visibility would make this add-on charge a risk for Intel whose thoroughly enjoyed Apple’s branding w/ Intel inside.
ROTFL. In the computer market Apple’s sales are a small drop in a rockpool next to a large ocean. But nice try.
Why does apple get the newest cpus from intel before dell and others get them?
Are they sure about doing this ?
Well, intel do a very good job for many opensource projects, but as a big company, some managers can take bad decisions.
Don’t know why/whether it is legal to sell something they know in advance it can do more after a simple update and with no reason.
People are they able to sue Intel for this ?
I still can’t understand how I can buy a physical thing (a piece of hardware I can touch!) but limited to use only some parts.
I don’t see why. Car manufacturers have been doing this for years now. The hardware is all there and all you have to do a lot of times is install better software on your car to unlock some features you’d otherwise have to pay money for. The only difference is that they don’t sell you software upgrades, but still the hardware is there.
IBM doing it for some time, you’re not gonna buy but rent.
It already happens with CPUs: AMD’s x3 range processors are just quad-core chips with one core disabled.
The difference here is unlocking the 4th core is free but unsupported as most of the time it’s disabled due to a manufacturing defect. However there are plenty of tri-core AMDs floating about that had a working 4th core locked purely to meet quotas.
Where do you think Intel Celeron processors originated from? Crippling their Pentium II/III/IV equivalents by hardware techniques for QA purposes. Later on they became their own line; but it was pure profit at first.
The difference now is that they are planning to disable parts of the processor essentially via the processor’s microcode; then add a card, which loads some special software to tell the microcode to re-enable those parts of the processor, charging a chunk for the card.
Of course the question then becomes: Where is this card going to plug in? And does it have to always be plugged in? Or do you plug-in it, install the update, then remove it and go on your merry way? If it has to stay in, then how many of these upgrades can you do?
Logically, this doesn’t make much sense unless they’re going to use something like a USB key to unlock it, then have you toss they key once it’s been used. If you have to have a permanent card in your system to enable the functionality, then you can only enable so many additional functionalities depending on how they run it.
And of course you come back to the whole problem of “Requires Windows or Mac”, though Intel does enough with Linux that it would be highly likely (but not guaranteed) to have Linux/UNIX support too.
Yes, but you’ve missed the point: Intels “Celery” processors weren’t upgradable back to Pentiums where as AMDs x3’s were upgradable back to quadcores.
Hence why I used my example and not yours.
Celeron processors were Pentium processors, just as much as those x3’s are really x4’s. The only differences were primarily (i) clockspeed, and (ii) cache. While you couldn’t do much of anything about the cache, you could up the clock on them – essentially the same as enabling that 4th core in the x3’s, with the same kinds of issues. Their clockspeed was lower because they couldn’t pass the QA at their full clock speed. So they were upgradeable just the same for the tech at the time.
Yes I know about the clock speed – I built a dual-celeron system (ABIT BP6) a ~10 years back and are well aware of their ability to over-clock (to this day, it’s still one of the most fun builds I’ve had).
As for cache, well you’re just proving my point as you can’t upgrade a CPUs cache! So that alone was an unavoidable hardware difference between them and the pentiums. However IIRC celerons were also missing one or two extensions that their big brothers had.
So while an over-clocked celeron could compete with a pentium, it still completely false to claim that a celeron is just an unde-clocked a pentium. There were subtle yet very real differences between the two processors, and that is why I used the AMD example I gave rather than this one, and why this example of yours isn’t the same as Intels new strategy.
Never said you could upgrade the cache; and the initial Celerons had zero difference in instructions or features; it wasn’t until later (P4 era) that they started making them drastically different. And you don’t get more cache in an x4 than an x3, since the x3+extra could would still be sharing the same amount of shared cache as it would have had with only 3 cores.
The Celerons differed in: (i) they were under-clocked, (ii) they had part of their cache physically disabled, and (iii) they had further things disabled based on what passed or did not pass QA (which is where the subtle differences came from).
AMD did the same thing for the x3’s – make an x4, sell it as an x4 if it passes muster, otherwise, disable a core and sell it as an x3. Saves some money by selling what would have otherwise been tossed.
And if you note – I said it was the equivalent for the tech of that era.
Seriously, why are you arguing with me?
You keep telling me these points as if it should be new information to me yet I made the very same arguments right from the start.
The crux of the matter is there are hardware differences between the celeron and the pentium (something you’ve since agreed with) – thus why I cited AMD when describing a processor you could upgrade without buying new hardware.
That should have been the end of discussion, but here we still are….
As far as I can tell you are trying to argue that AMD are already doing this, Intel aren’t (by trying to reverse the Celeron/Pentium example into a weak one to try and ‘prove they’re not the same hardware) and then trying to paint over and avoid talking about what Intel are trying to do now.
What’s your point, because you’ll find a comment from you at the root of this thread that kicked it off?
Edited 2010-09-23 00:55 UTC
The next cpu will be from AMD
I have a hard time figuring out if I think this is a kinda good idea, or if I hate it with a passion. If there’s no hidden costs (except the upgrade ending up being more expensive than having bought a faster CPU to begin with) to the end user, it could be OK…
As for crackability, this could be done pretty much 100% secure. There’s already (public-key?) cryptographic verification in the CPUs for accepting microcode, and we know Intel has the capability of doing per-CPU unique numbers (remember the P3 days?).
Famous last words.
Seriously though you should really reflect on what you’re about to say before making a statement like that. A free upgrade is enticing and will surely draw a lot of attention to cracking the Intel upgrade security. The problem with “100% secure” guarantees is that while something can theoretically be “100% secure” the devil is in the details. A lot of security schemes are broken because of poor implementation not poor cryptography.
That they can make it so that it can’t be cracked/unlocked by some sort of mod-chip.
After all if Intel can unlock features, they why wouldn’t others be able to.
Patents.
Ah yes, because crackers just care so very much about patents.
If they implement it right, it can be done pretty much unbreakable. Good luck trying to intercept internal CPU logic that doesn’t communicate across the bus; sure, initial transfer of “unlock key” goes across a bus, but from then on everything can be done internally.
Its quite a good idea but I am not so sure people will go for it.
I think it comes down to how much of an improvement it will grant. Will a user be able to SEE the difference? I would suggest most ordinary people dont use their CPU to its fullest. They will pay for the upgrade equating and feel ‘conned’ as its not super quick..
Please explain how this could be considered a good idea?
“Buy this computer for $XX less than that one, and if it’s too slow pay $50 to speed it up!”
What happens if $XX is only $20 or so? They just screwed the customer out of money they could have saved by buying the AMD based machine sans shenanigans. This kind of crap only serves to further confuse the ordinary PC buyer. It’s bad enough they have an obtuse numbering system for processors that has no bearing on speed or cores/threads, now we have deliberately crippled hardware with an enticement to spend more money to make it do what it should have in the first place!
I’m not going to buy a blender capable of 16 settings but only provides eight until I buy the upgrade card and punch in the code. That’s just senseless. Most PC buyers see computers as another appliance and they want it to just freaking work. Making things complicated can only hurt sales.
This stinks of bait-and-switch bullshit in the worst way. I bet Best Buy is drooling in anticipation though; they already train their employees to get the most money for the least product as it is.
And the worst part: We still are playing catch-up with the economy, and owning a PC these days is an expensive necessity. Why the hell should any American company be so willing to further gouge the already stretched-to-the-limit consumers?
Why buy an i5 over an i3? or an i7 over an i5?
Lets be honest, this complexity is nothing new and already in place eg the Core i7 800-series and Core i5 CPUs each adopt different permutations of the fanciest of the Core i7 900-series.
A user would then be enabled to purchase a ‘cheap’ PC and upgrade later. Intel would benefit as they are producing one chip design instead of 3 which Should reduce cost of manufacture. One hopes this would be passed on, but who knows?
If this idea comes to fruition or not is another matter but I see nothing wrong with offering the upgrade from an i5 to an i7 post sale as opposed to POS.
I would argue the point of ‘AMD based machine sans shenanigans’ as most basic users dont have the ‘anti-intel’ mentality. They know the name ‘Intel’ and trust it as a result. You have to explain WHY someone should by AMD. Intel is the default in mindshare. For better or worse, its the market reality.
I don’t buy this at all, it’s not an anti-Intel mentality. When AMD introduced their amd64 line people bought AMD because they were much better than Intel’s offerings, when Intel introduced core2 people bought Intel because they were much better than AMD’s offerings. If Intel goes down this crippled route I’m certain it will help AMD, but it won’t mean squat if their cpu offerings aren’t up to the task, people wants the best bang for the buck, very few care if it says AMD or Intel on the cpu box.
I will happily agree that during the amd64 days their offering Was better and I myself bought a couple. But in terms of joe public who dont compare benchmarks, they bought whatever was in the box HP/Dell/OEM provided. Most of the time, that was intel (in the UK anyway).
I honestly cant recall an AMD advert on the TV but I can Intel’s *bum bum bum buuum*
This coupled with many other factors meant that AMD was the ‘one with the strange processor’ the chap in PC world had to explain was different (badly)
I hope it fails or gets cracked.
An intriguing concept, but also incredibly confusing, and, dare I say it, misleading to customers
I don’t like the idea myself, but how is it misleading to customers? They’re being sold a chip marketed with certain capabilities, and they’re getting a chip with those capabilities. It’s not as if Intel are being deceptive over what they’re doing.
And it *does* make sense from Intel’s point of view. It means they can manufacture a single part, and sell it at a variety of price points. Someone willing to spend $X can have the fully enabled chip, someone willing to pay half that can have a chip with half the capabilities, without Intel having to manufacture two different chips.
Well, this kind of technology will add more flexibility in market – for the sellers. For customers there will be also some benefits – buy cheaper device, then if needed upgrade it. But the question is, how full featured CPU compares to CPU with all features bought. That “upgraded” in total may be more expensive, than full featured by default. Also the number marketing thingy here will make a trick. Well, a “feature” in CPU technically would cost like 11.60, but the number marketing will ask 50, because all functions cost 50.
It will be like getting a V12 but with only 8 cylinders working. And for 50 extra bucks you can open the trunk!.
Actually this idea is already implemented in modern cars. Things like Cruise Control and all the other fancy stuff in current cars is implemented in all cars (say of one specific model) but the customer has to pay for it to be enabled. Same applies to the engine control software by the way.
Customer don’t seem to care and I don’t know why this should be a problem.
Google “anti-feature’
This looks like an “anti-feature”:
http://www.fsf.org/bulletin/2007/fall/antifeatures/
If they start doing this on a large scale, I think I will switch to AMD processors.
… in fact this is just an complementary version of the marketing already in place based on binning. And most chip manufacturers (at least in the processor and GPU markets) do this already, Intel is just taking it to the next level.
Whether we like it or not, Intel really is not in the business of making CPUs, as much as they are in business to make as much money as they can. I don’t particularly like it, and I wish humans had evolved past these silly myopic systems and “isms,” alas that is neither here nor there…
Edited 2010-09-19 22:26 UTC
Not only that, but IBM and other big iron companies have been doing this for years, maybe decades.
There was a VAX in the 80s which could be “upgraded” by a microcode swap which, essentially, removed some no-ops put in there to deliberately slow down the machine. IBM used to do this on AS400s too…the upgrade was a hardware dongle IIRC, but I think it just told the OS to let the processor execute jobs for a greater percentage of the time (the other part being no-ops).
Anyway, is it lame? Sure!
Because for IBM mainframes you pay for licensing/usage of the CPUs. All of the mainframes come preinstalled with more CPUs than you ordered for failover and additional power when you pay. But this is in a nieche market where the buyer is well informed, not uninformed consumer market – that what makes it bad….
But naturally – Intel isn’t a charity, it’s a business that exists to make money for it’s owners. I’m not sure why people expect it to behave otherwise.
No, “binning” is marketing in a certain range of speed and function and/or disabling features that didn’t pass QA during testing, so the customer doesn’t get a buggy chip. They can sell it at a lower price as a less-featured item without having to deliberately fab such an item, and the customer is guaranteed a certain level of performance and features. It’s a win/win situation; the chip maker doesn’t have to throw out less than perfect product and the budget-wise customer can feel confident they are getting all the performance they need at a lower price point. A bonus is the tech-savvy customer can increase performance via overclocking, at the expense of warranty protection, without having to hack or crack anything.
This new idea is offensive and potentially confusing. Imagine this scenario:
Salesperson: “You can spend $600 on this super-fast PC with X speed processor and Y level cache” — already confusing enough right? — “…or, you can get this PC with half the speed and cache for only $580!! What a deal!”
Customer: “So I’m paying only $20 less for half the performance? But what if it turns out I need that extra speed for my favorite games or programs?”
S: “Well then you just buy this handy dandy processor upgrade card for $50!”
C: “Wait, that’s not a new processor or one of those fancy PCI thingys, it’s just a code to type in? How does that work?”
S: “Why, you just type in this code and all the features on your CPU are now unlocked!”
C: “…So you’re saying these two computers have the EXACT SAME HARDWARE but one is artificially crippled so you can upsell me a string of text? …Hey, I don’t see that bullshit on that AMD machine over there…”
No, what you described is only a part of the binning process. Not all binning is done to increase yield by creating SKUs which can reuse parts which are not 100% compliant.
By the time a process is mature enough, most manufacturers will actually sell literally the same part at different speed/performance grades that have nothing to do with the functionality of the part and more to do with the specific price targets the manufacturer wants to hit.
For example, many SOC parts are now sold with specific sections/functionality of the chip disabled to meet price points for specific models. Even though, technically most of those SOCs are fully 100% functional. It is cheaper to do the disabling before the packaging stage than creating a new mask/fab line for a specific iteration of the SOC.
Someone will point out that you can get really cheap CPUs from AMD. Sure. But for those of us willing to pay $500 for a CPU, $50 isn’t much to unlock some features that up ’til now would double the price of the processor.
The nature of cpu manufacturing encourages this type of model.
This will also encourage power savings which is a good thing. It doesn’t make sense to have everyone on 2.8ghz dual core cpus when most people are actually fine with a single core.
I’d prefer the opposite myself: a 1.8GHz dual core system tends to feel faster than a 2.8GHz single core, and will run cooler too. That’s based on my own experience of course. Certainly, tasks like video encoding — which a lot of people who own iPods and other such devices do daily — benefit much more from multiple cores than from astronomical speed ratings.
Of course, I’ve always been of the opinion that parallelization is the future of high speed computing. One day we’ll have a commodity priced 16-core single CPU at 1GHz per core that will outperform the fastest hexacore system available today, with much less power draw.
Edit: Sorry ozonehole, I didn’t see your post about 16-core chips, I just pulled that number out of my head. I like your thoughts regarding ARM taking over.
Edited 2010-09-20 02:08 UTC
I agree and this is due to one core being available even if the other is choking on a process. But for the typical user a single core @ 2ghz is fine if it is a recent processor. I do think all computers should come with a dual-core and 3+ gigs of RAM but mainly because OEMs pre-load them with so much crap and leave auto-scanning on.
That could take a long time due to diminishing returns. Only in gaming has software become more demanding. Most consumer software runs fine on a P4. There is more incentive for Intel and AMD to sell dual and quad cores as long as they can at the commodity level and keep trimming the die. People have enough power, what they want is more battery life.
That was what I meant by “much less power draw”.
Yea and I am saying that Intel and AMD are not going to be motivated to bring that level of scaling power to the consumer for a long time. They will just keep cutting the die while only adding minimal increases to processing power.
You’re right I’m sure. Anything to improve their bottom line, I guess.
Hmmm…maybe Intel can get their manufacturing line down to just one chip with 16 multicores. They ship it with just one core enabled, and you pay $50 for each additional core you want to enable. Works in every machine from a netbook to a mainframe. Though expensive to manufacture, Intel bets the company on this great new technology, selling the chips for a loss on low-end machines, expecting to make a fortune from the upgrades, and then…
…whoops! Somebody finds a way to jailbreak it. Google and everyone else decides to run their server farms from Asus eeePCs.
Intel goes bankrupt.
And at last, the ARM chips come into their own! Goodbye x86 processors. ARM rules the world!
OK, I just woke up.
Edited 2010-09-20 00:23 UTC
How is buying a computer an unpleasant experience. I love going down and buying one.
I hate shopping for clothes (which I will say is rare) and food.
Cars
Computers
TV’s
ummm
you get the idea
Well, you post on OSAlert so I’d have to guess that you’re a techie. For someone like you or me, this upgrade card BS is transparent and irrelevant. We’ll just buy a non-crippled machine, or better yet the parts to build or upgrade our own, and won’t be affected by this.
The victims of this scheme are the everyday person who sees a PC as just another appliance in the home. They are the ones who will get pushed into the crippled hardware and talked into the $50 string of characters to unlock what they already had. That’s almost pure profit for Intel as it requires no new hardware and only pennies worth of paper and plastic for the functionless upgrade card.
It’s a scam, period.
Where is the scam? The consumer is being told what he is buying up front.
The level of power provided at the default level is more than adequate for typical use.
You may not like the idea of an artificial limitation but as with software it makes sense from an economical point of view.
They have been doing this with cpus and video cards for years. There are single core AMD cpus that can be unlocked into dual core cpus with a hack. It often makes economic sense to produce a single die and then place an artificial limitation on it to meet low-end demand without cutting into high-end margins. Sure it would be nice if NVIDIA made one high-end gpu and sold it to everyone for $50 but that would mean far less R&D for future products.
The problem with that argument is that the customer pays more up front for the full featured chip and reaps the benefits immediately. There is no bait-and-switch involving activation codes purchased at almost pure profit. If you’re okay with lower performance, you pay less. If you want more performance, you pay more. Either way, you get what you paid for the first time around.
The scam is in the fact that the hardware is deliberately crippled so as to force the customer to pay $50 for what is essentially a string of text in order to unlock what they already paid for once. And, as others have pointed out, this scam is possibly limited to Windows based machines, and perhaps is even a pure software switch so that if you have to reinstall the OS you lose your performance and must shell out another $50 to get it back. I’m sorry, but I’d feel much more comfortable knowing that my hardware’s raw performance is not a variable based on a highly volatile software switch.
To put it another way, it’s being forced to pay twice (or more) for one physical purchase. It’s dirty and underhanded. Didn’t Intel follow the Sony debacle?
Well, this all depends on how it works. I mean, if the chip with all features unlocked is $100, but you can buy it crippled for $75 with an option to unlock the rest for $50 later on, then just buy it for $100 and be done with it.
In other words, I guess I don’t have a problem with it, so long as there’s an option to pay for it all up front and spend less than you would getting it piecemeal.
Citation needed.
Hi,
I’ve been looking into it a little, and (at http://www.intel.com/technology/product/ius/howitworks.htm?iid=ius_… ) Intel say:
“The upgrade enables changes to the firmware (driven by the Intel Active Management Technology Management Engine in the chipset) that in turn modify the hardware.”
From this I’d assume that changing/reinstalling the OS won’t cause the “$50 unlock” to disappear. A firmware upgrade might (or might not) cause the “$50 unlock” to disappear (but I’m not too sure which firmware is modified – the normal/system firmware, or the Active Management Technology Management Engine’s firmware).
I’m also unsure if only Windows can be used to unlock, or if you can use any OS to unlock. Intel say (same web page) that it’s “architected to be highly flexible and should be compatible with most third-party management consoles“; but I’m not familiar with Active Management Technology and I’m not sure which management consoles exist for which OS/s; and to be perfectly honest, every time I try to unravel what “Active Management Technology” actually is I drown under a sea of seemingly unrelated buzz-words.
– Brendan
I wish I could mark your post “Informative”, thank you for the legwork. I’m still pissed that they want to run a bait-and-switch game like this, but if the unlock is persistent in firmware it’s not quite so galling.
I bought a cpu before that was deliberately clocked low by Intel to encourage sales of high-end models. Was I scammed?
Early AMD triple core cpus were actually quad cores with the L3 cache and one core disabled. Was this a scam?
http://www.cpu-world.com/news_2010/2010041501_Unlocking_fourth_core…
This is very much dependent on your viewpoint.
Buisness interests will very much like to push the viewpoint that increased sales figures is better for everyone … they will say “better profit, better returns on investment, more jobs, etc, etc”. From this view derives the notion that the price should be set at “the highest price that the market is prepared to pay”.
Everyone else, which is the vast majority of people, would naturally have the opposite point of view. It is better for them to lower their costs. From this view derives concepts such as “value for money”.
Being a person myself who does not make or sell silicon chips, but rather one who simply uses them, let me tell you that from my point of view (which aligns with the vast majority of people who are also in the category of not being makers of chips) that the second concept, one of “value for money”, makes vastly more economic sense than the concept of buying something priced at “the highest price that the market is prepared to pay”.
This is especially the case where there is a tendency in a market to believe (against the actual facts) that only one supplier can deliver. For example, there is a belief in this market that only Intel can deliver on “binary compatibility”, when in actual fact there are other options from totally different manufacturers who don’t deliver binary compatibility at all, but who still deliver far better value for money because the customer didn’t actually need binary compatibility.
Edited 2010-09-20 06:36 UTC
“Where is the scam? The consumer is being told what he is buying up front.”
Like most PC buyers have any idea about specs? Wait until the Marketing department gets hold of this. The “upgrade” feature will be buried or left out of the advertising. Once they get the PC home and open the box they will be confronted with a friendly little note saying “Get the most out of your PC for only $50!”. This is basically only hiding the true cost of the machine in order to have a competitive advantage at time of sale.
Someone will point out that you can get really cheap CPUs from AMD. Sure. But for those of us willing to pay $500 for a CPU, $50 isn’t much to unlock some features that up ’til now would double the price of the processor.
So I buy a CPU for $100 that I can enable features on for $50. So I actually bought a $150 CPU in the first place. Apparently it didn’t cost Intel any more to produce that $150 CPU than it would have to produce a $100 CPU — otherwise they would either have to charge $150 for it in the first place, to make any margin, or make two completely different versions.
So in this case it looks like Intel has come up with a way to make two hugely different levels of profit margin off an identical profit. Makes me wonder if they haven’t been selling us all hugely powerful CPUs forever and just dumbing them down a bit. Good for profits, bad morally, I say.
AMD/ATI, NVIDIA and Intel have all done this before.
What is immoral about selling a cpu at differing clock speeds even if the die is the same? You are being offered different levels of power/price, why does the manufacturing process make it immoral?
AMD is selling a triple core laptop cpu. That is freaking awesome. I love Intel cpus as well. All these multicore low temp cpus are so much better from what we had 10 years ago. Has everyone forgotten how much 1ghz cpus used to cost? I remember getting a Duron 800 and thinking about how smooth it ran XP. That is 800 as in 800mhz. I think I paid about $130 and it was an excellent deal compared to the pentiums.
Who gives a flying F if people at Best Buy have to pay 50 bucks to bump up their already fast cpu. Intel is a great company, they deserve that 50 bucks. Americans will drop 50 bucks on pizza and drinks. I did it Saturday night and I didn’t get a boost in power that will last for years.
There is nothing necessarily wrong with it at all, if the reasons are not simply to rip people off, but rather to offer them better value for money.
For example … suppose the manufacturing process for chips throws up a certain defect rate. Suppose I have a wafer with a number of quad-core CPUs, and that wafer throws up a 50% defect rate. On 50% of the chips on the wafer, not all four CPU cores are functional.
Now suppose I make it so that I can isolate power from a dud core, and by doing that I can salvage what would otherwise be a dud chip to now be a working three-core CPU, but with one unpowered area of silicon. Now I can have a wafer yeild 50% working quad-core chips, plus say 40% triple-core chips.
Now I don’t have to throw away 50% of the chips, I can now sell 50% as full-spec quad cores, and a further 40% as lower-priced triple-core parts. Everyone wins. I can even sell the quad-core parts cheaper, because there is less wastage in the whole process. This is pure win-win.
But that is not what Intel are offering here at all, it is only superficially the same. What Intel are actually offering here is a win for Intel, and a tax on everyone else.
I don’t think “tax” means what you think it means…
I’m not sure about the bad morality part but:
It does cost more but only marginally. Having more cash means more stuff on the silicone cell, more energy consume… But this is measured in cents/CPU compared to development and handling costs.
They have and they let us know it:
All CPUs from a generation are, in theory, identical. But they test each CPU and find out the max speed it can run, and remain stable.
So they run the CPU at 2.4 GHz, if it overheats or smth, they run it at 2.2GHz and so on until they find a speed at which the CPU performs well.
I don’t see any reason that, if there are a lot of 2.4 CPUs, not to sell some of them at 2.2, where the price (much) is lower but the market is bigger. This will help them keep price up on the high performance segment, but keep them down on the (normal) consumer segment.
About the upgrade part, I don’t like it from the marketing point of view, but I don’t find any morality point of view.
I’m sorry but you put too much trust into corporations. Yes that is their explanation and that is often the case but there have also been plenty times where clock speed clearly could have been much higher at the low end but they wanted to use tiered pricing.
This is true for Intel, AMD/ATI and Nvidia. I remember an AMD case where the company was embarrassed over people who figured out how to unlock the full cpu with a hardware hack and then changed the die.
I want what I pay for, If in the future, I can pay an upgrade fee with changing equipment and get what I pay for, thats fine with me. I want what I pay for. I dont want it taken away down the road…(SONY).
I will consider the cost of upgrading, versus replacement, before I pay. Common sense…
If this is their business plan, fine, just dont try to pull some shabby SONY crap on the public…
…if I run Linux? BSD? Or hell, if I re-install Windows will it remember that it was upgraded?
This is not good… not good at all. Intel, you lost me, unless this quickly flops before getting out of “test market” phase.
I recall reading about many Intel processors, even ones you would expect to, not supporting hardware virtualization. That was strike one. Here’s strike two. Pretty soon my mind will be made up: AMD only.
What’s so hard about buying what you f***ing want in the first place?
Edited 2010-09-20 02:37 UTC
That’s the bug that bit me. Granted, my system was built from donated parts so I shouldn’t complain, but at the time I was under the impression that all Core2 chips had VT-x. It was interesting to see the vast difference from my old single core AMD chip with AMD-V to this C2D without it, when running VirtualBox.
It was also an eye-opener to find that the OS X version of VirtualBox requires manual editing of the configuration file for each VM to turn off the VT-x flags.
So apparently, It is intel’s fault they can’t read your mind and figure which processors should offer the arbitrary features you “expect to” because you are too lazy to check the actual specs for the product you are buying?
I have no clue how moving over to AMD is going to change that for the better.
If they are honest about what is inside the computer and what you have access to then it is not a problem. The problem is the consumer will have a hard time comparing computers at time of purchase if each computer has multiple specifications (what is currently enabled and what potential it has). Marketing departments will advertise the machine with the maximum specs whether they are enabled or not. That is just the way it works.
It is when someone gets a higher-end processor, expecting it to be more capable, and then they realize that it doesn’t have what many people would consider a higher-end feature. And then when they realize that some random lower-end processor they sell *does* have this feature, shit hits the fan. Is it really such a bad thing to expect a more powerful, expensive processor to have more powerful features? Really?
Well, just about every recent and semi-recent AMD processor has hardware virtualization, and a lower price in general. Which is why the last couple computers I recommended to people had AMD processors inside. And AMD, so far, isn’t pulling this shit on us (so far…).
So far, I haven’t been bit by Intel since AMD’s been a cheaper in general for approximate (or more) power and more features, but all this stuff adds up. My opinion of Intel is not too high right now. With AMD, so far at least you get what you expect… Intel’s line (all the way down to their completely non-descriptive names) is a horrendously complicated mess.
People are talking about ‘speed’ where as I think what will happen are features will be disabled and enabled when they pay extra – for example, why should an end user be forced to pay the virtualisation feature when they might never use it, or encryption acceleration features and so forth. The cost to Intel of keeping the features on the chip is probably cheaper than removing them but it enables them to sell a cheaper chip and then allow people to upgrade without replacing the chip. The cost to keep the feature on the chip maybe a few cents but the cost to upgrade will be $50 meaning that the margins will be massive in terms of after sale upgrades.
The cost to keep the feature on the chip maybe a few cents but the cost to upgrade will be $50 meaning that the margins will be massive in terms of after sale upgrades.
In other words Intel could nickle and dime you to death with their crippleware CPU in comparison with the lumpsum version.
You got it – just like the mainframe vendors will sell a machine with the CPU sockets maxed out but only the ones you’ve paid for are enabled. Another example of a high end feature scaled down for the little people thinking they’ve got a great deal – in much the same way Microsoft nickle and dime end users with half a dozen versions of Windows when a single version would do the trick.
Its a good deal for us technies if this gets cracked. Typical folks will still fall for the extra $$$ for more speed while those of us who are savvy will get better performance for much less.
To me this really indicates that the desktop is in severe stagnation. The old school cpu manufacturers can’t figure out how to innovate anymore so its come to this.
Another aside…this tells how poorly amd is doing. Last week I got ahold of a dual socket 24 core amd system. Running against a dual quad i7 the amd was clearly half the speed per core on heavy cpu load. Mixed loads the amd seemed to do not quite as badly. The good for AMD was that the system itself was overall faster than the intel on the multithreaded load. With intel 6 cores coming out that spells more trouble for AMD.
Edited 2010-09-20 05:04 UTC
Firstly, is this unlocking software only available in Windows? What if you’ve wiped it and are running a Linux or BSD distro instead?
Secondly, do you need this unlocking software running all the time or does it modify part of Windows and then exit (in which case, a simple software crack to simulate the Windows change would do)?
If it modifies the BIOS NVRAM in some way (i.e. sets a flag or something that’s remembered even if you re-install Windows/Linux/BSD), then again a software hack to simulate that would do (although BIOS code may need to be disassembled and patched if the “flag” is more than a simple boolean).
I suspect someone will crack it fairly quickly, but it could get messily exponential – it might be a BIOS hack that’s needed (and BIOS releases come out frequently and are unique to each PC model, never mind OEM). Sadly, there’s no actual technical details anywhere how this “upgrade” is done (other than “some software is downloaded and run” [and obviously has a code to authorise it]) so we can only guess at the moment.
This will get ripped apart by crackers. Let’s be honest if Intel can afford to ship the product (with all these dormant features) in the first place then people will be wondering why they should pay more for what they have already got. Most will take the free option, and good luck to them.
It just devalues their product, surely Intel can see that!
So, you buy hardware which will have its features locked unless you pay extra. Then you buy [non-foss] software which will have its features locked unless you pay extra.
You know, if everyone wants us to basically switch to some kind of subscription scheme both on hardware and software, then they’d f*ing better give us all the hardware for free.
Until then, f* the locked down pc hardware. It’ll just make new market opportunities for other hardware vendors (give me those multicore arm cpu’s) and create a new pc modchip black market.
It could be an interesting model if Intel is fair in its pricing scheme. If you buy a chip with 50% of the capabilities disabled and you only pay 50% of the full feature set, this could be a winner. It gives you the opportunity to buy at reduced pricing and afterwards get more power when needed without having to put in a new CPU.
It’s just that the temptation is always there for Intel to reduce the capabilities to 50% and still charge 75% of the price. The upgrade to unlock the rest at 50% pricepoint would mean you pay 125% of the market price in comparison to a chip that doesn’t artificially limit you.
I also wonder how the unlock works. Is it a microcode and hardware thing, or is it some tweak in the Windows drivers? If so, how do you perform an unlock on an alternative platform? If it is purely an OS tweak, what is to stop a “pirate” from saying “Arrgh, here is your unlock keygen en driver tweak?”
Yeah well, Intel are obviously not going to sell CPUs for less just because some features are locked. More likely the customers will pay more for the privilege of having all the bells and whistles.
Reminds of those “entrepreneurs” on ebay, that take a second-hand $200 SPARC server, break it up and sell the parts at inflated prices, totaling something like $500.
I mean are there people stupid enough to fall for that?
Oh your machine’s graphics performance sucks? Well that’s $50 to enable FPU.
Oh you’re not happy with your 100MHz CPU speed? Well it can go up to 3GHz, it’s just $5 per 100MHz.
Oh you want to run Linux/BSD/* operating system? Well that’s $100 to enable alternative operating systems support.
Intel go fcuk yourself!
Edited 2010-09-20 11:13 UTC
Ahhh…the smell of sweet, sweet editorial bias in the morning.
The title of the original article: “Intel wants to charge $50 to unlock stuff your CPU can already do”
Perhaps an alternative title: “Intel to disable features and lower prices by $50”?
The truth is that we don’t know what pricing model Intel is going to use for this system. The photo in the article mentions upgrading “from 2- to 4-way multi-task processing” among other things, which I’m assuming refers to cores. News: not everyone needs four cores, and plenty of people would pay $20 less, even $10 less.
Many of these people wouldn’t even upgrade a CPU anyway. For all we know, Intel’s target demographic with this model are the people who buy a new computer once it’s “broken” by malware and viruses. Those guys weren’t going to be upgrading anyway, but they’ll be roped in by cheaper prices.
Why would you even consider this to be a good deal?? It is designed not to offer you a lower price, but make you pay twice for something that you have already payed.
It’s like you go and buy a house, but you can only use half of the rooms, even though you own the entire house. The other rooms are bricked up, unless you make another payment of %10.
This practice is totally despicable. It is rife in software industry and now hardware manufacturers decided to jump on the bandwagon.
Did you actually read the article? Where are you getting this from?
Well, 50 bucks,,,,Meh,,,,,as long as it comes with a tee shirt.
Once upon a time, Intel was delivering 486 Overdrive CPU extension for 386 or 486 SX that was, basically, a 486 DX that disabled the obsolete CPU. You now had twice more silicon in your case, with only the ‘half’ working.
That’s the same, yet more ‘green’ in the way that you don’t have to manufacture/buy another chip, just unlock the current one. Intel just takes a little more money without real added feature, just added benefit :p
Kochise
Actually, all CPU/GPU manufacturers already do this.
When chips first come out, yields are low, so prices are high. But once the yields start getting good, they start producing lots of high end chips.
So, now AMD/Intel have a bunch of 6 core cpus that can run 3.33ghz, but they need to sell a bunch of quad cores running at 2.8ghz.. What do they do? They bin it lower and sell at a cheaper price.
Same thing with GPUs. Got 1800 SPs, but disable half of them and sell cheaper even though it cost them the same to make it.
Now, one step further. Instead of having all of this overhead of binning it at a lower speed, they can automate the process and just change the micro code. This would push down prices and reduce the amount of extra equipment is needed for certain special versions of some CPU lines.
But wait, because they can control everything via micro-code, for a small prices that is MUCH lower than a new cpu, they could let you “upgrade”.
Even though it sounds stupid, it still means lower prices and more money saved on average while giving more choices.
Can’t afford the 3.33ghz 6 core? Nope. Buy the 2.8ghz quad and wait a few months for the “upgrade” to come down in price then pay the $50 to activate more of your chip.
Or you can just stick with the current system where you have to buy a whole new CPU for $300 and AMD/Intel have to have a ton of wasted equipment to support multiple lines.
More money wasted for the manufacture and more money wasted for the customer.. yeah, sounds sooo much better.
It sounds stupid, but really, it better than the current system.
Actually, working for money is a stupid system. People should do the work that they like. I love working on computers and I would work “for free” if I didn’t have to worry about paying bills.
FYI. I remember circumventing AMD’s version of this with a graphite pencil back in the day.
Here’s the issue that everyone forgets. It would be f’n stupid for a company to make several lines. It much easier/cheaper to to create one line. Now, a chip manufacturer comes along and makes a chip. Some run fast, some run slow. What do you do? You sell the fast ones more expensive and the slow ones cheaper.
But wait. Later on, nearly all the chips being produced are fast. So now they have a ton of fast chips and only a few slow ones. They don’t want to sell the slow ones at a loss, but why would anyone want to pay only slightly more to get a fast chip?
This is where the problem is. Once a fab matures, nearly all chips are good. They need to *artificially* inflate the price of the fast chips. But they can’t charge more now for something. So the cost of producing fast chips gets cheaper and cheaper. What do you do when nearly all of your chips are fast? YOU SELL THEM SLOWER!
So what is to keep them from turning this in to a subscription model. They they could tell you that the L1 cache on your processor is something you can license for the next 12 months. Once it expire, the feature goes away unless you pay to re-up your subscription. Sounds like a way for them to make money in a world where CPU power has plateaued and software does not push users into hardware upgrades every few years.
Edited 2010-09-21 02:18 UTC
An intriguing concept, but also incredibly confusing, and, dare I say it, misleading to customers, to whom buying a computer is already unpleasant enough an experience as it is.
The computer buying might actually become less confusing as You can only buy 1 model of processor and increase the processing power incrementally if needed.