Intel wants $50 to software unlock CPU features

[OvRiDe]

Doesn't AMD do the same thing when they sell you a Quad core but call it a "Black Edition" Tricore, that can be unlocked? Or even just a unlocked multiplier? They don't use a different dye for the Black Edition CPUs.

Its all marketing anyways. This is so that you can go to your local big box store and they can have an SKU for an upgrade. They will make great Christmas presents. I have a feeling we will seem stuff like "Free Upgrade (50 value)" next to a bunch of Desktop computer ads in the circulars.

The one thing I see here is all the people saying.. I am already buying a piece of hardware that has the capability but getting charged extra to be able to access all of it.

BUT.. what about all the people that are forced to pay extra for features they may never take advantage of anyways? Some people could argue that they will never need hyperthreading or the extra L3 cache, and not having to pay and extra 50 bucks is awesome for them.

[x88x]

Doesn't AMD do the same thing when they sell you a Quad core but call it a "Black Edition" Tricore, that can be unlocked? Or even just a unlocked multiplier? They don't use a different dye for the Black Edition CPUs.

My take on that is that if you use the unlocked multipliers to overclock it, you're taking it beyond the manufacturer's guaranteed performance, whereas with the Intel thing they have to guarantee the "unlocked' features in every chip regardless of whether the end user chooses to buy the "upgrade" or not. ...though you do have a point about the Black Editions being more expensive for no really good reason, but then Intel does the same thing with they Extreme Edition CPUs, so nobody has the higher moral ground on that one.

BUT.. what about all the people that are forced to pay extra for features they may never take advantage of anyways? Some people could argue that they will never need hyperthreading or the extra L3 cache, and not having to pay and extra 50 bucks is awesome for them.

Then wouldn't it be cheaper for Intel to make a cheaper chip that meets these needs? I mean, this thing is basically gonna be a slightly lower clocked i3-530 (an i3-520, say) since they have to certify it to perform at the max settings. ..unless they up the clock a bit in the G9651 to 2.93GHz, in which case it will be an i3-530. I suppose it's a kinda like what AMD ended up doing with a lot of the later 3-core CPUs, where they just branded a 4-core as a 3-core and locked one core even though it worked fine...you have to find the balance between the lower cost of just printing more of the same chip, getting less profit per unit selling it as a cheaper CPU, but likely selling more units than you would have if you had kept it as the more expensive CPU that it really is.

[OvRiDe]

Then wouldn't it be cheaper for Intel to make a cheaper chip that meets these needs?

Actually probably not. The cost savings in not have to retool for a different dye would most likely make it much cheaper in the long run.

But there is also another thing that most don't consider. Businesses. I am not talking about your average small business. For instance my company has 15K+ desktops across the enterprise. My locations have roughly 1700 desktops. If all my desktops were exactly the same, I would have the ability to have a consistent image for all machines, and software deployments would be more consistent. The problem is that not all of my users are the same. Where as the majority of them do not need the extra performance, there are several that could use it, or require it. An extra 50 bucks later they are still on the same desktop as all the others, and cruising along with the added performance. The same could be used for promotions or shifts in job functions. If someone has a change in roll and will need the added performance, normally the company would need to buy a complete new desktop unit, not to mention the installation and switch over time that would be needed. Instead the upgrade could be deployed along with any other software changes they may need for that roll to their existing machine, meaning much less time spent making the changes. Even on a more global situation, say one of clients deploys a new tool that will need the added resource. We have around 1550 machines used by the agents. $50 x 1550 is 77.5K dollars. New desktops at say $400 a piece would be $620K. Not to mention all the man hours required to receive, stage, image, migrate, and deploy that many workstations. I don't know if any of you have done this but .. it takes quite a while to just unbox 1000 PC's.

I would consider us a medium to large company. Now take companies like EDS or American Airlines where they run COE's (Consistent Office Environments) and order 100's of thousands of PC's and warehouse them for deployment over 3-5 years. The benefits to them would be exponential compared to the company I work for.

We as enthusiasts tend to look and apply everything to our applications. There are plenty of reasons behind these kinds of moves that make a whole lot more sense from other aspects. Intel has been around for a while, and it wasn't because they made a slew of bad decisions. Some of them may just be outside our realm of experience.

[x88x]

Hmm, I suppose you make a good point about the enterprise environment...though it still irks me for plenty of reasons that I won't rehash.

Intel has been around for a while, and it wasn't because they made a slew of bad decisions.

I never said it was a bad business decision. I'm sure they'll makes a ton of money off it. I just think it's a crappy thing to do to consumers.


Hmmm, this brings a thought to mind... In the novel Snow Crash, by Neil Stephenson, computers have advanced to the point where a simple wrist computer (think PDA/phone/etc) has orders of magnitude more computing power than most anyone will ever need. ..given such a situation, if the manufacture of these ridiculously powerful computers got to a certain cost and reliability point, I could see companies doing their entire line using a single identical piece of silicon with different parts disabled by firm/software. It'd be like if, say AMD, got their manufacturing down to the point where it cost the same to manufacture a 6-core CPU as it did to manufacture a 2-core (I think already there, tbh), and the success rate reached 100% (not anywhere near there yet...though they're doing better than Intel, iirc, at least on their high-end chips), then they could make their entire line by just making craptons of 6-core CPUs and disabling however many cores wouldn't be used in a given line. Heck, nVidia is kinda doing that now with the GTX400 series. IIRC, there's a single die for "Fermi", and they just separate them out into the different models depending on how many stream processor nodes failed. Don't get me wrong, if a company did what I suggest, it would massively piss me off...until a method of unlocking all the features was leaked/RE'd, in which case I would just buy the $50 cheapo CPU and turn it into the $500 awesome one... .. Locking out portions of the chip for power-saving reasons or the like, I would go along with, as long as I could unlock those portions at no additional cost...I just really don't like this almost 'hardware as a service' approach..I bought the CPU, I should be able to use the whole thing without (essentially) buying it again.

[OvRiDe]

That does brings up an interesting concept. What if Intel decided to just produce a single dye that has say 12 cores. But they basically have feature sets that are unlockable by in place upgrades. So you start out with a basic CPU. For this example say the dual core 2.8Ghz non-hyperthread version is 100 bucks. Your on a limited budget so you start out with this. Later if you would like to go with say a quad core, instead of having to spend 175 bucks on a new CPU and then have to deal with either eating or selling your current CPU, you could just pay a fee of 50 bucks for an upgrade. Later as time goes on .. you have the money to go up to 6 cores for say another 50 bucks, and so on. What if they did the same thing with multipliers and bus speeds.. so for x amount of dollars that 2.8 could become a 3.2 or 4Ghz? You could slowly step up your performance over time with out having to deal with the depreciation of your current cpu on the used market, and eventually end up with the top of the line performance. To be honest this sounds really appealing to me. It seems like its a lot easier to free up fifty to hundred bucks every month or so, then to go through the hassle of buying a whole new CPU if you want to upgrade, AND deal with your current CPU. Also .. you wouldn't need to dismantle your PC to upgrade your CPU or better yet your water loop!

I dunno this could be the start of something really cool.

EDIT..

Don't get me wrong, if a company did what I suggest, it would massively piss me off...

I believe AMD has been doing similar practices since the days of their 486 and 5x86 chips, along with probably every other chip manufacturer as well. Most of the time they used physical means. I believe it was the early Athlon XP chips that used diodes on the top of the chips for multipliers. You could use pencil lead to complete the traces and change/unlock the multipliers. Its been a while so I may be fuzzy on the details, but that was the jist of it.

[SXRguyinMA]

very interesting points, I like where this is going

You did shed some light on the other side of things OvRiDe, I was just thinking of most people here and our applications, as you mentioned

[Oneslowz28]

And now like all other great ideas posted in a modding forum, this will be turned into a reality. The question is.. who will steal it and call it their own.

[FuzzyPlushroom]

So, I buy a new laptop and get a free 'upgrade' (read: unlocking what's already there) as a promotion. Fine - no initial change in my buying experience, really. However...

I see three possible ways of doing this. The most obvious would be including a tiny amount of flash memory inside the CPU die itself, so that once it's unlocked, it's unlocked. I know that heat and size constraints prohibit that option, though. I figure that it's actually a BIOS update, based on the description of and instructions for the tool itself on Intel's website ("After downloading, copy onto a bootable DOS USB drive"); as such, it shouldn't be that difficult to hack. The third option, of course, would involve a specific driver or some such under a operating system, or a tiny partition on the hard drive - and I imagine some folks here remember how well storing the BIOS on the hard drive worked for, say, Packard Bell back in the '90s. There's a reason they're out of business here, but I digress.

The relevant questions are many. What will this CPU appear as in an ordinary motherboard (something other than an Intel DH55*)? (After all, if someone pays for an upgrade and has their motherboard replaced, even with another identical board, the upgrade will no longer exist.) Is this upgrade code stored on a separate EEPROM chip, so that a corrupted/outdated BIOS can be reflashed without losing the upgrade? If not, can the upgrade code be used more than once, and how would this be enforced?

It's a good idea, in theory, for very select markets (OvRiDe's big-business example is the only one I can think of), but I suspect that the implementation is inherently flawed.

[x88x]

Looking at the idea of having one die for everything, after thinking it through more, I think as long as they couldn't sue people who managed to unlock them without paying them (I'm not sure if the recent 'jailbreaking' thing would affect that or not), I wouldn't have a real problem with it..heck, it would mean I could get a top-of-the-line CPU for the cost of an entry-level CPU, and be guaranteed that it would work properly (unlike unlocking the fourth core in a 3-core AMD chip). Obviously the warranty would be toast, but since when has that stopped us?

The relevant questions are many. What will this CPU appear as in an ordinary motherboard (something other than an Intel DH55*)? (After all, if someone pays for an upgrade and has their motherboard replaced, even with another identical board, the upgrade will no longer exist.) Is this upgrade code stored on a separate EEPROM chip, so that a corrupted/outdated BIOS can be reflashed without losing the upgrade? If not, can the upgrade code be used more than once, and how would this be enforced?

This is a very good point, and one I hadn't thought of...would it only be unlockable on certain chipsets (which would invariably only be Intel chipsets, at least at first)?

[mDust]

I'm thinking that the record of what chip you unlocked and what parts you unlocked would be stored in a digital 'locker' similar to what Steam and Microsoft do with games/software. It wouldn't have to be much different than flashing your BIOS. They could lock and unlock it mechanically instead of digitally. Just send the right commands to the chip and switches close. By making the switches internal or spraying some sort of lacquer over them, they can prevent users like x88x from getting his free upgrades... Don't worry x88x, I'm only saying this so when the bigwigs at Intel read my post they'll think it's safe to release these chips. We'll hack them later.