Intel Puts the Lock on Overclocking

Patrick Schmid writes "Intel included an overclocking-prevention mechanism into the 915/925 chipsets. So far, only Asus and Gigabyte know how to override it. You can start from the beginning or jump to where we discuss the overclocking lockout."

Comments

Market Share

[drsmack1]

Do these sorts of things hurt their business? I wonder how many would not buy into Intle knowing that there are these sorts of things built in? I imagine that the big corps don't care.

Re:Market Share

[Supp0rtLinux]

Remember... all the big OEMs are pushing the high-end gaming systems to the "new breed of computer users". Just as there's a resurgence in IT spending, there's a new breed of home PC user that doesn't mind paying $3000+ for a high-end system. No, corporate America doesn't give a rip about overclocking cause they care about their warranties. But there's a growing group that does... and choices like this make them turn to AMD. AMD has better, cheaper 64bit support. And they still make overclocking easier. Intel is cutting off their nose to spite their face. Funny... you'd think Intel would learn from Microsoft's example instead of repeating the same mistakes...

Re:Market Share

[lukewarmfusion]

No, these help their business. The percentage of people that overclock their machines is probably very small. However, the people that are willing to spend money upgrading their hardware are much more common. So Intel capitalizes on the fact that more people are willing to spend money on upgrades than to overclock.

Very few people will "not buy into" Intel because there are mechanisms like this in place... because most people don't know that overclocking exists, and many of the people who do aren't going to overclock anyway. No loss for Intel.

Re:Market Share

[NanoGator]

"I wonder how many would not buy into Intle knowing that there are these sorts of things built in?"

I find it unlikely Dell customers are buying machines thinking about overclocking.

Re:Market Share

[WormholeFiend]

I understand your point, but if that is the case, why put any overclocking mechanism in the first place?

If the people who don't care about it don't overclock and buy upgrades, is Intel trying to force the small minority of OCers to buy upgrades as well, instead of squeezing in extra speed/mileage from older chips?

Re:Market Share

[amRadioHed]

No, these help their business. The percentage of people that overclock their machines is probably very small. However, the people that are willing to spend money upgrading their hardware are much more common. So Intel capitalizes on the fact that more people are willing to spend money on upgrades than to overclock.

I can't imagine this will do anything but hurt them. The people who were going to spend the money to get a faster intel chip before are still going to. The people who were going to buy an intel chip and overclock it are now just going to buy an AMD. How is this possibly going to get them any new sales?

Very few people will "not buy into" Intel because there are mechanisms like this in place... because most people don't know that overclocking exists, and many of the people who do aren't going to overclock anyway. No loss for Intel.

Sure, very few people will "not buy into" Intel because of this, but ZERO people will buy Intel just because of this. few > ZERO, thus it's a net loss for Intel.

Re:Market Share

[angle_slam]

I agree. Back when you could overclock a Celeron nearly 50% (from 400 MHz to about 600MHz), there was big benefits from OCing. Now, who really cares if you can get your 2.8 GHz processor to 3.2 GHz. This generation's games don't even push the 2.8 GHz processor that much anyway.

Re:Market Share

[p0rnking]

AMD might be the one that people are turning to, but remember, they have locked their chips in the past (ie: with their durons, you had to use a pencil to connect the bridges on the chip to unlock it).
As for it hurting Intel? I doubt it. I think overall sales would go up a bit.
Personally, if I can overclock my chip by a few hundred Mhz, then thats a few hundred Mhz that I will skip out on buying, until a chip comes out that mine can't compare to.
Besides, a majority of the people who own home based Intel systems, don't even have a clue what "OC" stands for ... the overclockers, and "hardcore" geeks make up a small percentage of overall sales.

Re:Market Share

[morcego]

If the difference between a low/mid range and high end product is just a switch, or a simple transistor, they shouldn't ask extra money for it, because the features in the low end are only "hidden".

I used to think like you do. Unfortunately, there are other issues to consider, besides the production costs. Development costs (the cost of the technology itself) plays a big part on pricing. The fact that the low end product is the high end one with some features disabled (or even burned out) is a move to reduce the production costs, without stopping the offer of the low end model.

So, if they were to remove the price gap, you can rest assured it would be based on the price of the top model, once you pay not only for the production costs, but the features you get.

This is pretty much what you get on the software or consulting industries. Sometime you have to pay extra to have a few features of a given software enabled. Sometimes you will have to pay different price rates for a professional's time, based on the kind of work he will be doing.

Re:Market Share

[bentcd]

That may be part of it, but the main reasons are price elasticity and development cost. There are two different market segments out there, and the boys and girls at Intel know it. Their problem is how to cater to both segments - thereby making more money - while avoiding the astronomical development costs that would be involved in having two completely different products. Their answer is to sell one "crippled" line of the product to the low-cost market with a price designed primarily to cover unit production cost (with a markup) and sell the fully functional version to the high-cost market with a price primarily designed to cover the development cost, also with a markup of course.

You see the exact same thing in the software industry. If I buy JBuilder X from Borland, I can basically choose between the low-cost version at $500 or thereabouts, or I can buy the high-cost version at $3000-ish. The different versions are carefully designed so that high-end customers will really _want_ the expensive version because its features cater to high-end needs such as development of scalable server-side web applications while the low-end one contains features for general application development. There is nothing in the production cost that dicates the price difference - it's just a set of CDs either way and if there are a couple of extra CDs in the high-cost version, that hardly costs them $2500 per unit to manufacture. In stead, they create an artificial product divide in order to cater to different market segments, using one of them to help cover the development costs associated with the more pricey parts of the product.

This is going to be an active problem in any industry or in any phase of a business where the development costs completely overshadow the unit production cost. Until such time as the industry or business has managed to cover its development costs, you will see units that are massively overpriced when compared to actual production cost. This creates a dilemma for the manufacturer, because while he might technically make a profit from selling a unit at, say, $100, this is not going to be enough to cover the huge debt he accrued from development by the time the investors will want their money back. The solution to the problem is to create one market that will buy the unit at $1000 and another that can have it for $100. The $1000 market covers your development costs and the $100 market is just gravy. The way to maintain the high-cost market is to do something or other to cripple the low-cost market product so that the high-cost market won't want it.

So now I can't burn out my processor?

[Trigun]

Sounds fair. Now if only Intel would agree to not abuse this by artificially locking processors to manipulate market prices.

I can hope.

Anything to make a buck...

[Supp0rtLinux]

Sure... wouldn't want to be able to overclock easily... gotta make us upgrade to the next, best, faster CPU. Maybe if AMD stops gaining market share from Intel, Intel will lighten up on those of us that want to overclock.

Re:Anything to make a buck...

[BenjyD]

Would overclocking really hurt Intel's sales of high-end chips much? The kind of person who is going to overclock their processor is doing it because they can't afford the faster processor, or just because they can and want to be l337. Either way, that's not a lost sale.

The only reason I can see for preventing overclocking is to stop resellers selling systems with overclocked chips in to cut costs. You could prevent that by having the BIOS display a big warning message saying "This system is overclocked and may be unstable - phone Intel now" to alert normal users.

Re:Anything to make a buck...

[BabyDave]

Or maybe when AMD continue to gain market share, Intel will stop treating overclockers (many of whom are buying the top-of-the-range CPUs, and pushing them further than Intel do) like thieves ...

Over Clocking is over Rated

[Captain BooBoo]

I have been over clocking since the 486 SX and I can say its all just a fun game. You will NEVER have the stability and it just does not make sense to do any more. If you can't aford to buy a chip fast enough to do whatever job you need it to do then you need to rethink what you're doing. Granted it was "fun" and "neat" to put one over on the chip makers but in the end its all just meh.

Re:Over Clocking is over Rated

[swv3752]

Especially now when the real world performance gain is just not there. When overclocking a 300mhz chip to 500mhz gave significant performance boost. Now that we have Ghz+ chips and squeezing another 100mhz gives a few percentage point points more performance. So now it will take 1 second less time to encode an mp3 or mpeg? I know have to take superhuman steps to cool the processor. I'll stick with stock speeds and have a quieter PC to boot thanks.

Re:Over Clocking is over Rated

[Billobob]

Thats just flat out wrong. My friend purchased one of the athlon XP-2500 mobiles for ~$90 a couple months ago (stock at 1.83 ghz), and managed to overclock it to 2.6ghz with the same stability as the processor is stock, and nearly 800mhz is nothing to scoff at - it managed to be the small advantage he needed to run UT2K4 in 1600x1200 smoothly, and improved multitasking slightly. If its just for home use, then overclocking is a fun crapshoot - sometimes its a bust, but often you can gain a significant improvement out of your processor. Of course nobody would overclock any system that requires total stability and 24/7 uptime, but for your typical home PC a little experimenting couldn't hurt as long as you know what you are doing.

Re:Over Clocking is over Rated

[hawkbug]

Yes, I also have the 2500+ and it's amazing that it can go that high - if you get the mobile version, and you got a good one. I don't have the mobile, so I haven't tried to mess with it, but even at 1.83, it runs very cool for me - and having a quiet pc is more important than fast for me these days.

Re:Over Clocking is over Rated

[kju]

You must be a fool to do science on a system you are running outside of specification. It is a fact, that overclocking can cause miscalculations etc. which might not always be detected.

Re:Over Clocking is over Rated

[swv3752]

Sure there are some CPUs that you can push far. But you have to have a bit of luck in picking out the right one. But for many, you have to mess with core voltages and higher heat for a lousy 100-200mhz.

For real world performance, faster memory and hard drives and video will give a better price/performance ratio.

I tried and tried...

[ErichTheWebGuy]

but can find no practical reason for this that makes sense to me. The people who overclock know that they can burn up their chip, and the people who do not overclock don't have to worry about it. I guess maybe a small percentage of people might go poking around in CMOS setup and change the clock speed, but is that number large enough to alienate gamers and hackers who want control over their own boxes? I think not.

*grabs ankles* Thanks again, Intel. Gimme on-board DRM and I will be a happy camper.

Re:I tried and tried...

[ErichTheWebGuy]

It's to crack down on "grey market" resellers

I realize that, but is this the best way to go about it? IMHO, this only irritates users, who then run crying to AMD. There should be a better way, like revoking whatever reseller license Intel gives to these people (if they even give such a license, I don't know), or displaying something on the display at post "cpu overclocked!!!" or something. That way the bad-guy resellers won't last for very long, as they are now exposed for ripping people off.

It still makes absolutely no sense to me that Intel would punish end-users for the actions of a shady reseller.

How Long?

[BlindSpy]

How long before every semi-hacker knows how to unlock their intel chip to over clock it. When has locking anything kept anyone out?

as if...

[colinleroy]

...as if it was still useful at speeds above 3GHz.

It's official; Netcraft confirms; Intel is dying

[W2k]

Multiplier locks on new chipsets - in effect, new CPU's? AMD's Athlon FX is completely unlocked. How is Intel going to compete by continuing to offer an inferior product?

Agreed

[mfh]

Why Intel bothered to lock these chips from overclocking doesn't make much sense to me. Maybe they want to ensure that users are getting what they pay for, and not more, but if it can be unlocked easily enough, I don't see why Intel would bother. It doesn't seem like best practice is being utilized in this kind of prohibitive design mechanism.

Re:Agreed

[Anonymous Coward]

However, if this is the *real* reason, then surely this is met by one of the following ideas (there are probably plenty more):

1) Report the *attributed* speed in the BIOS bootup (you can then see what the rating is even if it is overclocked)
2) Flag something saying that the CPU is overspecced

These allow users to overclock, but stops retailers cheating.

Re:Agreed

[Anonymous Cowpat]

Maybe they want to ensure that users are getting what they pay for, and not more

Ummm, I was under the impression that they were paying for a whole pile of transistors and some copper (at lowest detail). Surely pumping more current/voltage through said bunch of transistors is still only getting what you paid for, you're just using it to it's full potential.
I paid for a bunch of transistors which, when used optimally, will run at speed X, whether or not speed X is what Intel tells me that it will do.
Now, if I was activating a transporter device which stole additional transistors from Intel HQ and put them onto the chip, that would be getting more than I paid for.
If it was there when I bought the thing then I paid for it, whether or not you told ME that it was there, YOU knew that you were selling me the extra capacity.

First Dual-Core Hotplates, then locking?

[LordKazan]

First It comes out that Intel's making Dual Core Prescotts what would do better as hotplates than processors, and now they're announcing that they're preventing you from overclocking?

Will someone PLEASE remind me of Why I would ever pay $499 for a Pentium 4 3.4Ghz Prescot, or $990 for an 800Mhz 2MB Extreme? I can hop over to AMD and get a better processor for less, and to boot I can overclock it if I want!

Intel = Morons

i love these articles...

[moondo]

Intel included an overclocking-prevention mechanism into their new chipsets... only Asus and Gigabyte know how to override it.

There you go! As long as Intel doesn't make an "unoverridable" chipset we'll have crazy geekz trying to figure out how to get around it and making a webpage about it.

Intel cutting its own limbs off

[Spl0it]

I can't believe they would do something like this. As far as I've seen AMD has been stealing market share faster and faster every day and by doing this they're only encouranging every gamer who still has intel to purchase an AMD system. Everyone knows that the gaming industry is driving computer hardware and software to new levels; For intel to be cutting it's ties with the gaming industry is like shooting yourself in the foot.

This has nothing to do with warrenty issues, and everything to do with Intel wanting people to continuely upgrade for the faster chips right when they come out. (When prices' are high).

How does it work?

[maxwell demon]

Yes, I've RTFA, but to me the image looks as if it were comparing with the PCIEX clock, but the text explicitly states that PCIEX isn't doing the trick. Maybe it's just because I don't know what BSEL[2:0] means, but I don't understand the mechanism.

Especially: Since the only reference for a clock can be another clock, shouldn't it always be possible to simply increase all clocks available to the processor (assuming the rest of the hardware doesn't make problems)?

Support

[rsilvergun]

The problem with overclocking is support. That "new breed of computer users" are idiots. Now that overclocking can be done in the software on most computers, it's accessible to people who don't know better than to kick the speed up 2x. Remember people who would run the Add New Hardware Wizard in Windows 95 to add stuff they didn't have? These are the people Intel is afraid of (and rightly so).

Yeah, they're not supported, and they don't get warranty support. But that's not gonna stop them from getting mad at Dell/Gateway/etc when they won't replace the CPU they fried (remember, these people are dumb). Their attitude is likely to be: "if it was going to break the computer, why'd you let me do it?".

Also, Intel's interested in making installing a new processor as easy as possible, which means idiot proofing the things. If you can't overclock it, that's one less way to fry it. Again, fewer support calls, fewer stupidly angry customers.

Re:Support

[Cecil]

Now that overclocking can be done in the software on most computers, it's accessible to people who don't know better than to kick the speed up 2x ... These are the people Intel is afraid of (and rightly so).

A nice theory, but if this were the case, there would be no reason for the stepping B chip to remove the ability to disable the 'feature'. The fact is, Intel would much prefer power users to buy a very-high-margin Xeon or Extreme Edition or whatever new multi-core CPUs they have coming up (all of which they currently avoid, because the hefty cache and other features make them very poor for overclocking). They decidedly do not want them taking the latest bargain-basement equivalent to the Celeron 300A and overclocking the shit out of it.

At worst, a conspicuous bridge that needs to be soldered on the CPU like AMD used to do is more than obscure enough to keep out the people who don't know what the hell they're doing. It should not require trickery from motherboard manufacturers to work around actual electronics on the chip. The last thing I want is to have to my CPU and motherboard and other components engaged in electronic warfare with one another.

I'm really really tired of people trying to protect the 'stupid people'. Let natural selection eat these morons (or at least their money). Please. For the good of the human race.

1% is hardly "cutting off their nose"

[msobkow]

I'm quite certain that even a 1% estimate exagerrates the number of overclockers rather significantly. Sure they spend more money on their systems than regular users, but not on the CPU -- it's going to the latest bleeding edge 3D accelerators, the fastest CAS2 memory they can find, the fastest HDD's they can find, etc.

If most overclockers were busy overclocking the fastest CPU's Intel sells, maybe it would make a difference, but most overclockers are trying to get that bleeding-edge performance without paying the bleeding-edge price. Intel loses nothing by stopping the practice.

We're also getting well into the hardware performance ranges where overclocking by even 10% is a major accomplishment that requires very serious cooling. It's not like the PII/III days when you could get as much as a 50% boost over the rated speed (rare, but it did happen.)

Even most overclocking fanatics I've known over the years don't bother overclocking their latest systems. It's not worth the risk of frying the CPU and destabilizing the system for less than a 10% performance boost when you can go with a dualie board of cheaper CPUs instead.

Re:1% is hardly "cutting off their nose"

[mprinkey]

It's not like the PII/III days when you could get as much as a 50% boost over the rated speed (rare, but it did happen.)

It wasn't at all rare. The Celeron 300a overclocked to 450 MHz without so much as a whimper. I built several systems (single and dual) using them. I had dual 366 celerons that ran at 550 in SMP. That was the heyday of overclocking and it was created by Intel's marketing practices. With competition from AMD, Intel hasn't been able to "downbin" and terrace the market to its whim anymore, so overclocking isn't that big of a deal.

Re:1% is hardly "cutting off their nose"

[mcrbids]

I'm quite certain that even a 1% estimate exagerrates the number of overclockers rather significantly.

But it's a very vocal 1% of users. The 1% that many others look to to discover what's "best". How many people bought athlon CPUs back in the day because it was very overclockable, because of the implied quality?

Witness Tom's hardware and a million other sites, which have long been a big proponent of overclocking...

Re:AMD fanboys everywhere

[Anonymous Coward]

You obviously have not put together an Intel P4 lately!

The heatsink that they ship with the P4 OEM kit is humongous. The OEM fan included with my son's AMD 3000+ was much smaller. Fanboys indeed!

To Intel's credit, I recently built 7 P4's (building a cluster, got a deal on P4's; no vendor bias here) and every CPU idles at ~45C. These are 2.4G, 400 M FSB with 512k cache.

Overclocking is NOT overrated. It's got value.

[Anonymous Coward]

I built a system a couple years ago with an ASUS P4T-E mobo (dead-end rambus memory and 100MHz core FSB speed technology). Yes it was once a top performer back then when a 1.6GHz Northwood P4 was first released (which BTW, oc'ed nicely to 2.13GHz by cranking the core FSB speed up to 133MHz, long before the 133x4=533 FSB speeds officially arrrived). I got 2+ good years of service from that machine, rock-solid stable. I wanted to upgrade, but thanks to the current economy, I simply didn't have the $$$ to buy all new innards for this machine. The last of the 2.4GHz (400FSB) P4 chips were being sold pretty cheaply but need a pretty healthy cooler to overclock them stable to 133 core FSB speed which gets you 3.2GHz total CPU speed, so thanks to NewEgg and 2Cooltek for about $200 total upgrade cost I now have a 3.2GHz machine which is pretty close to the new contemporary gaming machine expected cpu speed. BTW, the old PC-800 rambus memory must be clocked with the 3X multiplier instead of the 4X multiplier when using a core 133MHz speed in this mobo... but still with a 399MHz memory clock, the memory thruput performance in this machine is about equal to PC3200 DDR memory running in single channel mode, which ain't too shabby. I run this machine 24x7 too, it's doing Stanford protein folding when I'm not gaming on it, so the CPU never gets a rest. Once I got my SP-94 cooler and case fans set up correctly, I've not had one single crash due to heat or OC instability since I installed the upgrade components about six weeks ago. I'll be able to squeeze another year's useful service life at modern speeds from this pile of parts, and hopefully by then, I'll be able to afford to build a new box around a 64-bit Athlon... the new king daddy paw-paw processor.