Bypassing Intel's Overclock Limit Reveals DDR2-667

BatonRogue writes "Slashdot posted a Tomshardware article talking about Intel's 10% overclock limit on their new chipsets not too long ago. The situation has just become even more interesting. AnandTech just posted a roundup of DDR2 memory that sheds some light as to why Intel may have implemented the lock. It seems that on the Abit board they tested, which supposedly bypasses the overclock limit, the first generation of DDR2-533 memory modules had no problems working at 667MHz. Could it be that Intel is keeping DDR2-667 support for yet another revision of their new chipsets even though the memory support is clearly here today?"

Comments

Dancing with stability...

[tepples]

Will DDR2 memory let me hit twice as many arrows at the same BPM?

Wasn't there an AMD related article last week?

[XMichael]

I seem to recall last week almost the exact same posting, however it was using AMD's name rather than Intel.

Anyone got any intel on that? (-;

Mike

A long-running conspiracy

[salemnic]

You know, I think we've all heard rumours of an intel conspiracy to make us buy the same product again and again for years now...

However, one must at least consider that they have a valid reason for this. Long-term stability maybe?

Oooohh... is that machine stability, or cashflow stability....

-s

Re:A long-running conspiracy

[AKAImBatman]

That was hardly a deception. Intel sold chips with damaged Co-Processors as 486SXs. This allowed them to increase their chip yields, and make money on something that would have otherwise been garbage. That's not to say that some SXs didn't have working coprocessors that had been disabled. In those cases, Intel simply remarked DX chips to meet the demand for the SX line. (How many damaged chips can you manufacture?)

Re:A long-running conspiracy

[Short Circuit]

There's more to the competition than just out-performing your competitor for this cycle.

You have to have your next product ready for your anounced (or even widely rumored) release date. It's a lot easier to resell the same overperforming device at lower performance levels than to keep producing new overperforming devices at the same rate.

Basically, you take bigger, slower steps, but you only report a frequently increasing fraction of each completed step in order to keep a comparable product out there competing with your competitors'.

Also, having a single product for a longer period of time allows you to tune the hell of that product while R&D is working on pumping out the next generation. So by the end of your cycle, you'll have a reputation for a stable product.

Finally, having a single design with a long lifespan gives you insurance in case there are delays in the R&D while they're trying to produce the next generation. You don't want your competitor to come out with two or three iterations of their product while you've only got a single relatively lackluster device on the market. (This has happened to Intel in the past...look at some of the old benchmarks on Ars Technica.)

Re:A long-running conspiracy

[ePhil_One]

in essence, you paid for two chip and were only using one. Yes it was deceptive. The users beleived they had a cpu and a seperate expensive add-on doing the math.

Let me see. Users paid the price for two chip solution that would actually be slower than the single chip solution. Users actually got the faster single chip solution. Instead of forcing users to pay full price for a 486DX, they offered a mislabeled 486DX that they sold to users who already had a 486SX at a large discount.

Yes, lets curse those bastards! Never buy anything from a vendor that gives you more than you paid for!

Yo know, I heard a rumor that Intel is still up to these tricks! Appearantly, all P4 Northwood CPU's are identical, cut from the same slice of Silicon even, burned from the same mask! and they have the gall to sell "3.2" Ghz parts for hundreds more that "2.4" Ghz parts.

Cripes people! It costs about 50cents to actually produce the chips, billions to design the damned things, and this causes some weirdness in the marketing of them. AMD does this as well, the test only enough parts to meet the "demand" for the high speed parts, everything else gets tested at the next lowest speed. There are hundreds of "Overclocking" sites that point this out all over the web. Are you going to accuse Intel of ripping you off because that 2.4Ghz chip could have run at 3.2Ghz? You paid for a level of performance, you got that level of performance.

Well...

[kemapa]

Could it be that Intel is keeping DDR2-667 support for yet another revision of their new chipsets even though the memory support is clearly here today?

I'm suprised one must even ask this question, because the answer is absolutely Intel would do something like this. Keep in mind that the benefit of locking overclocking to only 10% is twofold for Intel. Not only do you disguise the now revealed fact that DDR2 is ready for faster speeds, but you stop people from buying cheaper chips and overclocking them. At least, that's what intel intended before the mobo manufactureres bypassed the locking. Think about it in this theoretical situation, why should I buy the $1000 3.4Ghz chip when I can buy the $500 3.2Ghz chip and overclock it 200Mhz?

Re:Well...

[Dracolytch]

Simple, because I want a machine that is reliable, that I don't have to dick around wi... Did you say $500?

~D

Re:Well...

[foxtrot]

you stop people from buying cheaper chips and overclocking them

Of course, you and I, the enthusiasts, aren't the people Intel is worried about when they turn off overclocking. (After all, we've already moved to AMD, right?)

The problem is grey-market processors. If the processors (or RAM) are easily overclockable, then Tiawancorp Computers may buy 3 gigafoo parts, overclock them to 4 gigafoo, and sell them in computers "with 4 gigafoo parts!" to unsuspecting consumers.

The hazard here isn't just that Intel loses money-- after all, for every three people you know who overclock processors and have rock stable systems, there's always that guy who can't get the darned thing through much past a boot before the thing locks up. Intel systems become known as "unreliable", which is definitely not a position Intel wants to be in.

Losing money _and_ losing your reputation is a heck of a double-whammy. I can't say I blame them for wanting to stop overclocking.

I just can't imagine why they'd do it in a mechanism that could be defeated by the motherboard manufacturers....

-JDF

Re:Well...

[eggoeater]

This reminds me of a story an older IT guy told me years ago. He was in charge of several mini-computers in the 70s. The company who made the computer told him they could upgrade it when he needed more capacity. When he ordered the upgrade, the computer company sent a tech out to do it. The tech opened up a panel on the computer, cut one wire with a pair of clippers and closed the panel. I don't think they even had to reboot it.

Re:Well...

[Jennifer E. Elaan]

It used to matter a lot more back when you could really overclock a processor. My old dual-celeron used to run decently at 550MHz, when it was originally specced at 366. I think ion migration finally started to set in, though, it doesn't take the higher speed properly anymore. Otherwise, it's still running (at 366) to this day.

I'm one of the original overclockers. I had a 486DX36, back when overclocking required replacing the clock module on the motherboard. I'm also an occasional chip designer.

The trick to overclocking is to know what your limit is. Until recently, thermal load was not the limiting factor. The real limiting factor was a condition called "metastability", where a digital transition fails to finish before being latched in the next register (usually due to violations of the setup and hold time restrictions of those registers). The smallest case of metastability can flip a bit.

A larger case can cascade through multiple stages, flipping lots of bits or even pushing the state tables into illegal states. This is where the first real danger lies: a processor that uses one-hot encoding to improve the speed of the controlling state machine can be pushed into illegal states that may cause several circuits to drive the internal busses at once. This leads to large current dissipation, and in some cases it can burn holes in the thin metal layers of the IC.

A less common hazard appears in cases when the CPU is massively overclocked. The CPU in such a case will never exit the metastable state. This causes each clocked circuit on large areas of the chip to dissipate maximum current during those metastable states. This can also lead to high current dissipation, although it is less dangerous than the abovementioned one. It's also worth mentioning that a chip in this state will not function normally.

Generally speaking, you have a wide margin between the onset of metastability and the onset of serious damage. Unfortunately, there is now *another* danger of overclocking processors. High thermal load can cause ion migration. In fact, most processors are now designed to only last 5 years (!) before ion migration renders them useless. (This is also why I personally don't overclock anymore).

It's frightening to notice that mainstream CPU's are less and less overclockable and have higher and higher thermal loads with smaller and smaller featuresizes, though. The manufacturers are simply not leaving as much margin as they once did.

A company trying to manage its products

[jj_johny]

To imagine that a company would manage the release of products. Its as if they had a plan. How could they keep new things from us that they have already finished? My goodness what is the world coming to? The next thing you will tell me is that they have products planned that go even faster.

Flashbacks of a failed mechanic

[Havokmon]

The 8 DDR2 memories that we received were all rated at or near the Jedec standard for PC2-4300 of 4-4-4-12 timings. Jedec standards for PC2-5300 (DDR2 667) call for 5-5-5 timings, but our Corsair DDR2 667 was rated at the faster 4-4-4, at 667, which already shows improving timings for DDR2.

Then I suddenly remembered the warm-buzzy feeling I got the first/last time I tried to check my timing belt in my '80 Capri.

Demon Dance

[Chagatai]

Well, it is DDR2-667, which means that you will be dancing on twice as many arrows with the Neighbor of the Beast. In short, nothing but songs by Aqua.

It's all about the Benjamins

[Dark Kenshin]

Could it be that Intel is keeping DDR2-667 support for yet another revision of their new chipsets even though the memory support is clearly here today?"

Wow, a company may be holding back on technology for a future money making opportunity? This has never happened before!

Dont like Manufact. control but Love Stabilty

[Linus Sixpack]

While I really dont like companies crippling the hardware I buy. Could you imagine a shovel that would stop you from moving sand too quickly ?

I have to say that if my systems are %10 slower but even %05 more reliable its a good trade off. I'm not saying that they are super unreliable now, given all that goes into making a computer behave and the number of suppliers involved. Its just that some of these machines are so fast that I there is no value to me personally in overclocking.

I think that businesses generally have enough to worry about with proper operation to not want to overclock their processors. If Intel is working with memory makers to insure stability they should be up front about it -- people would understand the overclocking then.

Overclocking adds another possible source of problems that most companies don't need.

Re:Dont like Manufact. control but Love Stabilty

[mopslik]

Could you imagine a shovel that would stop you from moving sand too quickly?

It's commonly referred to as a "pitchfork".

It isn't that complicated...

[Theovon]

1) A major reason to not support the higher speed is that chip production yields are increased.

2) Why not speed-grade the parts and sell two models? Not cost effective at this time.

3) Futhermore to release one model how and then another model later maintains a more even revenue stream than two models now and then none later.

4) Most likely, spreading it out also increases total revenue due to the people who buy one and then upgrade to the next.

Turn the question around

[mec]

Turn the question around. Instead of asking "why buy $1000 3.4 GHz chip instead of $500 3.2 GHz chip", ask this: how come Intel doesn't label that second chip "3.4 GHz" and sell it to you for $1000?

That would be $500 extra revenue for Intel. How come they don't do it?

Perhaps it's something to do with increased failure rates, warranty returns, and a negative hit on their reputation.

Intel grades their chips. They mark each chip with the speed that they feel comfortable selling with Intel's name and warranty.

If you want to overclock your chip, it's your chip; you bought it, you didn't license it with a stupid EULA! But the problem comes when another company (not Intel) buys a $500 chip from Intel, overclocks it, and then marks up the price and sells it in a system as if it were a $1000 chip. Intel gets nothing all the trouble, the cheater company gets the markup, and the end user gets the shaft.

Re:Turn the question around

[LqqkOut]

They sell it for $500 for the same reason that Canon is selling toned-down versions of the EOS 10-D

There's a huge market segment that purchases equipment that's one step below bleeding edge just for the price break. Could it be that there's more money to be made by soft locking a single product than to manufacture multiple products?

This sorta reminds me of the boon brought by the Celeron 300A

I've come to expect this from Intel...

[ameoba]

In Intel's long tradition of pushing a new technology before it actually surpassed the previous generation (P2, P4, P4-Prescott) DDR2 is, at best, on par with systems already in place.

If you look at some benchmarks of DDR2 performance, you have to wonder why anyone would even consider buying it right now.

"Expect DDR2 memory at 533MHz to be comparable to DDR1 at 400MHz, but don't expect to see any "noticeable" memory bandwidth performance gains till DDR2 667 and above with low timings!"

Marketing Decision

[NoSuchGuy]

That's a marketing decision.

Engineer: Our new chipset is capable of running DDR2-533 memory modules at 667MHz. We can jumpstart shipping...
Marketing: STOP! Right now there is no demand for such a thing. Let's wait till we are ready with a new promotion song from Christina Aguilera.
Engineer: But no one buys our chips because of the promo and the songs we have. Our customers buy because of the specifications. Right now we have...
Marketing: Who are you to tell ME what our customers want? Stick to your soldering gun and chips!
Engineer: But the Intel guys in their funny radioprotection overalls didn't accelerated sales....
Marketing: No buts! We start marketing this feature for christmas.
Engineer: But...
Marketing: SHUT UP, I'm the guy with the MBA! That's why I am here. I know what to do!
Engineer:...
Marketing:...

This is pathetic

[Anonymous Coward]

Intel reliability standards call for infant mortality less than 500 units per million and a service life of 7-10 years depending on product. The goals of overclockers are very different.

Ever heard of a schmoo? It a contour graph of what frequencies a part will work at at a given voltage and frequency (and that's only at one set temperature). Ofthen is is very blob like and can have holes of instability or islands of stability. You need to pick product offerings that not only yield well, but are far enough on any contour line from the schmoo that you can test the part at a few values of voltage and frequency with confidence. Tester time is a bottleneck, and Intel test more than anyone else.

Yes, downbinning occurs for marketing reasons, but generally for economy products. Believe me, Intel and every company fights for yields at the top bin for every flagship product. Even when downbinning occurs, it is done to match actual yield to orders, not out of any desire to hold back.

So you can beleive the last three paragraphs or you can believe that Intel is being forced to slow the rate of product advangement by the Carlye Group. It's up to you.

Re:Spikes in the electrical current.

[spectecjr]

So how exactly do these magical 'spikes' get through a modern, self-regulated switching power supply?

The modern, self-regulated switching power supply creates those spikes in normal operation.

(One of my coworkers once worked on ECGs... they had to synchronize everything so that all data acquisition occurs directly out of phase with the power supply... even the fans in the case had to be timed with the spikes it would generate... and this was a low-noise, custom generated, several hundred thousand dollars worth of R&D supply).