This Is Not Hard

From OCP:


Well, AMD gave us their OFFICIAL ANSWER about locking Durons and TBirds.

“AMD’s Socket A processors (The AMD Athlon and AMD Duron processors) are produced with the multiplier locked and the correct operating frequency laser etched on the processor die. This is done to help deter remarking of processors, which can lead to consumers not getting the product they think they are buying.”

I have to say there are a LOT of folks who have their panties in a wad about this and I simply see it as AMD protecting themselves. Just remember that before you buy into some big conspiracy theory, AMD could certainly lock the bus down as well as the multiplier very easily. With that said, we have not yet seen one of these CPUs with the pins removed. We are hoping to get our hands on one soon and it MAY NOT be as bad as everyone thinks. We will let you know as soon as we know. Then again, we may be totally off base here…just have to wait and see


Let’s look at this.

I don’t understand this at all.

Man is entitled to his opinion, but so am I.

Perhaps I’m missing something, but this remarker argument doesn’t work.

Does AMD have a point?

I’ve gotten some responses from AMD forwarded to me from people who’ve on this issue. Curious that such “personalized” messages follow practically identical scripts.
Kindly as they sound, they just nicely tell the overclocker, “Go away.”

Here’s the two I got:

“You have a valid point. Many PC enthusiasts are annoyed that the new Thunderbirds and Durons have a locked clock multiplier. But the reason for this was not to become more profitable, or to annoy you. It is to protect the end user. There is a large remarking industry that is taking AMD processors, overclocking them, remarking them, and selling them for a higher cost. In most cases this decreases the stability of the processor, thereby reflecting poorly on AMD as a quality manufacturer, when it is quite the opposite. AMD had to take a stand on this issue, and we took the side of the unknowing end user, and chose to protect them against the remarking industry, and keep the AMD quality standards.

“Being in tech support, I understand your frustration. But I feel this was a step in the right direction for AMD to protect it’s quality standards and it’s end users.”


“Thank you for your concern. This issue has problems on both ends. Lock the
multiplier and protect unkowning customers, but anger PC enthusiast like
your self. Leave it open, and loose customers who are angry because they
unknowingly bought a remarked part and it goes defective after a few weeks,
but keeping the PC enthusiasts happy, and the shady gray market remarkers
of the world. AMD had to take a stand for either side, and this is what led to
locking the multiplier, and hopefully you can see why this decision was


Remarking? What Remarking?

For such a threat, AMD seems remarkably vague about the details. All we know about is some Slot A chips found in Australia that some Asian group apparently tampered with a bit. That’s it.

Funny, but when Intel was introducing their CPUID program, they were much more specific about the problem. They had news conferences. They talked about specific raids. They explained methodologies. In short, they presented a case just to justify the introduction of a software program which actually does the job.

AMD’s done none of that, but even worse, the actions taken do not solve the problem.

We Only Care About the Dead and Crippled?

Every CPU has a CPUID built into its microcode. This is used to identify the processor. Intel has used the CPUID instruction to include the processor’s rated speed starting with the PIII. AMD currently does not, but easily could.

Once you have the speed encoded into the CPU’s microcode, it’s available to be called up by all kinds of programs. For instance, rather than say “AuthenticAMD,” you could have Windows report “700Mhz Duron” or something like that.

The sad reality is that current AMD processors will be wide open to remarking so long as the L bridges governing the speed of the CPU remain exposed. Change the L bridges and you change the speed of the processor, with nothing inside the computer’s microcode to tell a user otherwise.

All that getting rid of the pins does is add one other thing to look for in a remarked chip. Under normal circumstances, the average person doesn’t even think about looking at his or her CPU. Things will have to get pretty bad before somebody’s going to open up the box him or herself and inspect the chip.

Unless you have some extremely greedy and stupid remarkers, odds are that most remarked chips will either work fine (in which case AMD gets ripped off) or not badly enough (after all, this is Windows we’re talking about) to be physically inspected (in which case AMD’s reputation gets ripped).

If I’m worried about my reputation, I’m not just going to identify the worst-off of the victims. I want to identify all of them.

Mind you, what is the remarking victim most likely to do with a quirky machine?
Why, bring it back to the place where he or she bought it. Do you think all or most resellers will immediately own up to selling a remarked processor and happily replace it? I’d bet a healthy chunk of those most likely to use such processors, knowingly or semi-knowingly, wouldn’t.

If AMD were truly concerned and serious about protecting the innocent user, they’d add the speed to the CPUID instruction to ensure prompt, undeniable identification of remarked chips. If they do that, they don’t need to do the other things to protect the customer. If they don’t do that, they aren’t protecting the customer.

If this is bad, isn’t it bad for everybody? If not, why not?

FSB Locks

Actually, if you’ve read what I’ve had to say about this , they and/or Via have done a fairly good job of that already.

A true FSB lock would actually be rather difficult and expensive to implement, though, this was discussed extensively when Intel looked like they might be contemplating it.

You’d have to have a circuit that would have to measure very precisely the timing of the bus’s clock cycles. To make a long story short, we’re talking about a circuit capable of measuring sub-nanosecond differences in timing. That’s not cheap, and probably why Intel has never tried such a thing.

The way AMD’s effectively done it is to have the clock generator limited, which actually is a much better approach than a high-tech circuit. If you make FSB overclocking more trouble than it’s worth, you prevent all but the diehards from trying it. If you can stop 95% by doing next to nothing, but it takes a lot to stop 100%, you’re better off with the 95% solution.

I’ve raised some questions, and came up with some possible answers. I’m happy to hear any better explanations; I even provided an alternative myself. I just want to get to the truth. You don’t answer the questions by calling me nuts; you answer the questions by proving me wrong.

The real issue here

The issue is not that removal of these pins makes overclocking impossible, it is that it will make it difficult enough to preclude most current overclockers. We’ve already mentioned on our pages two ways it can be done, but both will essentially preclude most overclockers. That bothers us.


If AMD were truly concerned about protecting users, they would provide protection for all users just like Intel. Since they haven’t, and since it makes the task of the serious remarker just a bit more difficult than it was before, it’s difficult to take the presented arguments or arguers seriously.

AMD has the perfect right to do what it likes, but we have just as much right to reject nonsense.

If AMD said, “Look, we think you overclockers are costing us a lot of money, and we think we charge a fair price for a good product,” that would probably approximately what they truly believe, and at least I’d respect their honesty, even if I disagreed with their conclusions.

I just don’t see that here.

Email Ed

Be the first to comment

Leave a Reply