Slashdot Mirror
AMD Ends Overclocking On Durons
Jai writes: "AMD have finally done it. The crime has been committed and now we must reflect on this and look at setting ourselves a new pathway to overclocking AMD's successful Duron and Athlon Processors. Basically this means the pins for the multiplier are gone and no longer physically exist on the chip. Insane Hardware have a Duron and they are showing everyone as the overclocking world is filled with gloom."
Comment removed based on user account deletion
Digital electronics don't work that way. There's no such thing as "prone to errors" or slowly breaking down. Either it works or it doesn't.
Digital electronics is still made of analog transistors, which are made of diffenrently doped Si-regions (n/p). Every increase in temperature will increase the mobility of dope inpurities and thus change the n and p regions. This can lead to damaged transistors or transistors that do not provide the proper voltage levels. Of course there is a safety margin in voltage levels, so it will take considerable diffusion of impurities to render the electronics useless. Breakdown of digital logic will occur suddenly (thats probably what you mean). Basically, every temperature decrease will increase the lifetime of ICs.Whilst I'm not disputing the possibility of AMD nailing the multiplier hack in their new chips, I doubt they'd be removing pins. In fact, on closer inspection, there are four perfectly blue squares in that image -- where there should, probably, be a few golden pins and a pixelated blue background. Use a decent image manipulation program and zoom in on the area highlighted with the red lines in the image, till you get to about 12:1. It's not all that conclusive, and it's a damned good job if somebody did doctor it, but it's suspicious all the same.
Eh, this is OLD NEWS.. And UnaClocker has already found a work around... http://www.procooling.com/html/the_athlon_conspira cy.shtml
Because AMD is stopping overclocking it's preventing vendors who are dishonest. Now you or I can not be sold a chip which is overclocked and be told a lie to which the real fabrication limits are.
Justen Stepka
This is only the end of the mulitplier unlock. It was nice while it lasted, but Duron owners will still be able to toy with overclocking by cranking up the bus speed. There's still a lot to be played with-- Celerons and P3s are still commonly overclocked huge amounts by computer hobbyists! Mantle
What is needed here is not a solution consisting of "crippling" the processor to make it impossible (or more difficult) to overclock, but a clear method of telling if it's overclocked or not.
AMD, how about putting in a microcode program into the processor's initialisation sequence to check if the processor is overclocked and display a big warning (on standard VGA hardware) for a few moments (or until a key is pressed on a standard PC keyboard) if it is?
That would stop the grey-marketeers in their tracks and also give the legitimate overclockers something to point to and say "Look what I did!"
Well to tell you the truth this happened about 2 weeks ago. But many overclocking sites have stores which have thundrebird and durons for sale. These chips are pre-unlocked and ready to overclock. One such site is the Over clocking store at athlonoc.com.
"I INSTALLED LUNIX AND FPROTTED HIS TARBALL!!!!!@#"
How does this help AMD?
Unfortunately, overclocking is not restricted to a few hobbyists. I remember taking a huge heat sink off a Micron 166, rubbing off the compound and see a clear "Intel 133" mark on it. Manufacturers have been playing fast and loose with us for years, and when they screw with the CPU, it's just asking for Intel or AMD to come up looking bad. The average user will never know that it was overclocked (or even what that is), just that, "AMD makes junk!".
1Alpha7
Live to be Moderated
It's really a shame... At first, AMD supported the overclocking fully. The problem was that nasty resellers (like CompUSA) would grind the original markings off the faceplates and mark it up one jump (a 600 would become a 650, etc.) and people installing the chips in their computer would set the spped at that. This was _real_ common in the Socket 7 days, when there was no protection whatsoever for this, which is why K6 series had engraved metal faceplates instead of paint (it made remarking easier to spot).
Then AMD put the copper bridges on the top of the Socket A chips to set the multiplier, a technique which made it harder (while still possible) to change the clock, and which made remarking also quite apparent. If I understand the insane hardware article, this no longer works either.
This shows the problem that poor reselling techniques (i.e. fraudulent ones) can have with the community at large.
"Your mouse has been moved. Windows 95 must be restarted for the change to take effect."
Doesn't AMD have a responsibility to limit overclocking in order to eliminate the fraudulent sale of overclocked systems?
Most retail PC buyers couldn't tell if a system has been overclocked or not. And if John Smith's home PC eventually burns out, doesn't it negatively effect AMD's reputation?
This is hardly a cause for concern, seeing as how Tom's Hardware has an article on how to change the settings without needing the pins.
AMD know fine well that most of us (spare the die hard freon cooling bunch) would likea bit of extra CPU power.
My next system upgrade was planned to be a duron 700 running at 900mhz which would give me 87% or so of the power of a 1ghz Thunderbirdie.
I could afford the thunderbird. Sure it would have meant no voodoo5 and maybe even no film scanner but by locking the duron they want to push me into spending a higher proportion of my budget on a cpu that I otherwise would.
Plus it will raise the overall percieved stability of their cpus since as mentioned before the vendors will get away from it.
That's still pretty naieve to think that that eliminates the problem.
Ultimately the BIOS in most modern systems can easily be replaced since it's only software.
If a manufacturer is sufficiently dodgy to be selling marked up cpus then they probably woudln't realy mind flashing the bioses to hide their sins.
Anyway you are already expecting all motherboards to have their existing bioses replaced with new ones that support this extended CPUID. No novice user is going to recognise that their motherboard is a few months older than it should be...!
Anyway there is also the further problem of how the fsck do u expect AMD to do this. I guess they'd have to generate difference wafers for each speed of CPU.
At the end of the day AMD and Intel only manufacture a few different chips and then sort them by tolerance to feed the different markets.
In transistor terms theres really no difference between a 700 and an 850 but in order to keep the overclocking market you want AMD to double their overheads..?
one of my friends worked in an intel fab lab over the summer, and he is all worried about overclocking now... the smaller the pathways get on the actual chip, the larger the risk is that overclocking can actually destroy the chip.
;)
In stopping overclock, AMD is merely doing the same thing that manufacturers did when they made a safer system for electrical outlets: protecting the end user from damaging their product or themselves. Granted, overclocking a chip and burning it out is not going to physically hurt the end user (not compared to sticking a finger into an unprotected wall outlet) but when you find out that you just melted your new processor--you might punch a wall rather hard--which could physically hurt you
mov ah, 0
mov al, 13h
int 10h
mov ax, 13h
int 10h
AMD is working with AMI and Award on having the BIOS be able to recognize the clock that the chip is supposed to be running at as well as the current clock speed. So when you boot up you'd see:
Duron 700 running at 1003
or something along those lines. Would be a big plus, as it would allow them to let us overclockers have our way with our chips, and prevent oems from being able to sell chips at clocks they arent supposed to be running at.
The multiplier pins may be gone on your new Duron, but the multiplier can be unlocked anyway!
Read about it here: http://www.tweakers.net/reviews.ds p?Document=150
Woohooo 900Mhz, here I come!
And on top of that, I've heard that this Insane Hardware article is a hoax...
;-)
So until I actually see a chip for myself that's missing these pins, I'll continue the direction I was headed before I read this crap.
Of course, there has never, ever been a hoax or rumor posted on Slashdot.
--- "So THAT's what an invisible barrier looks like!" - Time Bandits
I know there's a large (or small but vocal?) contingent of people who think that overclocking is Sticking It To The Man, but it's a pretty pithy form of rebellion. You'd be hard pressed to find someone who really benefitted in a measurable way from a 10% speed boost. Does Word load faster? Are you getting 357 frames per second instead of 348 in timedemo? Is your internet connection faster? Are you saving months of time running those complex numeric simulations you're always running, the kind that take months of computer time?
Heck, you already bought your Athlon or Pentium. You're giving money to AMD and Intel; you're not putting one over on them. They love it if you run your CPU too hot and have to get a new one more frequently. So let's stop acting like this is subversive, okay?
here's one. Unfortunately not for the faint of hart. It requires some soldering...
Well, it's interesting that you would say that. It's hard to tell how much AMD is actually stopping remarkers by doing this, compared to making it harder for overclockers to changer their multipliers...
Taking out the pins only disables the functionality about being able to easily adjust the clock multiplier in BIOS. The other way to change the Tbird/Duron's clock multiplier is to modify the copper bridges on the top ceramic of the processor, by cutting some with an X-acto knife and closing others with a conductive pen (or even pencil lead).
When the processor is overclocked in BIOS, it's much easier to tell, since you can go into the section that has FSB and multiplier adjustments and see if they're set at default or at a specific clock lock. If the bridges were modified, you can't tell at all until you open up your computer, take off the heatsink, and look at the top of the processor's die.
So as far as remarkers are concerned, I'd imagine that they'll go ahead and do the bridge modification, especially since it's harder to detect. Meanwhile, the overclocking hobbiests can't take advantage of the overclocking features of their new Asus K7A or Abit KT7 motherboards they just bought.
Does no one else remember an article posted just last month dealing with this? AMD has already announced that its future chips will DECLARE ON BOOTUP what speed they are made for and what speed they currently are. This will make it blatantly obvious for any users who buy an OC'd processor w/o knowing it. So with this technology being implemented, why hurt the OC niche market? I imagine that AMD gets a TON of great word-of-mouth advertising from AMD fanatics (myself included), so why shoot themselves in the foot on this one? I don't get it...
------
Let me give you the lowdown
It's not just the cheaper end that the Duron is good for, though. Since it has such good FP, it's a slick chip for a rendering farm - It's cheap, motherboards are now or soon will be cheap, and it has very good Floating Point math. I bet you could slap them into some 1U or 2U rack cases and get some sick flops that way, running NT/Win2k or Linux, depending on what software package you're fond of. Ahhh, ScreamerNet...
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
When Intel released the Pentium there was a huge trade in remarked Pentium chips.
:)
In my opinion the Pentium was a robust chip, but when an OEM was putting a remarked chip into a customers machine they might have been effectively reducing the MTBF of the device.
Whilst the majority of users will never know that they have been conned, (whats a few blue screens between windows anyway) the ones that do find out tend to do so after a terminal failure.
All this did is hurt the Intel trademark, whilst some users came to the conclusion that Intel chips inherently fail, other users realised that their chip was in fact remarked and they thought to themselves why isn't Intel doing anything to protect me as a consumer?
It turned out to be a double edged sword which hurt Intel.
I imagine that this phenomena has been a significant factor in the development of AMD's market share.
This is the reason that Intel released the 100Mhz pentium II's with a multiplier lock.
Now that AMD have managed to release high quality chips (shh be quiet you RISC zealots) which are capable of overclocking. They are facing the same scenario.
Its important to note that they aren't trying to put an end to enthusiast overclocking, they are trying to put an end to large scale remarking cartels.
PS:- I am not implying that overclocked chips are the cause of all bluescreens in Windows
I can understand that they'd be worried about unethical retailers. There seem to be a lot of them around. I'd think that they could come up with some method of verifying the chip's true speed without locking it, so the folks who like to do that sort of thing could still do it. AMD's in a great position to stomp Intel at the moment (as I've said before) but they really need to keep all the friends they can keep.
I'm trying to teach myself to set people on fire with my mind... Is it hot in here?
As for "professional", unless you're involved in video/MPEG-2 rendering, realtime digital audio processing, or heavy-duty CAD rendering, there's a fine chance that any CPU still made will be just dandy for professional use. Most professionals spend their time in compilers, word processors, spreadsheets, or other non-CPU-intensive apps.
Most people find Internet just dandy on anything made in the last 5, maybe 10 years. Despite all of Intel's consumer ads, most of the time you're bandwidth limited... and I'm saying this from the other end of a serious optical pipe, not a 56K dialup.
As for games, sure, some of those kiddie demand the fastest performance they can get their hands on. So they're running 1GHz Athlons if they're dedicated. Compared to most computers in most regular consumer homes, the Duron is very fast. After all, it's mainstream, and it's relatively new. You might expect Intel's next major Celeron upgrade to run a little faster than the Duron, if they can swing it.
-Dave Haynie
Those Wackos at AMD have done it this time! Now they're just some chip manufacturer who sells CPU that are better, clock-for-clock, than their Intel equivalents, for less money!
I don't know if I can continue to give them business in light of this news.
:-)
AMD, like Intel, doesn't give a rat's ass what you do to your own cpu. Burn it out? They don't care one way or another.
It's the grey market area that they want to stop. These shaddy OEMs, and local PC shops that like to OC a 600 chip to 800, then sell it off as a genuine 800mhz system. While the novice computer user that buys these secretly OC'ed systems isn't any the wiser.
What happens? Their cpu burns out way too eary, and when their computer stops working due to a dead cpu, they don't give any though out to OEM that sold it. The consumer just thinks that AMD makes crappy cpus, and will just get an Intel chip the next time.
So, I dont' blame one bit for AMD to stop as much as they can on OC'ing their chips. It's not to stop your or me from taking one of their chips up a few extra mhz. It's for the bonehead local OEMs from selling a Duron OC'ed to 850+ as a genuine 850+ chip. Which sadly to say, I've seen it done all too often.
Tier 1 OEMs and authorized vendors were pushing for this. You see, if a part is readily overclockable, it is easily remarked as a higher speed part and sold at a higher price.
(I have experience in this arena, I used to work in the anti fraud group for a x86 mIcroprocessor maNufacTurer that will remain namELess)
Remarking microprocessors can be big business for remarkers. Once fully automated processes are put into place, they can "reprocess" (that's the term they use) hundereds of processors a day.
When housed got raided, there would be many hundreds of parts in process of having their appearance altered.
Remarkers don't just focus on the chips. They have the ability to produce retail boxes, appearing to all but the most highly trained eye to be authentic.
Remarking the chips and making phony boxes is the easy part. Getting the parts back into distribution is the hard part.
Just how many computer shows are there on any given weekend? Not many. How many mom/pop computer shops are just itching to save $5 on each processor, not really caring if the parts don't come from the authorized channel? Tons!
You see, if the grey market proliferates, it does so at the expense of the authorised distis and the tier 1 OEM markets.
Expect to see an even greater embrace of AMD's processors by big name manufacturers now.
Joe Goldmeer
Comment removed based on user account deletion
If you spend less money on the slower chips and make them faster, in theory AMD is losing money.
It also allows you to stay on the trailing edge of technology by buying last year's chip and keeping it for as long as you would a new chip.
The final reason may just be a preemptive strike against returns/et.al. An o/c'd chip is more likely to show manufacturing defects, too.
I don't know if any of these reasons lose them money because they are creating brand loyalty by selling o'clockable chips and you will also buy a new one when the old one burns out.
IANAL, but I play one on
oh contrair...
I had an athlon 300a overclocked to 450 for a little under a year. It never got terribly hot, but eventually it just died. By being dead I mean that it just started spewing out errors, more and more often until I got a new chip. Then all was ok.
Do you even know anything about perl? -- AC Replying to Tom Christiansen post.
I have always wondered if a company A could make a chip with the same performance as company B, but send it out on the market at 100mhz faster than it's rival. (which would then be compared in price with company B's faster chip)
*COUGH* 1.13GHz PIII *COUGH*
2) they are pushing a bit too much on their new chips and don't wan't people to notice their "safety margin" is smaller?
This has nothing to do with it. The only way to judge the safty margin is stability and failure rate. The multiplier can always be figured out by dividing the rated chip speed by the FSB speed.
There is more to bin sorting (determining CPU speed grades) than stability or other technical considerations.
Currently AMD is marking chips at LOWER than they are capable of running in nearly all cases due to marketing strategies, not technical limitations. AMD's K7 class chips (Athlons and Durons) are known to run stably at higher speeds, but AMD is biding it's time and keeping this headroom available for the coming mindshare fights with Intel's Pentium IV.
This is smart for AMD. The top chip, whatever speed it is, will sell for about the same price. So they can sell the lower clocked chip until Intel is able to counter with something with a faster clock speed. When that happens, AMD just changes the speed marking on the the chips they are already producing.
Yes, technical concerns are a limiting factor on chip speed ratings, but they are not the operative one for AMD right now.
Tastes Like Chicken
Why don't chip manufacturers like AMD and Intel simply put an instruction in the CPU that can be called to get internal information about the chip? (e.g. Company, intended clock speed, etc.) This especially goes for Intel, seeing as how they've already done it for serial #'s.
Arguing about vi versus Emacs is like arguing whether it's better to make fire by rubbing sticks or banging rocks.
Would be to burn the CPU spec in some kind of ROM on the CPU. I know intel puts some scrach ROM on their chips for info, and both of them have CISC->Microcode lookup tables on their chips. Return that number with some of the other CPU-ID stuff. Instaid of "Athlon 800" the system would say "Athlon spect @ 700mhz (Overclocked to 800)" at boot.
Bios's could even be set up to display a warning if the chip is running out of spec.
Anyway, I doubt that Remarking is really what the chip-makers are worried about. I'm sure AMD would rather have us pay the $500 for an Athlon 900 then get a Duron 600, overclock it to a Ghz and get the same performace for a tenth the cost. Sure, these chips would cheap if AMD just gave them away, but I think they are more intrested in making money then being good semaritans.
I ordered my Duron 600 today, I hope I get a good one, but if I don't, I won't be to pissed, I only paid $50 for it, I certanly wouldn't think of myself as stealing anything if I can run it faster. I havn't got any money at all, and I couldn't afford to get a highpowered athlon. but if I could, I still wouldn't if AMD didn't pull these pins.
ReadThe ReflectionEngine, a cyberpunk style n
The submitter read the article wrong. AMD hasn't blocked overclocking. What they've done is re-locked the front-side bus multiplier on their chips.
This renders the AMD Duron supporting just as much overclocking as your favorite Intel CPU. Before this, they were _more_ overclockable.
Remember, Intel just got burned a few weeks ago (remember the 1.13 G pIII recall?) by pretty much deliberately overclocking chips that couldn't take it. So, it really shouldn't come as any surprise that AMD would do some testing, figure out how fast these chips can run SAFELY, and then take steps to ensure that they run at those safe speeds.
Sure, they might lose a bit of favor with the geek crowd, but the money, especially concerning the Duron, is in run-of-the-mill discount PCs. Having them stay sturdy does a lot for AMD's bottom line. All hail the dollar!
Ok. I'm entering the fray quite late and nobody will pay attention to this post.. but it has to be said..
with AMD and Intel, this whole MHz race is bullshit. There is no *real* advancement in technology when all these guys are doing is tweaking these old technologies to run faster.
But its this "hype" about MHz, driven mostly by you stupid overclockers, that keep AMD and Intel locked in this MHz race that makes everyone lose.
I mean, go look at MIPS who actually give a hoot about decent design - their R12k 400MHz is twice as fast as a PIII 800MHz at seti! And go check out PPC's Altivec, which is pretty much MMX except 10 times better.
But of course, the rest of the world seems to neglect these achievements in computing, and think that more MHz == better. WRONG. Go have a look at SPEC and see the current trend which shows that even though intel and AMD are making their chips run at a higher MHz, the performance gains really aren't that much.
So, we have this whole overclocking and MHz frenzy, which is basically driven by a bunch of clueless kiddies who believe that making a 5 year old technology run at a fast clock rate is the be all and end all of chip fabrication.
pfft
Folks,
While some of you folks want to overclock the heck out of the CPU, have you all considered that there are other ways to speed up your computer?
How about installing more RAM and a faster hard drive? Getting more RAM into a computer can make a BIG difference, especially in graphical environments. And definitely get a 7200 RPM ATA-66 hard drive, too, because you want to be able to read and write data on the hard drive faster.
In my personal opinion, once CPU's went past 500 MHz in speed, for most home users they would be better served by getting as much RAM as they can afford and making sure the system has a decently fast hard drive.
Raymond in Mountain View, CA
AMD right now is hurting a little because a LOT of people are returning chips that have "gone bad". Now, do you think the chips really just "GO BAD" or do you think they're overclocking their chips??
AMD needs to save money to beat out Intel. I love AMD, and if they think they need to disable the OverClocking of their chips to stay ahead of intel, so be it.
Plus now the chip differences in price are praticly nothing. You can afford a 900MHz T-Bird these days!!!!
AMD is doing everything they can to get us the best chips possiable and keep compitition with Intel realistic.
Frankly, if I spend $500+ bucks for a processors, I'd be less inclined to take risks with it. Cheap processors on the other hand make it easier for us to experiment because if we burn one up it's no great loss and replacing it isn't a wallet-emptier.
When I spend a lot of money on a CPU specifically for performance, I tend to lean towards the reliability of it's stated speed rather than the risk and relative unreliability of overclocking - after all, I've already got performance. But if I want to see just how far I can push a CPU - well, I expect unreliability and I expect to crater some chips. That's part of the fun...
Mike