amd's big blunder

[acejj26]

For the record, according to the email Tom Pabst got (and in accordance with Anand's latest findings) the production durons/tbirds are multiplier-locked. Initially, Tom got a pre-production duron for review purposes and its multiplier was not locked, so he could overclock to his heart's content. Maybe AMD saw that the Duron had so much headroom for overclocking that it would take away sales from their TBirds??

If someone wasn't sure if their processor was true blue, couldn't they just look at the die on the processor since AMD is using flip-chip packaging to see what the processor is rated at?? Just a thought.

 

[Adul]

The averag user would not even think as to opening their case. What makes you think they would look at the CPU to see its rated speed. Actually, how many of them can even identify what the CPU make and speed are? Has for us, I think to may of us could still be fooled pretty easily.

 

[acejj26]

Adul,

If that's the case, then remarkers wouldn't even have to solder slot A processors to remark them. They could sell them in a computer with a GFD attached, no? The remarkers of slot A athlons made a lot of money by soldering them and the reattaching the heat plate and sluffing them off to OEM's as genuine chips when they weren't. With the socketed part, is that even possible since the core is exposed??

 

[BoberFett]

jpprod and Rogue are right. A motherboard won't suddenly overclock your CPU just because it supports 133 FSB. If you leave all the settings on Auto Detect, you can swap a Celeron and a P2/P3 and neither one is running at the wrong speed. The Celeron uses a 66 FSB and the P2/P3 uses 100 or 133.

And when you think about it, the AMD/VIA platforms have an added perk, the memory can be run asynchronously. You can choose to run the memory -33Mhz or +33Mhz. So if you put a Duron on a 760 with a 100 Mhz DDR bus, you could probably run the memory at 133 Mhz DDR.

 

[DaddyG]

Hold it !! There are too many Durons out there that overclock using either BIOS or Dips for them all to be Engineering Samples. These are Production Chips for sure. AMD can, and probably, will change the chip with a later stepping, but I believe that all the Durons in inventory right now will overclock. Besides, overclocking with mobo controls is not remarking. Cutting the links on the chip and then resoldering or using silver ink would be remarking but this would be obvious to a chip reseller or consumer. (The remarked chips are always OEM, no heatsink/fan attached). I think that AMD is reacting to a problem that existed with the classic slot A Athlon, which because of its SECC package hid the fact that it was reworked. The socket chips will obviously show remarking efforts.

 

[acejj26]

DaddyG,

That is my point there...remarking should go down considerably with the socketed chip with the core exposed. Someone who buys an OEM processor will know right away what the chip is rated. With the "golden bridges" still on the package, hardware modification can still be done to reset the multiplier, but the consumer will still see what rating the processor is. So if a company sells a computer with a duron/socket tbird in it and the computer is unstable, the consumer will take that up with the manufacturer. Such an RMA is costly, so that should defer most scrupulous OEM's from doing so. Then, those trying to sluff off the processor by selling it bare won't be able to remark the chip since it is stated right on the core what speed it is. I really don't see why AMD can't just let us use BIOS/dipswitches to change the multiplier. I really hope they rethink this strategy.

 

[MadRat]

Hey I broke the story of the "now overclockable" Durons. The guy here in Bellevue didn't say he resorted to overclocking the FSB; he said he adjusted the multiplier! His Duron did 950mHz from a stock of 700mHz.

Overclocking the Duron/Thunderbird through the FSB is not very feasible. First of all the AGP/PCI ratioes increase with the FSB. (i.e. 133 FSB = 44.3mHz PCI and 89mHz AGP!!!) Secondly, the DDR overclock on the controller is less likely to succeed in older Slot-1 boards.

I wouldn't mind overclocking the FSB if there were PCI/AGP dividers.

 

[Howard]

The Duron and the Thunderbird are still overclockable, guys.

All you have to do is modify the bridges on the PCB, not unlike the original modding of the resistors on the Athlon, although it's likely a tad bit more difficult.

Read that little overclocking guide by Tom (yes, I know *shudder*).

It explains in detail.

 

[KoK_Thunder]

Read Anand's article and you will see that with a production AZ-11 board the CPU's WILL NOT overclock.

 

[MadRat]

But the guy I speak of was using a Microstar with BIOS adjustments!

 

[jpprod]

My MSI-6330 (K7Tpro) doesn't have the BIOS multiplier adjustements anymore. I wonder whether it's simply a matter of BIOS - in that case there'd still be hope for me to overclock this Socketed Tbird/800...

Hey, you gotta have optimism

 

[mschell]

Both Intel and AMD have no problem motherboard induced processor overclocking. Intel sells millions of chipsets to manufactures like Abit who specialize in producing CPU tweaking motherboards. What they don't want is people physically modifying the chips to run at a higher clock speed. CPU multipliers not locked in the core can easily be manipulated with a little soldering work. There's enough profit in the remarking to make the modifications look factory done but things are a bit tougher with the exposed FCPGA CPU's.
AMD's problem(not overclockers) is they implement a system that fixes the multiplier after the chip has been tested which means it isn?t locked at the core level which means it can probably be changed. That's good for hobbyist but bad for AMD because if it can be changed, it will be and the losers are the people and companies buying OC'ed CPU's. - M.

 

[Euclid]

There was a serious problem with people remarking processors in and around Australia I believe. If you buy a remarked processor, and put it in your machine and it crashes...who's responsible...

Most people would hold AMD responsible. That's the problem. It would be the equivalent of you buying a car from a dealership that was "tweaked" to get an extra 20 miles per gallon. But Every now and then it would stall and die. You would say, "Damn, Ford sucks." In reality, it was the shady dealer who screwed you.

In order to keep people from getting screwed, AMD has to eliminate easy remarking.

The % of people who overclock their processors is very very small. That's why it makes good business sense to do this.