>Don't nForce2 mobos unlock the processor by themselves?
A myth that won't die. It orginated when the first Nforce2 mobos and the completely unlocked TbredBs hit the market at about the same time, and review sites did not quite understand that it was the TbredBs that were unlocked.
There are two things that AMD has left available that made it possible to change the multiplier, until now. (Even when Athlons were Slot A cartridges, a similar system was available.)
1) So-called bridges exposed on the top of the CPU. The bridges do various things, and there have been several systems. One of the things the bridges do is provide a factory set multiplier, which because the bridges are exposed, you can could change.
2) A set of bridges, which when joined connect the multiplier bridges to certain CPU pins. Then if the mobo maker implements the appropriate circuity, you can override the factory set multiplier. There are both VIA and nForce2 mobos which provide the circuitry, but it is almost universal with nForce2. Fpr instance, my several year old ABIT KT7, which only has 100MHz FSB, could set the mutiplier of my TBB to different multipiers up to 20. Until Tbred Bs, it was very rare to see these bridges joined coming from the factory. You had to join them youself.
There are very few mobos that can set all five of the CPU pins that control the multiplier. That's because before a certain point in history, Athlons only had a 4 bit multiplier, but TBred Bs have 5. The ASUS and ABIT control 4 of the 5. Therefore you need a pin trick (to set the 5th pin) if you want to override what it is factory set to.
BTW in all public AMD documentation, those pins (let alone the bridges) are not identified as controlling the multiplier, or as having any function. There has been no public documentation or support from AMD since the first Athlons, even though the possibility has been maintained, in different ways, since then. It is obvious that AMD could have set the multiplier at the factory using the same inaccessable method Intel does during all that time. It may even be cheaper to do it Intel's way.
Somebody said somewhere that AMD is losing $300 per CPU because of the overclocking. Ridiculous, of course. Sometimes I wish people would shut up on naive financial theories, but at least when they say something stupid, there is an opportunity to say something reasonable in contradiction. AMD is probably making money on the CPUs that sell for $50. IAC, people that buy Barton's for $80 to overclock are not going to pay $300 more, now that they are locked, to get the same CPU speed. But sure, there may be a little more pricing power gained by locking the CPU. Unfortunately, basic economics dictates that they will probably sell fewer CPUs as the price goes higher too, so there is nothing obvious about raising prices being a good thing financially. However, where AMD is really going into the red is on the new generation Athlon64s/Opterons, regardless of the price, which they have been spending oodles on to develop and buy the production capability for, and will contiinue to for some time, but have not yet sold enough to recapture the expense. Even when they go from the red to the black, probably this quarter, and will therefore be profitable, they will still be behind, because remember they have not made up the losses. This is just normal capital investment, where you have to put out a lot of money first before you can make money. Over the lifespan of the Athlon64, AMD will likely be in the black, and so this talk about bankrupcy is foolishness. It won't be until the next product cycle, and the next economic downturn after, that there could be any reasonable probability of serious finanacial difficultly, and that is way too far to foresee. So check back in five to ten years. At worse, AMD will still be hanging in there, and they could just as easily be a big money maker.