Those supporting AMD should remember that they did a similar thing after it was discovered that some processors could have entire cores re-enabled.
On the other hand, if I remember correctly one of the mobo makers soon figured out how to use both the new microcode and the old one to get the best of both worlds; it might've been AsRock too...
Those supposedly triple and dual core CPUs that got unlocked to quad cores were apparently a way to sell damaged silicon that the remaining cores passed QA so they just disabled the broken core. To meet demand of these product lines perfectly good quad core CPUs got disabled and sold off as dual/triple editions. It was the luck of the draw if you got one of those and not a genuinely flawed chip.
Now, How much of that is true I don't know but that's what I remember reading at the time. A friend of mine did succesfully unlock his triple and ran for a long time.
> Those supposedly triple and dual core CPUs that got unlocked to quad cores were apparently a way to sell damaged silicon that the remaining cores passed QA so they just disabled the broken core.
It's called "binning", and it's not just for defects. CPUs from the same chain can have different tolerances, one will reach 3.5GHz easy and the next one won't be stable beyond 2.8. Those are also put in different bins and sold as different models.
> Now, How much of that is true I don't know but that's what I remember reading at the time.
It's completely correct. Selling a model from the exact corresponding bin is ideal, but if you have more demand than the bin provides you get parts from higher bins and gate them. A few years ago that got very very common with Intel parts as they reached tremendously low defect rates, and more or less any CPU you bought would come from the highest bins gated and undermultiplied to whichever model you'd buy.
Floorsweeps are done when a part of a chip isn't passing tests. The defective part gets fused off and the chip gets sold as a dual core instead of a quad core, for example.
Binning is orthogonal. It's a qualitative measurement of (the part of) a chip that passes all tests. The electrical and thermal properties of a chip are tested, e.g. the leakage current and temperature are measured at different clock speeds and voltages (this is infinitely more complex for a battery powered device where voltages and currents fluctuate). The ones that pass with best results are sold as a premium product and the rest are clocked down and sold for less.
So a "K" model Intel chip comes from the best bin. An i3 chip dual core is a floorswept quad core i5. (This is what I assume, I don't work for Intel and don't know the details)
But these chips may not sell in the proportion they get manufactured in, which means that some perfectly functional quad cores get fused to dual cores and sold for cheaper. If you're lucky, you're getting one of these (and un-fusing, if possible, will work).
Floorsweeping and binning are required because the tolerances of modern semiconductor manufacturing are so tight. To get the best bin to perform well, the manufacturing process is really pushed to the extreme, which means that there will be chips that have manufacturing defects as well as lower performing chips.
On the other hand, floorsweeping and binning cannot exactly match market demand. Towards the end of a run of a given chip, most may qualify for the highest bin - but due to demand many have to get moved down a bin or two. So you end up with totally capable of being overclocked CPUs being sold as lower clock.
This has been going on for a looooong time. Old time intel 486 "SX" chips without math co-processors (floating point unit really) were originally 486 "DX" with either the co-processor disabled or which failed QA. At some point, they made actual SX without FPU built-in.
The "stand alone" math co-processor you could add to an SX based system to gain h/w floating point was actually a full blown 486 dx CPU with a different pin layout (one extra pin) so it couldn't be used as CPU and could therefore be priced differently.
I'm still using all 4 cores in a "triple" core phenom to this day. I believe it is also overclocked roughly 400mhz, my wife sometimes plays games on it.
I'm running such a system at this very moment. It has a Phenom II 550 Black Edition (i.e. sold to be overclockable/unlockable), and by toggling a BIOS option, I was able to unlock two more cores. I run it at rated clock speed, and it has been rock-solid since the day I built it up.
I can build a large project in NetBeans as fast on this system as on a hex-core FX-6100 (and I can watch my CPU monitor clearly showing all four cores running during the build).
CPU temperatures stay well within limits with the stock cooler.
The difference is that Intel is forcing this update (via Windows updates and by forcing mobo makers hands). Don't think AMD ever had that kind of power.
I am curious which AMD chips had the unlocking disabled?
I am still rocking AsRock 970 MB with AMD B50 which actually is an Athlon II X3 unlocked into a full blown Phenom II X4 (3 cores with no L3 cache into 4 cores with unlocked L3 cache).
On the other hand, if I remember correctly one of the mobo makers soon figured out how to use both the new microcode and the old one to get the best of both worlds; it might've been AsRock too...