Coen Rijkaart kwam langs fietsen met een link naar deze Silicon Investor post, waarin ene Pete Gerassi een interessante observatie doet over de mogelijkheid van multiplier-overclocking op de Thunderbird en Duron. Als het onderstaande verhaaltje juist is dan zouden mobo fabrikanten vrij simpel de multiplier kunnen overriden. QDI lijkt deze methode te gebruiken in hun KinetiZ 7T Socket A plank, die volgens speculaties in deze nieuwsposting al over multiplier-unlocking beschikt:
I believe that all is required is a logic gate to force the FID pins to the proper states for the desired multiplier. All that needs to be done is to and the inverted RESET# signal with the PWROK signal. Then the result is connects a 4 pin buffer to output to pins FID0 through FID3 the settings of a DIP switch or memory location. This buffer should be set to output no more than 2.4V when driving high and 0V when driving low. Since this driving is for a very short time and the resisters that are "laser etched" can be overcome without any resistor overheating. This time period between PWROK being true and RESET# raising to true, is when the processor is looking at what multiplier to set. The bit pattern to set is seen in the diagram on page 62 of the pdf document. Since this can be done via a small amount of circuitry on a motherboard, the processor multiplier can be overridden at cold boot (hard reset) and warm boot (processor already going but software reset like in BIOS).
This is a better way than by attaching a GFD or placing it into a socket adapter. This does allow overclocking but the CPU will always tell the correct settings, if not driven. This allows one to read the AMD suggested multiplier. This shutdowns all but the "high end" remarkers while still allowing overclocking at the customers desire.