Intel Puts The Lock On Overclocking
Maybe you've already heard rumors that Intel's new platforms do not offer much leeway for FSB overclocking. As VIA is currently fighting to get both AGP and PCI Express x16 working in their PT890 and K8T890 chipsets, insiders assumed that the implementation of PCI Express would cause these limits.
Intel does not want history to repeat itself. A common approach for avoiding undesirable situations is to do away with all unnecessary options, such as additional multipliers, but this strategy doesn't seem to be enough any more. This time, the chip giant not only cuts away functions that are regarded as non-essential, but it is actively adjusting hardware in order to prevent the industry (and as a consequence, the users) from anticipating upcoming features like DDR2-667 and FSB 1066.
Of course we tried to get some DDR2-667 DIMMs in order to see what they can do. The small memory makers do not have these chips yet, while the big ones claimed not to have any samples for the press. Interesting, however, is the fact that a well known memory specialist will launch its DDR2-667 memory as early as July 7.
Anyways, as long as there is no chipset that supports DDR2-667, these fast DDR2 DIMMs only make sense if users can overclock their systems. What Intel did is implement an overclocking limiter to the MCH chips: If the CPU clock exceeds the threshold (we determined that this is 10% over specification), the required PLL (Phase Lock Loop) will reset and won't refuse to lock that frequency. Basically that is a very simple way of throwing a spanner in the works, as it causes a system crash. Decent motherboards automatically restart and you may try again. The easy way would be to limit all overclocking ambitions to 10% max, but now that Intel added this extra obstacle, I somehow felt compelled to break some barriers.
