So, the 925A01-8EKRS2 and 915A01-8EKRS2 mainboards from Foxconn aren’t the overclocker-friendliest mainboards out there, but they do permit to adjust any of the most important settings. We only have gripes about the insufficiently wide range of the memory voltage, since there are advanced DDR2 SDRAM modules available that are rated for 2 volts, which is 0.2v above the default.
Well, the selection of overclocking-related options in the BIOS Setup is just one of the prerequisites for a satisfying overclocking experience. To make it possible, the manufacturer should also take care about an optimized BIOS code, about an appropriate design of the mainboard’s PCB, about the use of high-quality components. And this means we can’t judge a mainboard’s overclockability without some experimentation.
So, we wanted to find the maximum FSB frequency the 925A01-8EKRS2 and 915A01-8EKRS2 mainboards were stable at. To perform this test we took an engineering sample of the Intel Pentium 4 570J processor (3.8GHz frequency). Since this processor has a non-locked CPU multiplier, we dropped this multiplier to 14x to be able to step up the FSB clock rate without any limitations. For our overclocking tests not to by limited by the capabilities of the memory, we used overclocker-friendly Corsair CM2X512-5300C4PRO modules capable of working at up to 675MHz frequency.
The Foxconn 915A01-8EKRS2 mainboard, based on the i915P Express chipset, was the first to take place on our testbed.
Starting out cheerfully we soon stopped at 217MHz FSB – the system would not start up at a higher FSB clock rate. That’s not much, yes? One might even believe to people saying the i915/i925 chipsets were made protected against overclocking, and the mainboard makers had to avoid this protection. So, are these 217 megahertz really the overclocking peak for the 915A01-8EKRS2?
Of course, there’s no special protection against overclocking in i915/i925, but mainboards based on these chipsets are really difficult to overclock because of the high-speed PCI Express bus whose implementation has made the chipset’s arbitration logics more complex. The arbitration unit is a highly sensitive mechanism, which can be influenced by a slightest change in the speed parameters of the outgoing and incoming signals. And this unit affects the stability of the whole chipset! Thus, the ratio of the frequencies of various busses is directly connected to the arbitrator’s ability to correctly process the signals under the changing circumstances.
In practice this means that the ratio of PCI Express and FSB busses becomes a crucial thing for good overclocking: increasing the FSB clock rate alone we bring some misbalance into the system, rendering it inoperative. So, our overclocking recipe is simple in this case: we must increase the PCI Express frequency, too!
And really, setting 105MHz PCI Express frequency on the reviewed mainboard we managed to achieve 225MHz FSB frequency, keeping the system stable. At 110MHz PCI Express, the maximum stable FSB frequency was 238MHz and so on. That’s already better than those 217 megahertz of the FSB clock rate we got to at first.
Cutting the long story short, we followed our overclocking recipe to get to the maximum stable FSB frequency, which we found at 248MHz.
The frequency of the PCI Express bus was 115MHz at that, but our PCI Express graphics card, a Tul (PowerColor) X800 XT, remained perfectly stable.