The second peculiarity, and quite a useful one for the end-user, is the info about the voltage increase relative to the nominal. It looks like that: if the voltage is set a little (I suppose, by 10%) above the nominal, the option’s name is displayed in white. Like that:
If the voltage goes higher above the nominal, its name becomes yellow:
And if the voltage is set extremely high (in the mainboard’s opinion), the name is given in red:
This is no doubt a useful feature, especially for beginners. Even experienced overclockers may be pleased since the mainboard prompts the “safe” voltage range (if you are too lazy to do the math1 of adding 10% to the nominal in your mind).
Now, we are in the DRAM Timings Control. There are no significant changes compared to the BIOS of the KT4A Ultra. The range of available settings is still very wide:
This is all about the BIOS of the reviewed mainboard. Let’s get to the tests. Overclocking comes first.
I used the methodology explained in the ASUS A7V600 review. The idea is simple. I increase the Vcore to 1.85V, the FSB frequency to 166MHz and the multiplier to 13.5x. Thus, we get a 2250MHz processor and see if it works right. However, this simple method failed in the present case. The mainboard simply doesn’t support 13.5x multiplier. It offers a joint 12.5/13.0x and 14.0x, but none of 13.5x. Well, I set 204MHz FSB and 11x multiplier, thus achieving about 2250MHz. This combination didn’t work. The mainboard didn’t even try to start up. Well, it tried, but looped up (according to the D-LED) at an early POST phase. Thinking this might have been a processor problem, I replaced it with an Athlon XP 3000+ (166MHz bus) and ran it at its regular frequency of 2167MHz and at 1.85V voltage. The mainboard started up, started booting up Windows and then re-started. Seems like we have found the root of all evil – the CPU voltage regulator circuit doesn’t stand this voltage at all. I set 1.75V. The mainboard starts up all right and is quite stable. Well, I wonder why MSI offers the upper Vcore limit as 2.3V? Just for fun? Regrettably, the mainboard doesn’t pass our first stability test.
Now, let’s try to overclock the FSB. The ASUS board showed an average result of 207MHz, maybe the MSI is capable of something better? So, we reduce the CPU multiplier to 8x, increase the timings of the Corsair XMS3200C2 memory to 2.5-3-3-7 (with such timings and 2.7V voltage this memory is operational up to a frequency of about 225MHz), drop the AGP speed to 4x (I had to do this through ATI Control Panel, as the BIOS said the meaningful “Auto” instead of the AGP speed) and increase the FSB frequency, starting from 200MHz. The stability was tested in 3DMark2001 SE (three times), four standard demos from Unreal Tournament 2003 Demo and SPECviewperf 7.1. To double-check the system stability I also launched the distributed calculations client aka Find-a-drug to load the CPU to the maximum. I guess this check is more than comprehensive.
I won’t go deep into detail. In short, the maximum frequency the mainboard notched was 214MHz. It’s better than what we obtained on ASUS board, but worse than nForce2-based mainboards do, for example, DFI LAN PARTY NFII Ultra. Well, the AGP and PCI frequencies depend on the FSB frequency in Socket A mainboards on VIA chipsets. Thus, it would be unreasonable to hope for a really high result.