Soltek QBIC EQ3801: nForce3 250Gb goes SFF
by Evan Lieb on July 4, 2004 10:35 AM EST- Posted in
- Systems
Soltek EQ3801: BIOS
As with the vast majority of motherboards in the world, the Soltek B9A-FGR motherboard utilizes the Phoenix Award BIOS.
The EQ3801 allows surprisingly low Vcore adjustments in its BIOS. Normally, 1.70V or so is the norm for an enthusiast desktop motherboard, but not the EQ3801's B9A-FGR motherboard. This motherboard's BIOS only allows adjustments up to 1.55V. And even when Vcore is set to 1.55V, the actual operating voltage is lower, between 1.52V and 1.55V on average. Hopefully, Soltek can remedy this and release a BIOS with better Vcore adjustments for overclockers. Though, as we've seen before, ultra high Vcore adjustments actually don't make too much of a difference with Athlon 64 processors.
The EQ3801 BIOS also allows AGP voltage adjustments up to 1.8V in 0.1V increments, as shown above. VAGP adjustments aren't very useful beyond extreme video card overclocking, and even then, its use is mostly minimal. Anyway, despite the fact that the default AGP spec calls for operating voltage of 1.5V, we suggest 1.6V for maximum stability. In fact, some motherboards already set VAGP to 1.6V manually, including the EQ3801's motherboard.
Another AGP-related adjustment, AGP bus speed, is available in the EQ3801 BIOS, up to 100MHz in 1MHz increments. We highly doubt that you'll be able to reach this type of speed, even with separate hardware voltage mods, but speeds between 75-80MHz AGP are quite attainable on stock settings/hardware on a regular basis. Unfortunately, because there are no PCI bus adjustments, overclocking your AGP bus will also similarly increase your PCI bus' speed. Knowing that PCI speeds of over 40MHz are hard to attain reliably, this will inevitably make significant gains in AGP exceedingly difficult. Though, if you're just a FSB overclocker, the nForce3 250Gb's AGP/PCI lock will be much more useful.
As shown above, VDIMM adjustments aren't terribly spectacular. Even though adjustments available up to 2.8V in 0.1V increments aren't poor, it's fairly average nowadays. Still, if you have little use for memory overclocking, your DDR module's default operating voltage of 2.5V will more than suffice. However, setting VDIMM to 2.6V isn't a bad idea, if you'd like to take advantage of your memory module's lowest possible memory timings; the extra voltage aids in keeping stable low latencies, like CAS 2-2-2-6 (though, those low latencies would probably require BH5 chips, which are all but extinct, save for the promising new Samsung chips).
Arguably, the most significant overclocking feature of a motherboard is its FSB adjustments. However, to be absolutely clear, Athlon 64 systems don't really have a FSB. There is no such thing as a FSB in this type of system because the memory controller has moved from the North Bridge onto the Athlon 64 CPU itself. Therefore, technically, the bus that you're overclocking is the HT (HyperTransport) bus, AMD's name for their Athlon 64 (a.k.a. K8 Hammer) bus. Still, many manufacturers refer to the Athlon 64's HT bus as FSB, even if they know it's technically inaccurate. But that doesn't change the fact that you can adjust an HT bus exactly like you can a FSB, such as those found on Athlon XP and Pentium 4 processors. In the EQ3801's case, you can adjust the Athlon 64's bus up to 250MHz. A 250MHz FSB is a noticeably lower speed than what is available on retail nForce3 desktop motherboards and, as you'll see in the overclocking section of this review, it becomes a limiting factor for the enthusiast overclocker. UPDATE 7/7/2004: Soltek has just released a beta BIOS that allows FSB adjustments up to 300MHz. It will be available on their web site shortly. In addition, Soltek is now bundling AMD's cool 'n' quiet driver on their motherboard CD.
22 Comments
View All Comments
Swaid - Sunday, July 4, 2004 - link
The graphs for most of your benchmark only display one motherboard and leave out the rest, so its pretty much impossible to distinguish what the numbers corrispond to.Z80 - Sunday, July 4, 2004 - link
Your encoding benchmarks are not displaying properly