DFI LANPARTY UT NF680i LT SLI-T2R mainboard is claimed to be designed for extreme overclocking. It is really convincing, taking into account the richest theoretical features we have just discussed in the previous parts of this article. Maybe it will help us hit unprecedented heights of extreme overclocking, but our experiments proved that it doesn’t quite suit for traditional overclocking attempts.
Some time ago I reviewed abit IP35 Pro mainboard, which proved ideally suited for overclocking. Since I already knew what Intel Core 2 Duo E6300 processor was capable of, I set the FSB frequency to 490MHz right away, raised the processor Vcore, Vmem and North Bridge voltage a little bit. It took me a few minutes to do all this. After that we checked the system stability and performance in different applications, and that was it.
I tried to perform the same on DFI LANPARTY UT NF680i LT SLI-T2R mainboard this time, but it took me not just a few hours, but even a few days to finally get through with it.
We performed our tests in an open testbed of the following configuration:
- DFI LANPARTY UT NF680i LT SLI-T2R mainboard, rev. A1, BIOS 05.21.07;
- Intel Core 2 Duo E6300 CPU (1.86GHz, FSB 266MHz, Conroe-2M, rev. B2);
- 2 x 1024MB Corsair Dominator TWIN2X2048-9136C5D;
- NVIDIA GeForce 8800 GTS 320MB graphics card;
- Seagate Barracuda 7200.10, ST3320620AS HDD, 7200rpm, 16MB, SATA 320GB;
- Scythe Mine CPU cooler;
- OCZ GameXStream GXS700 PSU (700W).
I increased the processor voltage to 1.45V, memory voltage to 2.18V and set synchronous memory work mode. However, the board wouldn’t boot at 450MHz FSB. Then I decided to take it easy and start with 400MHz FSB, because most mainboards work stably at this frequency without any additional voltage changes. Not the DFI board: it froze dead during POST. The board didn’t react to processor Vcore increase, but started working fine as soon as I pushed up the North Bridge voltage setting.
So, looked like overclocking success depended a lot on that particular parameter. I tried 450MHz FSB again – no luck. 425MHz – yes, success! However the board couldn’t perform the stability test at this speed even for a minute. I kept increasing the chipset North Bridge voltage step by step until I moved from the nominal 1.3V to 1.51V. This is when I managed to get the board run stably in OCCT for 56 second. If the NBcore was set to 1.54V or higher, the board shut down immediately. The changes in any other voltages or corresponding parameters (such as CPU Drive Strength, for example) didn’t help, too.
The board didn’t restart after over-overclocking and I had to use the jumper to get back to the previous operational configuration. Sometimes, it would start, though, and I could see all keyboard LED indicators blinking very fast up until Windows started booting. However, the keyboard didn’t work in Windows and didn’t work before that. And the mouse didn’t work either. However, since POST had been passed, the wrong setting got recorded in CMOS, so the jumper couldn’t help any more. To clear CMOS in this case I had to either use the other jumper or simultaneously press Power On and Reset buttons. In the latter case, unlike using the jumper, the power shouldn’t be cut off: the board should keep receiving default power.
I used the buttons instead of the jumper once. CMOS did get cleared, the mainboard stopped on boot-up offering to access the BIOS and correct the settings or continue booting with the defaults. Unfortunately, the keyboard was still blinking with LED indicators but didn’t work, so later on I only used the jumper, because thanks to CMOS Reloaded I had a few profiles saved for situations like that.
I was pretty sure that the board will work just fine at 412.5MHz. This strange value comes from the fact that Nvidia based mainboards for Intel processors set the FSB in quadrupled values, and I was changing it with 50MHz increment. For example, for 425MHz it equals 1700MHz QDR. If I lower this value by 50MHz and set it to 1650MHz in the BIOS, we will get this “non-rounded” value of 412.5MHz. Later in this article I will lose the decimals.
Well, my hopes turned out vain. The mainboard worked for a few minutes and even a few tens of minutes at 412MHz FSB, but in the end it would inevitably freeze, or the OCCT utility would report errors. In despair I tried to get beyond the non-operational frequency interval by setting 475MHz FSB (1900MHz QDR). I picked this frequency at random, but my guess was a success as we will see later on. I would have failed at anything above or below that, but at 475MHz DFI LANPARTY UT NF680i LT SLI-T2R passed the short 15-minute OCCT test for the first time!
You may think that 475MHz FSB (1900MHz QDR) is only one step away from the 490MHz maximum for our test processor, but this part of our experiment also took a lot of time to complete. No matter what I did, what voltage settings I chose – the board just wouldn’t work.