Monday, January 02, 2006

Article: A Tale of Non-Overclockable Boards turned Overclockable

Oftentimes, for enthusiasts who overclock, they base their purchase on reviews and how motherboard overclocks. If a certain motherboard, reviewed at a "respectable" and "huge" site, failed to overclock, it is dismissed as that: board is not "overclocker-friendly".


But what is "overclocker-friendly"? For most overclockers I know, it can be the options in the BIOS where users can tweak everything, from stepless frequency settings, to fine tuning RAM ratio/timing, to changing voltage settings on the fly, to having option of adjusting processor multiplier. There are many motherboards that indeed offer such options, while some, just to be called "OC friendly" do have these "features" but these options aren't really that "great".

Take a look for example, in the case of a known great Socket478 motherboard: Asus P4C800. Even if this board offers great flexibility in terms of "tweakable" options, it offers mediocre choices in the voltage options, specially, in the RAM area. In the end, this board suffered various hardware modification, from droop mod, to vDimm mod.

II. Overclock-friendly Motherboard?

Now, what's my point in all this? Simply, focusing on the term "overclock" friendly. If a certain motherboard has not much option, say, less voltage control compared to another motherboard that has many, and produce the same results, which can be called "overclocker" friendly? It can be said that both isn't, but look at how Asus P4C800 was the boon of Intel® s478 users, which needs lots of modification to make it a "great" OCing board, and if not, it's much like another board with less OC options.

So saying both isn't OC friendly will not be fitting since Asus proved that all it needs are several hardware modding. So, can it be said then that both as OC friendly? Maybe, assuming both boards are hardware modded to perform just the same. Now, some people will say that, what about voltage stability design? Well, look at server motherboards, they have great voltage stability, yet, most of them aren't really geared as a good OC-friendly board. In the end, imo, if you have a board that can overclock pretty well, no matter what the options there are in the BIOS, then I can call it OC friendly.

III. Proof Of Concept

Intel® D875PBZ

A simple Google of this board and you'll find out that this board is like a curse to overclockers. While the default BIOS can give as much as 4% OC, there isn't really much offered by this motherboard in terms of OC options. The Intel® Desktop Control software gives the same options: 4% max. However, many failed to see the gem on this board, as is evident by comments from various web reviews:

Every great board has a few needy areas though, with the D875PBZ being no exception. The biggest area of concern is the lack of FSB overclocking available with this board.

In the same terms you should regard the "overclocking" options integrated in the Intel® D875PBZ board. For example, increase of memory voltage and adjustable latency timings, which gives the system integrators increase the list of memory types used. But the possibility to raise the processor clock speed is not more than an extra load for various stability tests.

And there other similar outcome from Motherboards.Org, SharkyExtreme, HardAvenue, and CyberCPU.

However, this same motherboard hits 300FSB, which, as far as my fragile mind can recall, was beyond the Holy Grail of s478 overclocking, which was hitting around 250FSB back then, 260FSB-280FSB as average highest, and only a handful can even get past 285FSB. So, is an Intel® D875PBZ overclocker-friendly or not?

Intel® D925XECB2

This is another great example of a motherboard, which isn't blessed with many high-reviews re OCability, and yet can also be overclocked. As is the case with an Intel® D875PBZ, I worked with abo
to develop a tool for tweaking the FSB. In December 31, 2005, I started playing with the motherboard, and found that it has options for at most, 10%, OC. I consulted this with abo (SetFSB developer) on January 01, 2006 and by January 02, 2006, he already developed a working tool to tweak it. He spent most of the time developing, while I spent most of my time to actually test it. The results speaks for themselves in this CPU-Z validation link.

As of the time of this writing, we are is still testing the tool, but you can visit his site and check the update which pertains to the IDT CV1125 Clock generator. I would like to give a shout-out to go visit his site and check if he has something for your motherboard. Chances are, he has it.

Other Proof of Concepts

Other motherboards, such as the MSI 865PE-Neo2-P and EPoX EP-5EGA+, didn't get a good review for its OCability. But my experience with them here (registration required) and here, shows that they can OC really well.

IV. Closing thoughts...

Looking back at the term "overclocker friendly" board, I can safely assume that any board, that can be modded via hardware or software, can be called overclocker friendly. A motherboard, that offers options, and yet can't even live up to the stability and extent of its options, can't be said "overclocker friendly". I mean, come on, you have the options, and yet, can't even use it well. What good these options are if it's next to being useless. Having a useless option can be equated to having no option at all. Of course, I never said and never meant that a board that has options and can live up to its options isn't overclocker friendly, they are just uber friendly =)

V. Into the future...

I plan to play with Intel® Desktop Board D975XBX, another board that never see good OC reviews and check how it can be tweaked with FSB adjustments. And if abo is still up for it, I'll request for an Intel® Desktop Board D975XBX OC utility and work with him in testing and ensuring the tool works well =)....

Till then, happy OCing...

All materials ©2005 by DM, reproduction by permission only

No comments: