What started out as a simple motherboard and processor upgrade in an Antec NSK 3480 case turned into a major project due to the immature status of so many socket 1155 motherboards available today. One thing that I have noticed is the choices for enthusiast motherboards in the uATX form factor is far more limited than in ATX. The scene is dominated by H67 and H61 boards intended for office use. This is not a full review. You may refer to the manufacturer's web pages for detailed specifications.

Firs up was the Gigabyte GZ68MA-D2H-B3. My first one was missing the IO grate that fits in the back of the computer so I had to return to Microcenter for an exchange. Once running I thought it was pretty good. The F2 bios shipped with the board was missing several important features like Virtu, Smart Response and overclocking the on chip graphics so I flashed the F3. After that the board would overclock only until it went to sleep. When the machine woke up, it was set to a 33 multiplier, stock for my i5-2500K. I reverted to the F2 bios and the problem went away. Like the moth and the flame I tried this again a few days later and the loss of an overclocked multiplier returned. This time flashing back to F2 did not fix things, not did removing the CMOS battery. Epic Fail.

This Gigabyte board appears to lack LLC or Vdroop control. It does have a dynamic voltage offset which allows the user to increase or reduce the voltage applied by the Auto Vcore mode. I did not verify if the board could be undervolted below 1.2V, but it looks that way.

In all fairness to Gigabyte, I did not contact their technical support nor do I know if this problem is isolated or widespread. If you don't overclock I suppose this board is OK and eventually the bios will mature. Check carefully, as this board has limited analog audio connections.

When I brought the Gigabyte board back the only other micro ATX sandy bridge board available in either P67 or Z68 was an Intel DP67DE. It was the only board with eSATA ports, but that left it with only 4 SATA ports. It did not look very well built. On the box it said that processor speed adjustments were possible. When overclocked this board would throttle back to a 33 multiplier within about two minutes. This problem is not isolated. I found a thread about it. Intel has since redesigned this board to go from a 3 phase to a 4 phase power supply. This board may not be undervolted. The IO back plate was badly made and difficult to install.

If you don't overclock and absolutely must have the Intel networking chip, this board is a possibility. I was disappointed and rather annoyed as I had now been through two system builds. The board was exchanged. At Microcenter, they are very nice about this.

The final try, and it appears to be a keeper, is an MSI Z68MA-ED55. It's more expensive than the other two. I was lucky this board showed up in Microcenter's inventory on Saturday night. This board has the new style EFI bios which takes a bit of getting used to. It is very easy to overclock. I have mine running at 4.3 Ghz with auto voltage. This was tested with prime 95 for 2 hours and Intel Burn Test for 20 minutes. The later benchmark will draw an additional 30 watts AC and run 8 C hotter than prime 95. It will run at 4.4 Ghz with this setting too. Both are with base multiplier changes and turbo disabled. When switching to a user specified voltage there will be no voltage drop at idle and no increase if turbo is used. It looks to me the board could be undervolted, subject to these limitations. It could probably be run faster than 4.4 but will idle hot. I think around 4.3 Ghz is the limit with quiet cooling or 4.2 if one is conservative.

Any questions?

Just one question. Why?

Seriously...the CPU maker goes to a lot of trouble to determine the maximum safe clock. Why mess with it? What's the point?

Well there's one clear error here. The cpu maker does not determine maximum safe clock. They determine some sort of clock they want to sell chips at along with some binning. Otherwise chips would not be easily overclocked by 25% or more. Or the fact intel makes K-series chips that only make sense to buy IF you overclock. And the point is simple, speed. What other point is there?

You have no way of knowing whether the chip you buy was factory underclocked to meet a marketing demand or not. For obvious reasons, the CPU maker isn't going to tell you.

Speed for what? What can you do with a 4Ghz CPU that you can't do with a 3.8Ghz CPU?

A lot has been written about overclocking. The short explanation is that CPU speeds are rated for the cheap motherboards that Dell and HP make. the speeds will go much higher when the motherboard has the ability to deliver power to the processor. On the motherboard there are voltage regulators that convert 5 v to the CPU voltage of around 1.3 v. What I encountered with the Intel motherboard was a level of engineering and construction about the same as what would be found in some cheap desktop from Dell or HP.

What can you do? When editing photos it is important for the image to update as the various slider controls in Photoshop (or whatever) are changed. The lags really kill productivity. There are all sorts of CPU intensive tasks in media creation, especially video. It is a pretty big jump to go from 3.3 Ghz to 4.3 Ghz. It's like the difference between a Camry and a Corvette. Like the man said, why is Intel selling K series chips with an unlocked multiplier at a premium price?

Just wait until next year when Ivy Bridge is out with the 3D transistors, 22nm, and another 35% performance boost. Those babies are going to be crazy fast and not use a lot of power.

What cpu heat sink are you using?

CoolerMaster Hyper 212 with a Scythe fan turning at 850 RPM.