My goal with this board is to find a thoroughly stable 24/7 operating environment with a Q6600 OC'd to 3.8-4.2ghz and with four to eight gigs of RAM. Achieving this is not always a trivial task on Intel-proc nForce boards, especially when paired with a quad core. If the MSI P7N Diamond can do it w/o a struggle, then it'll have an advantage (in my opinion) over the ASUS boards and the 680i/780i reference designs.

Unlike the reference boards, the P7N features all solid capacitors and a better PWM system. This should help to allieviate the RAM and Northbridge burnout issues so prevelant on the EVGA and XFX reference boards. Note that while the non-reference 680i boards had their own host of problems, they didn't have the same burnout issues, likely due to the better materials used.

Unlike the ASUS 780i boards, most early impressions of the P7N are quite positive, noting that the board is easy to work with and OC. The ASUS P5N-T has had some reported compatibilty issues with 45nm processors, while the Striker II has been very hit or miss in supporting even moderate overclocks.

Right now I have the board running on top of its box, with components taken from my main rig. So far, I've been pleased, as setting everything up has been a plug-and-play experience w/o the slightest issue. I started with a never used Q6600 G0 L737B266 proc, 1 DIMM of 2 GIG PC-6400 OCZ SLI RAM, a single 8800 GTX card, a Lite-On IDE DVD+-RW drive, and a SATA WD 250 Gig hard disk. With Windows XP installed, I began adding my other components, and now the motherboard is supporting nearly my entire rig, with the exception of a three extra HDs and another DVD burner.

.. unpacking the box ..



.. everything installed (note the board in the case is an evga 680i, the p7n is on the box) ..



As seen in the pic (or not, thanks to the GTX cards,) a second DIMM of RAM, a second 8800 GTX, and a Creative Labs X-Fi Platinum card have been installed. Again, I've been impressed so far with my experience, as everything works without issue. Even the troublesome X-Fi is behaving perfectly.

Note: The P7N manual states that a 2-card SLI setup must use the blue PCI-E 2.0 slots to function properly. I used one blue slot and one PCI-E 1.1 x16 slot (the white one,) so that I have room to install my PCI X-Fi card, which is better than the included PCI-E card (although the PCI-E card is still very good for included audio.) Note that I had no issue enabling SLI and that it appears to be working properly.

The BIOS is from American Megatrends -- the setup is logical and the feature set is rich, with far more voltages to tweak than either a reference 680i or 780i board. The only real negative I can comment on at this moment is that the BIOS interface lags more than the Award interface on the reference boards. Nothing too bothersome, but still noticeable.

One odd function of this BIOS, however, is how vCore is set. Unlike other voltages in this BIOS and unlike vCore settings in other common BIOSes, you don't set an actual voltage. Instead, you set an allocated increase from the default voltage. This means that when OCing, you must know your chip's VID in order to gauge how your chip compares to others. Unfortunately, this information is not available from the BIOS, as the default VID is represented by the AUTO setting. Therefore, as I OC beyond what my chip can handle at stock voltage, I'll need to do some simple addition to calculate how much actual voltage I'm sending to the chip.

Another thing I've noticed is that your only options for setting RAM speed are Auto, Manual, or Linked. When choosing Linked mode, there does not appear to be a means to choose a divider, and the default divider under Linked mode does not appear to be 1:1 / Synced. When I upped the FSB to 333mhz (represented in the BIOS by a quad-pumped setting of 1333,) Linked mode had my RAM set to 532mhz (266 actual.) To maintain a 1:1 relationship, I had to set RAM speed to Manual and input 667mhz. Now this may be an error that will be fixed in an upcoming BIOS. At worst, all this means at the moment is that some more simple math is needed to keep a 1:1 divider when OCing.

Preliminary OC Results

With the system still running on top of the box and with the stock cooler attached to the CPU, I tried a modest OC to ensure that the board posts correctly on non-stock settings.

I simply bumped the FSB from 1066mhz to 1333mhz for 3.0ghz proc, set RAM to 667mhz for a 1:1 relationship, and set vDIMM to my RAM's stock of 2.0v. I left all other voltages on auto and I left RAM timings to SLI (EPP) optimized settings. Note that when vCore is on Auto, the MSI board will leave it on the processor's default voltage. This is different from the behavior of the reference boards, which try to guesstimate an appropriate voltage for your OC. I must say that I prefer MSI's approach, since a motherboard's guess at vCore is rarely accurate or helpful.

With the FSB set and the RAM voltage set, I saved settings and booted into Windows w/o issue. After a quick 10 minute Prime95 test to ensure that the chip could handle 3ghz on stock voltage, I set out to do some benchmarking. Here are my results:



(Please click above for a larger, readable image)

Certainly not too shabby for a first-run, modest OC.

As for temps, the cores were at what one would expect with the stock cooler. Idles in the 30s, with 10 minutes Prime95 bringing them up into the high 50s, low 60s. Again, 3ghz is likely as far as the stock cooler can handle, so of course I'll be putting this board into my case and hooking the cpu to my water loop for further benching.

What did impress me during this run, however, was the chipset cooling. Note that as of now, an accurate reading of NB and SB temps is not possible. The BIOS only reports a system temp, and neither Everette or any other program I could find is currently configured to read the temps of these specific chips. However, throughout the prime run and the benchmarks, the system temp did not pass 52c (with ambient at 21c.) Furthermore, and most importantly, the chipset cooling pipes remained relatively cool compared to other 680/780 chipset heatsinks/pipes (and good contact between the pipes and the chips has been confirmed.) By relatively cool, I do mean that they were of course warm, but they were certainly not scorching, as is often reported the case on other 780i boards. By another comparison, the pipes were no warmer than the bottom of my Yonah-based Macbook Pro on a typical day, and considerably cooler than my Macbook Pro after intensive use.

So far I've seen one other impression stating that the P7N manages to keep the 780i's SSP and MCP relatively cool, and my preliminary tests do seem to support this. We'll see how things keep as I reach for higher OCs, but so far what I've seen is encouraging, especially knowing that the cooling mechanism is entirely passive.

Next Steps

With my prelim OC out of the way, my next step is to move the board into my case and hook up some beefy cooling. Given my schedule this week, I doubt I'll have new results for a couple of days, but when I have the time to bench, I'll be posting results right here.

Thanks for reading