Well all that synthetic stuff might not be ya cup of tea. Some might buy these boards, just to build the ultimate game platform with high quality components. Yep some of the most expensive boards find their way into a 24/7 platform. Some buy boards for the looks, some for the onboard features,...

Conducting the game tests at 1600 at 1200 resolution, the same resolution I use daily for my online gaming experience. My GTX 285 got no problem to power the two games into what I call consider the jerk-free plus 60fps zone.

We took Resident Evil and Far Cry 2 due to their excellent build in game benchmarks. The zombie slaughter fest Resident Evil 5 is up first. MSI's Xpower being the top performer, though between first and 4th place there's only half a FPS difference. Which is in fact negligible, some of the previous synthetic tests might have indicated a worthwhile difference for the benching crowd. For the 24/7 user, buy whatever suits ya needs best.

Far Cry 2 somehow seems to favour the Asus boards. Don't ask me why, but these are the average results out of 3 runs. With each run executed right after a reboot. Plus minus one FPS difference lads, that's all there's too it. Shocked maybe ?

Maximum Bclock Test From BIOS

Let's move onto the infamous Bclock test. I had one Bloomfield i950 and thanks to the Tones a brand new i970 ES Gulftown sample. Well lately we see a lot of hype created around the Bclock performance of these boards. Nobody is surprised anymore when we see a +250 Bclock screenshot. But usually this is at the cost of 3D performance as mostly Slow Mode is enabled. 3D performance gets crippled, though CPU performance remains pretty on par. Below are the results of the i970, cooled by the Corsair H70. Why not also the 950 results you may ask ? My Bloomie was far from brilliant and walled at 228-229 Bclock ( even under phase or LN2 ) on the REX III and UD9.

235 bclock with QPI way over 4000Mhz to me sounds great. Right up the alley where you think these boards should be. Though the MSI XPower walls hard at 220Bclock. There's an issue with this board a it refuses to post anything higher. By using soft reboots (so no uncore or touching the ram settings or dividers) you can work your way up to 230. However a lockup or crash would mean you would have to start again from scratch at 220Bclock or lower. This is no problem for benchers, most will safe a 200-220 bclock profile, reboot, gradually up the bclock in the bios, save and reboot and voila eg 230 Bclock at your disposal. For non extreme CPU users this might be a big culprit and a definite no buy reason.

The REX III and UD9 are head to head here and pushing this particular CPU to its limits. Sadly we experienced lesser stellar boards of the same type. Our preview UD9 sample ( UD7P ) was limited at 225. The retail sample was far better. Tim Hendley of Gigabyte, told us at the GOOC 2010 press meet that all IOH chips on the UD9 are binned for maximum Bclock action. (how they do the binning is unknown) Now this is what we would like to see for all high end motherboards. Sad news however for the REX III fans. Madshrimps worked with a 215 bclock walling sample. This low performance should be avoided at all costs. The buyer should somehow get a warranty that this high end board must outperform any lower classed/specced board.

For those that got no clue what Slow Mode means for ya 3D performance let me show ya 2 screenies :

Pretty abominal score isn't it ? Even the screen build up in windows is tearing. So please keep this in mind when looking at some impressive bclock numbers. Don't get caught in the hype and closely watch them QPI speeds plz.

Maximum Memory Frequency Test

The rams for the max frequency test are Corsairs 6GB triple GTX 2 kit. Rated at 2250Mhz CL8-8-8-24 they provide a challenge for all X58 motherboards and CPU's. Our test CPU was put under a Phase unit for this test as breaking 4000Mhz uncore with just air cooling is rough. The rams had to complete several 3DMark06 and Superpi 32M tests, before we called the clocks stable.

Both the ASUS and Gigabyte UD9 had no issues to reach the rated GTX2 speeds. The uncore for the Bloomfield CPU was at +4500Mhz during the 2260Mhz test. The Xpower board shows another of its weak spots here. While it perfectly ran though the 3DMark06 test, this without breaking a sweat. It failed epically in Superpi32m. Backing down memory speed to 1980Mhz finally gave us an error free Superpi 32M run.

The MSI board got issues to run 2000mhz ram speed with a high uncore on Bloomfield CPU's. When extreme cooling (read LN2) is used on the CPU the board is able break the 2000mhz barrier too. For daily Bloomfield users with high speed rams, this can be seen as a big issue. For those that use lower speed specced rams there's no problem at all. Massman is currently testing new biosses and giving feedback to the MSI engineers. Hopefully they can get the board up to the expected level.

2260Mhz was what we reached at the rated timings and Vdimm voltage. With some more Vdimm and a bit looser (sub)timings there was more to go for...

Cooling: Temperature of the IOH Chip

Last but not least we tested the heatpipe combinations. Being very sceptical of the MSI as it seems to be completely under dimensioned compared to its two rivals. MSI also opted not to link the IOH and ICH via a heatpipe. Personally I find this a very good solution as the most heat is put out via the IOH. This MSI solution also isolates the heat. An isolated active cooled IOH solution would be the best solution to keep its temps low.

We tested all the boards during a 20minute LinX session at 200Bclock, IOH set voltage of 1.2. No extra cooling besides the CPU being cooled by a Thermalright VenomousX. A temp probe was stuck next to the IOH, as neither the Gigabyte nor the MSI have got temp sensors. The Asus board was retested with the active fan solution, the Gigabyte UD9 with the Silent Heatpipe and with water-cooling!

First up, I redid the thermal paste of the Rampage III Extreme. With the stock TIM, temps were over 75°C at bootup. Very bad contact was the cause of these elevated temperatures. I gained 9°C by just using OCZ Freeze! cooling paste. Via mounting the included active fan solution I gained another 4°C, just beating the humble MSI Xpower heatsink. The UD9 at stock is the hottest of them all. But think of it, this board has got 2 extra hotheads to cool : the Nforce200 chipsets. Once we mounted the Silent Heatpipe, which got some extra airflow from the CPU cooler, the IOH temps dropped to 60°C. This without any added noise. Including the foreseen waterblock into my WC loop the IOH temps settled at 58°C. Not bad, but if the contact was directly on the IOH, the temps would have been even lower. Madshrimps received a full cover waterblock for the REX III and temps are lower than 40°C on the IOH and ICH. (But that's for another review.)