All benchmarks below were run with the CPU at 3.0GHz and RAM at 1066MHz.
I used FutureMark's 3DMark06 and a few tests from SiSoft Sandra 2007 to see how the motherboard performed in synthetic tests. All of these benchmarks were harvested with the CPU and memory running at their overclocked speeds. While not directly comparable, I did this to see what sort of gains were had when compared to results I had received previously with the same CPU and memory at stock speeds on the ASUS P5N32-SLI Premium nForce 590 motherboard.
My previous 3DMark06 score in the EVGA GeForce 8800 GTS SLI review was 10618. Thanks to the increase in CPU and memory speed, I've tacked an additional 2,066 points onto my score. I tested the CPU twice in Sandra 2007 using both the multimedia and arithmetic tests. The memory was tested with their memory bandwidth benchmark. I tried to keep the same computers in all three of the benchmarks, but in some cases, various setups weren't available.
SiSoft Sandra 2007 Memory Bandwidth
It was somewhat disappointing that Sandra does not have any memory faster than PC2 5300 (DDR2 667) considering that at 1066MHz, the Super Talent memory has an effective bandwidth of 8500MB/s. It is also slightly disappointing that the memory topped out at only 6893MB/s, but it is still quite a step up from where it had been.
SiSoft Sandra 2007 CPU Arithmetic
Considering that the X6800 and my E6600 clocked at 3.0GHz are essentially the same CPU, it's not surprising at all that mine pulled ahead slightly due to the increase in clock speeds. The advantage was slight, but it still made me smile in nerdy joy that my "mid-range" CPU is faster than the $900+ top-of-the-line dual core Intel offering.
SiSoft Sandra 2007 CPU Multimedia
Once again, it is no surprise the overclocked E6600 pulled ahead of all the rest of the CPUs. As awesome as it is to have the fastest CPU my copy of Sandra has ever seen, it is important to keep in mind these are just synthetic benchmarks and not completely indicative of real-world performance. Regardless, the fact that this board remained so stable and the CPU temperature remained at 60°C or below left me very pleased and very excited for a future of even faster gaming than anything I have previously experienced.
F.E.A.R. is likely my favorite game to benchmark thanks to the incredibly handy built-in time demo. Besides that, the game looks phenomenal at 2560x1600 with settings completely maxed out. To see how much of a difference an additional 600MHz on the CPU and 266MHz on the RAM made, I compared the results of this run-through to that of my results on the ASUS P5N32-Premium nForce 590 SLI with the EVGA GeForce 8800GTS SLI setup. The top results are with the CPU and RAM at stock speeds and the second set are overclocked.
I was disappointed the results weren't that much greater than the stock speeds, but I saw an increase in minimum framerates across the board which is always welcome.
Thanks to the embargo on this review being pushed back 11 days, I was able to get S.T.A.L.K.E.R. and benchmark it for this review.
For those who still haven't picked up a copy, the oft-delayed S.T.A.L.K.E.R. is well worth the time its long development cycle took. The graphics are stunning, the atmosphere is creepy and realistic and most importantly: it is fun! I used the area right after emerging from the Underground to benchmark. The area is densely packed with lots of houses, guard towers and enemies. I've been playing the game at 2560x1600 with no antialiasing enabled, but full dynamic lighting (and AA isn't allowed in that mode) on. Other than that, all settings are maxed except for grass shadows which I had to disable due to continuous compounded performance degradation. I would have to close to the desktop and restart the game after a few savegame loads (I die a lot in this game) because the game would chug so badly.
This graph shows second-by-second performance of a 60-second run-through of the aforementioned game area.
Performance was very good with the average hovering around 35 frames-per-second. This was during combat with 9 enemies around and firing almost constantly, running in and out of buildings and into foliage cover. Considering how good this game looks, I'm surprised the scores were this high.