Before I get to the benchmarks, a couple things of interest:
1) RAM
I originally had the OCZ Gold VX dimms in this board but couldn't get it stable at much past 210fsb regardless of multiplier. I realize now that I was too quick on the trigger finger when I bought that RAM; I didn't realize that it needed higher voltages than this board could supply (but I hear it's very popular in the DFI board, which can go significantly above 3.0 on the vdimm). I know there's a vdimm mod you can do on the MSI board, but that's a little more than I'm prepared to bite off. Instead, I switched to the G.Skill PC4400 LE stuff and it's been amazing.
2) Temps
Before I bought this mb, I was running on an Asus A8V (AGP) board. I had a heat problem (according to BIOS, anyhow -- I never used an external thermometer to verify..and I know MB thermos are notoriously inaccurate) because I was seeing mid 50's idle and into the low 60's under load. I tried the Thermaltake SilentBoost HSF, but it performed worse than the stock HSF. I then went to the Thermalright XP-120 (close fit) and it helped out, dropping the idle temps to around 50, and load temps to the high 50's. I was using the "half grain of rice" AS5 appplication method and was religious about cleaning the HSF and CPU surfaces with 90% iso alcohol.
When I switched to the MSI, I also switched to the Thermalright XP-90c all copper HSF. I mated that with a medium speed panaflo fan, and the temps I get from MSI corecenter are low 40's idle, low 50's underload. The highest I've ever seen is 55C, which was running /w max vcore a couple hours into a Prime95 blended torture test.
Like I said above, I know the MB temp sensors are very suspect, so I'll try to take some DMM temp readings this weekend to collect various temp readings. I'll post them as a follow-up. I gotta say, however, I really like the XP-90c. Especially after that "adventure" mounting the XP-120 on the Asus board, this thing was a piece of cake.
3) Stability
As I said in the intro, stability is critical. I use Prime95 overnight to verify stability (or lack thereof). I have found that SuperPI gives a good (and faster) indication of when you've gone "clearly too far" because he reports rounding errors, etc. After getting him working, however, I like a nice long Prime95 run to make sure things are kosher.
4) OC'ing
A couple things here. I've been doing the PC thing since around '85, and did the Coleco Adam thing before that. I was really big into mucking /w hardware back then, but took something of a hiatas from since the late 90's. I recently decided to get back into things, so I had to relearn all the current tech. I still have a LOT to learn, however, but I've done a lot of RTFF'ing and forum lurking over the last couple of weeks to educate myself.
That said, my OC'ing so far is limited to mutiplier modification, FSB modification, and use of the RAM dividers. I switched the RAM to 1T, but I haven't otherwise messed with its timings/etc. I imagine I'll get into that one of these days. I'm also going to be staying on air for the time being...but you know how these things tend to snowball. ;-)
Now, for some benchmarks