Yep, your memory could be the issue with your OC. Being PC3200 means that it runs @ 200 Mhz (400 DDR) by default 220 FSB=10% OC of memory. Since running memory asynchronously is not a huge penalty on Athlon64's as opposed to P4's, some of us with limited budgets are using the asynch route for OC'ing. The reason that your scores did not change much is that some benchmarks are more sensitive to memory speed and some are more sensitive the the actual clockspeed of the CPU. By lowering your memory speed and increasing your CPU speed as you have in your experience (if you further up your FSB, memory will return to baseline since it is not fixed at 333, but starts @ 333 if FSB=400) you have effectively come up with different way to make your system perform like it would have if left stock (in that particular application).
The other things I'm noticing with your setup are these:
1. You have new memory coming-(I'm assuming faster memory) This may allow you to overclock in synch with FSB which will give you a benefit in most if not all apps.
2. You started out with a 3500+, which is damn fast to begin with, so if you can OC that by 10% as you have, that's nothing to scoff at. Reaching higher will not give you drastically different results, but IMHO it's always worth getting the most out of what you have.
3. If you're only benchmarking video-based apps like AQM then you'll notice a more recognizable difference by OC'ing your video card if you haven't done that already.
4. Get a copy of Prime95 (google it) and use the blended test for at least 4 hrs. (many run it for 24+) to check for stability of overclocks. Be sure to keep and eye on temps when you do this, because it is a very taxing application and it'll heat up your CPU quickly if you don't have enough cooling. (The stock cooler is pretty good, but not great).
Facebook
Twitter