Well, the first thing I did when I removed the heatsink was rip off the stupid aluminum pad + some light lapping with 2000grit, so I have no idea how it would of done without lapping. One thing I did notice is that there isn't enough pressure being applied to the CPU. After you fasten the heatsink as firm as you can, you could still wiggle it a bit because the springs aren't applying enough downward force. I'm looking to buy some washers to slip between the springs and the screw to make them compress more... From my experience, that should lower temps another 2C or so.
As for the ramsink, right now it's a bit too high off the heatpipe, so it creates a slight 'bulge' underneath the keyboard. This gave me an idea, I'm going to sand the fins of the ramsink down until it just touches the underside of the keyboard. But, since the keyboard base is made of metal - it conducts heat. I'll then rip off a part of the plastic protecting the base of the keyboard and apply some AS5 ontop of the fins of the ramsink. Voila, the keyboard doubles as a heat spreader
Another idea: since there is a gap between the fins of the CPU heatsink and the exhaust port, why not solder some extra fins onto the heatpipe and fill that gap?
1.564v seems to be the sweet spot for me, it primed small FFTs for over an hour before I stopped it, temp stabilized @ 70C with the fan on full. Thank you so much for the Voltage vid chart, that made things a lot easier to look up.
Take a look @ your 755, what does the batch code start with? I assume it's because the CPU is a lot older, maybe a week 20 of 2004 or earlier, or even a 2003 chip. Usually, the newer the CPU, the better it OC's. I think you should purchase a newer CPU, retail and give it another go, it might do 2.66Ghz stable with only 1.5v, since my chip isn't that new anyways, and it's an OEM, retail chips are usually better as well. If I had the cash, I would of picked up a 755 retail, Malaysia, week 35 of 2005. I got my chip off ebay for $235 so I guess I can't complain.
Originally Posted by lucidchaos
You are a brave soul. Sounds like you are already having better luck than my attempt. I didn't get 30 minutes out of mine until I grounded both vid4 and vid1. Try doing a custom test with nothing but 12k FFTs. I ended up stable at EVERYTHING but 12k FFT testing at the prime of my 2.66 overclock. (vid4, 2, and 0 @ 76*C if I recall correctly)
I'll be back to track your results. If the lapping and ramsink helped a lot, I will try putting the ol' 755 back in mine and give it another whirl.
I found out a fact you may want to consider a couple months after my 2.0 stock > 2.66 hardware > 2.53 software overclock. I had it rock solid stable, maxxing out at 76*c when priming it during testing. A couple months later, the dust accumulating on the HSF caused my temp to go up a couple of degrees. I blame that temperature increase to my breakdown of stability. A simple can of air fixed my problem; however I wasn't comfortable knowing that my system isn't stable after getting two degrees hotter. :P I ended up taking the 755 out and put a 745 in. (and to my disgust, it got nearly just as hot and ran exactly the same as the 755... not stable with just a vid4 mod. I think I got a junk 745.) It is voltmodded with vid4 and vid1 and running stable at a maximum of 76*C after a couple of months of dust accumulation.
Let me know if you have any questions, I haven't been checking the forums lately but I am genuinely interested in helping you make your mod work so I'll be back to check your progress.
PS. I actually had SetFSB working on my 9300... just not very well. I took it from 2.66 to 2.68 and was stable enough to run a 1M superpi (30s w00t!). Since you seem to know your stuff maybe you might know why I can't get any better speeds with setfsb. I tried locking PCI to every value with no luck.