This is kind of a continuation of this thread, but it pertains more to overclocking the chip than cooling, so I figured it would be a good idea to move it over this way.
I have successfully unlocked the additional two cores, and it remained stable for the half hour that I ran Prime95 before I decided CPUTIN @ 60c (24c ambient) was hotter than I would like it to be. Same with ACC disabled and a 3.8ghz overclock (1.375 vcore) that hit core temps of around 65c. For some reason, at least when I'm running it as a dual core, my CPUTIN is always lower than my core temps, which is why I was concerned with the quad core CPUTIN temps.
With that said, I don't plan on going over 1.45 vcore and 4.2ghz, and I'd say it's more likely that I won't go over 1.4 vcore and 4ghz (if I can make it to that point) with ACC disabled. As a quad, assuming it lasts a full 8 hours of prime at 3.2ghz just so I'm more comfortable with it's unlocked stability, I don't plan taking it over 3.6ghz. It will end up being a pretty light-weight overclock.
I went ahead and purchased a Hyper 212 Plus and got it installed last night. My load temp with the cooler installed at 26c ambient is 42c (dual core, core temp) and my CPUTIN stays around 6c higher than my dual core core temps. For quad, I'm hitting 52c CPUTIN, which is perfectly acceptable for me. I'm idling at 30c across the board. MUCH BETTER than the 45c idle I was seeing with the stock cooler.
I ran Prime95 for 8 hours, and didn't have any cores fail at 3.2ghz, so I think I'm going to creep on up to 3.6ghz and give it a shot at 1.4v.
I have quite a bit of experience overclocking Intel C2 processors, but I'm pretty new to the AMD thing, especially multiplier overclocks. Is there a point where I should kick back the multiplier and adjust the CPU Frequency? Does that offer additional stability at certain levels?
If I hit a wall, or stability just isn't there, what should I look at first? Vcore? NB Voltage?
My experience went like this:
550BE asrock 880G coolmaster Hyper TX3. Processor unlocked to 4 core straight out the box with voltage at stock. The RAM was running slower than it should so I set the RAM speed and timing to what they should be at stock. I increased the voltage to 1.45 and increased the multiplier only in increments running Prime 95 for 5 mins each time @ 3.8GHz I had to increase the voltage to 1.5 and it seemed stable for 5 mins but the temps went into the 70s so I stopped it. I went back to 3.6GHz and 1.45V (1.5 was running alot hotter) and ran Prime 95 and after 1 3/4 hours one core produced an error. I dropped back to 3.5GHz and it ran prime for 2.5 hours all OK. I tried to claw back to 3.6 by increasing the frequency but it was only stable to 3.55GHz without more voltage. I then rolled back the voltage in stages and found it was OK and stable with only 1.3875 in Prime 95 with temps 29C idle and never goes above 50C running prime (51 on a very hot day). I played around trying to get back to 3.6GHz but anything that looked stable increased the temps by 5C+ so I thought will I really miss that 50MHz. I then tried to benchmark with Sisoftware Sandra and there burn test crashed it instantly and I had to raise the voltage to 1.4375 to get it to run stable this only increased the temps around 1C since then I left it at that and its been running fine 17.5 x 203 1.4375V.
Ran P95 for 8 hours again overnight at 3.6ghz, no errors. Going to run LinX while I'm at work today.
My vcore is 1.4 in bios, but displays as 1.424v in CPU-Z. Still not sure about that one. Memory is running at stock specs, 1600, 9-9-9-24. Tried to push it to 3.8ghz with 1.45 vcore, but really wasn't anywhere close to stable. Tried bumping the NB up a bit as well, no dice. Windows would load, but as soon as I started up LinX or P95 I'd crash. Fairly certain it was a memory thing. Wondering if maybe I should clock my memory back to 1333 just to see if I can hit 3.8 stable...but I'm pretty happy with 3.6. Max CPUTIN at 3.6 is 55c, which means my cores should be around 50c, about 24c above ambient.