• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Help with overclocking a C2 955

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

ice445

Registered
Joined
Apr 21, 2012
I have a chip here that I'm trying to squeeze for a bit more. I'm running into a bit of problems though. The board I'm using is a Gigabyte GA-970A-UD3 (Odd I know, but my previous board failed and was out of warranty).

I'm also running 8GB (4 2GB Dimms) of A-Data GS 1600mhz ram at 9-9-9-24 and 1.65V.

Ram divider is 1:4, I'm using the multiplier to overclock. This chip is gold as far as I can tell, able to do stock 3.2 at 1.175V and 3.8 at 1.41V. Getting to four requires too much voltage for my cooler to handle, but is possible (I can boot at 4 at the same voltage it will run at 3.8.

My main problem is getting a proper northbridge overclock and making the whole thing stable. I'm not sure how many volts to use considering the extra strain of the ram speed and amount. I'm shooting for 2400mhz (according to dolk's table) and would like a guidline. 2600 seems to require too many volts (I *think* i got it stable at like 1.275V, as in it will pass prime, but that brings up my other problem. This chip runs quite hot. I haven't found a cooler able to tame it as of yet. Prime temps at 3.8Ghz hit about 57-58C, and temps at 3.6 (at 1.33V) hit about 55. With that said, is there really a heat wall? I definitely noticed a speed increase between the two settings, and it didn't seem to impact stability. Especially considering it's not going to be running at those temps outside of the stress tests.

So basically, any help to tune this properly would be helpful. Should I drop ram speed? Disable CnQ? Etc


Also the stock CPU-NB voltage on this board is 1.105
 
Last edited:
So I have it seemingly fairly stable at 3.8/2400 at 1.425V. The only thing that bothers me is the 58C load temp in stress tests. I need to do a full night of prime, but it passed IBT and 2 hours of small FFT already with no issues. Game temps don't break 55C.

I'm probably gonna try and redo my cooler and fan to see if i can get it a bit lower, but if it's stable there's nothing to worry about right? I'm not going over 58, which is still under that conservative 62C limit.
 
So I have it seemingly fairly stable at 3.8/2400 at 1.425V. The only thing that bothers me is the 58C load temp in stress tests. I need to do a full night of prime, but it passed IBT and 2 hours of small FFT already with no issues. Game temps don't break 55C.

I'm probably gonna try and redo my cooler and fan to see if i can get it a bit lower, but if it's stable there's nothing to worry about right? I'm not going over 58, which is still under that conservative 62C limit.

I would say you're safe at this point as long as your CPU socket temps aren't reaching 70c. Core temps are the critical ones but it's not wise to ignore socket temps altogether because core temp sensors vary in the accuracy of their calibration.

The most impressive thing here is that you are able to run the ram at 1600 mhz stable. Most Denebs won't do that. Especially with that high voltage ram that takes 1.65v to do 1600.
 
I would say you're safe at this point as long as your CPU socket temps aren't reaching 70c. Core temps are the critical ones but it's not wise to ignore socket temps altogether because core temp sensors vary in the accuracy of their calibration.

The most impressive thing here is that you are able to run the ram at 1600 mhz stable. Most Denebs won't do that. Especially with that high voltage ram that takes 1.65v to do 1600.

That's what I've heard over the net. I thought it was affecting my overclocks but it seems not. This chip is just not bothered by 1600 RAM, not even four sticks of it. At stock speeds it runs at 1.1V on the CPU Nb even, somehow.

I think I can even get away with NB at 2600/1.275V, but I'm just gonna leave it at 2400 since according to Dolk's chart the difference is negligible.

And my socket temp during prime/IBT is about 53-54C, the spread is pretty close between socket and core. idk if that's bad or good (it seems like a good thing).

Heres my settings, any where I should improve?

zFgUv.png

YVha9.png

I dropped the memory to 1T command rate and it passed prime and stuff when I was at 3.6, not sure about 3.8. I didn't think it was worth the gain though.
 
Last edited:
what you could try, and this works for me but might not work for you, look for load line calibration on your board, if it has it, turn it off, and set your core voltage higher, to about 1.47, it does not affect your idle temps too much but it does drop your load temps a a few degrees, I actually do have to run the cpu/nb at 2600 at 3.8 though, it does not seem to want to be stable at a lower speed, but I can get away with a 1.25v on it.
 
My new Bulldozer board doesn't allow LLC to be disabled, only "regular" or "extreme". With that said, regular mode seems to be decent enough at keeping the volts pretty flat around 1.4.

It sounds like your board has overcompensating LLC and lots of vdroop going on. But thanks for the suggestion.

HOPEFULLY it's nice and dialed in. Will prime blend overnight and see what happens.
 
My new Bulldozer board doesn't allow LLC to be disabled, only "regular" or "extreme". With that said, regular mode seems to be decent enough at keeping the volts pretty flat around 1.4.

It sounds like your board has overcompensating LLC and lots of vdroop going on. But thanks for the suggestion.

HOPEFULLY it's nice and dialed in. Will prime blend overnight and see what happens.

I wouldn't say lots of vdroop, its goes from 1.47 to about 1.43 which is normal, but with LLC on it goes from 1.43 to 1.5 or so, it was a great mobo for the time, but 3 years is a long time in the PC industry, every time I want to replace my C2 it makes me want to redo my whole water loop so I never changed it lol, but good luck with the OC, the C2 can be a real pain to clock sometimes.
 
Yeah I'm going Ivy as soon as I have some dosh, but until then I'm just squeezing this for all I can get out of my $30 air cooler. It still performs remarkably well in modern applications. Although that's probably just because games haven't advanced all that much computationally yet.
 
Back