• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

3930K Overclocking help

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
I wouldn't bump LLC up so high. It may actually result in your voltage being higher while under load.

Set Linpack to run for a long time. Let it go for about 12 hours. If you manage to Linpack for 12 hours, you should be pretty damn stable.

How do you set it to run for 12 hours? Didn't see an option for that, all I see is 'times to run' should I set it to 400 or something ? And which preset should I use, normal, high or extreme ?
 
Last edited:
Just go negative on your offset Voltage. Your chip is probably better than mine (I've got an old one already, as tech goes). I've gotten mine down to -0.065 Offset (4.6GHz@1.25V). Works great for everything but, you guessed it, Prime95. Can game, crunch numbers, run all the benchies, serves breakfast in bed (don't much care for that). It's just easy and reliable to run 4.4GHz. It doesn't seem to make a difference with a 6990+6970 when I jump it to 4.8-5.0 with massive voltage increases. All the 3dMarks (01,06,V,11) & Heaven 3.0 only show negligible increases in speed. Even massive video overclocks don't change the numbers that much. I guess that's when you grab a few 7990s and a really efficient power supply. Enjoy your Quest! I finally killed the Emporer today at level 46! 😉
 
Water cooling is really the only way to efficiently overclock.
DSC00086.jpg

TECs suck massive amounts of electicity (probably more than AC) and fan/heatsinks don't work very well. Highest temp on a 3930K for me is 53C on core 4. That's with a 32C Ambient. That's Prime 95 and FurMark3 at the same time. I'm still tripped on anybody going over 1.4V.
 
I found something important about this chip. I was going for 4.5ghz also and was forced to keep voltage in the 1.36 area and even then it wasnt fully stable. Linepack and prime passed for 10 hours, but BF3 would crash every other day or so. Temps were through the roof etc.
I backed it down to 4.3 and omg it made all the difference. 1.28v rock steady stable. LLC set to regular (off) and all other current/voltage settings are all regular, off or auto. Temps are back to an acceptable level and nothing has crashed once in a long time. I would have to recommend going to 4.3 if you are having any difficulty at all. I have read a few guides and they all show that 4.3 is the breaking point. After 4.3 the voltage requirements suddenly sky rocket and it quickly becomes not worth it (not to me at least).
 
Hey Mom,

Why post about *your* computer in someone else's thread? The OP doesn't know about or probably care about your TECs, assuming that's what you're using based on the contents of your post. Your screen shot is largely irrelevant to this thread. Your 53C comment in conjunction with the rest of your post, sounds like "oh, I just made an account let, me find somewhere half way relevant to brag about *my* computer."
 
I wouldn't bump LLC up so high. It may actually result in your voltage being higher while under load.

Set Linpack to run for a long time. Let it go for about 12 hours. If you manage to Linpack for 12 hours, you should be pretty damn stable.

Usually about 30 iterations should be sufficient; never heard of anyone running it for 12 hours. Some say 50 iterations. For Prime95, I'd run two tests: one with SFFTs and the other with LFFTs. As much as 10 to 15 hours each. Usually, if it passes LFFT test, it would pass SFFT test.

Then, with LinX or some LinPack variant that reports GFLOPs, if you set up the program properly (for hyper-threading, if the processor has HT), determine the processor's maximum GFLOPs, a few sets of 10 runs each should allow for finding the minimum, rock-stable voltage.
 
Back
Top