• 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.

Sandy Bridge Extreme Overclocking Thread (X79, 3930K, 3960X)

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
( lot of clutter in the thread.. Be nice if a Mod would clean it up )

I'm not totally through tweaking yet, but I'm settled in at 4.6 @ 1.35v ..
The CM 212+ has the temps at mid - high 60's with ambient of ~20 ..

Couldn't get stable at 4.8 @ 1.45v, and the power at full load goes to 430w versus ~330 @ 4.6 .. Temps go into 80's at 4.8 also ..

I've tweaked some of the power settings, but not sure what affect they have on stability .. I'll post a full list later. I think the fine tuning will have more to do with people trying to go above 4.6, but I'm not re

It will be fun to try to get it to 5 later, but for now I' just going to continue at 4.6, cranking out some F@H work units.. It appears to be at least 10k PPD better than the 2600K.

I still have this set up on top of my desk, but have a new CM RC-692-KKN2 case on the way, and this will replace my 2600k as my main PC ..
 
You are right on the 1st account and wrong on the 2nd..
However, there are significant BCLK adjustments that can be made with SBE, while leaving the multiplier alone if you choose..
 
Exactly, I'm using the 125MHz base clock so that I can run my memory at its rated 2000 DDR setting at 9-10-9-27-1T timings. Otherwise the memory speed selections go from 1866 to 2133 with a 100MHz base clock.
 
It's nice to see that SNB-E gives IPC + clock speed improvements over Gulftown. If Intel's tweaking their mature 32nm process like this, I am salivating over the upcoming IVB-E on 22nm goodness.
 
i checked the vcore and AI had it hanging right around the 1.4 mark. i took it off for now, hence why i looked. even at 4.68 WEI still won't max out! my hopes and dreams stolen! 🙁
 
I KNOW, RIGHT? I thought a 4.14GHz 980X would ding a 7.9...nope.

wtf!? what is required for a maxed cpu!? everything else is easy, but that cpu is 1 tough cookie to crack! i should take this opteron 24 core server and put windows 7 on it before i put it in commission. wonder how close that would get!


on a serious note. what's the max oc anyone has gotten on stock vcore? is the turbo mode clock about the max for stock clocks?
 
Last edited:
Not sure where that came from, but it is incorrect, the L3 cache is shared across all cores.

yes it shared across all of them. But each core only have 2.5mb of cache in reality. If you disable some of the cores for let say heat issues when overclocking disabling 2 leaves you with 5mb less cache. Every core you disable you loose 2.5mb of cache as well. So each core is limited to the amount of cache it has in reality.
 
yes it shared across all of them. But each core only have 2.5mb of cache in reality. If you disable some of the cores for let say heat issues when overclocking disabling 2 leaves you with 5mb less cache. Every core you disable you loose 2.5mb of cache as well. So each core is limited to the amount of cache it has in reality.

It is shared cache.. what part of shared is so hard to grasp?
 
Mine isn't all that extreme.

4.5 (45x100) on 1.33v 3960 via H80 Corsair.

Ram at 2400 10-9-10-28

Motherboard Gigabyte Assassin G1
 
Does the G1 Assassin 2 have a PCI-E aux power connector?

X79 does need it if anyone uses three or more graphics cards. Don't ask me how I know 🙁
 
Does the G1 Assassin 2 have a PCI-E aux power connector?

X79 does need it if anyone uses three or more graphics cards. Don't ask me how I know 🙁

Really ? Are you sure this is not strictly an Asus issue ?

I checked before getting my UD-5 with Gigabyte, they told me for 4x SLI I would need the UD7 because it has the connectors, but for 3-way I would be fine.

I hope their information is accurate. How did you find out your board couldn't handle it. A refusal to post ? Or something else ? :/

And FYI the only Gigabyte board that has the connectors is the UD-7, it's also the only board that supports quad-SLI. Seems flaky to me for Asus to sell a board with the ability to accept three cards and run Tri-SLI but not be able to power it.
 
Last edited:
Really ? Are you sure this is not strictly an Asus issue ?

I checked before getting my UD-5 with Gigabyte, they told me for 4x SLI I would need the UD7 because it has the connectors, but for 3-way I would be fine.

I hope their information is accurate. How did you find out your board couldn't handle it. A refusal to post ? Or something else ? :/

And FYI the only Gigabyte board that has the connectors is the UD-7, it's also the only board that supports quad-SLI. Seems flaky to me for Asus to sell a board with the ability to accept three cards and run Tri-SLI but not be able to power it.

http://www.evga.com/forums/tm.aspx?m=749507&mpage=1

Happened to my AX1200 yesterday. The 12V line was melted where it connects to the power supply. The motherboard is perfectly fine. Got another AX1200 today and have the P9X79 Deluxe modded just like it's done in that thread. It was pretty easy soldering it on there.

I should have known better. I did this same mod to my P6X58D Premium and never had an issue since doing it. Just thought the X79 may have fixed that issue and an aux PCI-E power connector wasn't needed. I was wrong.
 
http://www.evga.com/forums/tm.aspx?m=749507&mpage=1

Happened to my AX1200 yesterday. The 12V line was melted where it connects to the power supply. The motherboard is perfectly fine. Got another AX1200 today and have the P9X79 Deluxe modded just like it's done in that thread. It was pretty easy soldering it on there.

I should have known better. I did this same mod to my P6X58D Premium and never had an issue since doing it. Just thought the X79 may have fixed that issue and an aux PCI-E power connector wasn't needed. I was wrong.

Well that's unfortunate it, sorry to hear that.

Who would the end blame for this sit with ? Intel when they lay out the specs for motherboards, but I don't know that they have to stipulate specifications for exotic GPU configs, or the MB manufacturers ?

Well now I am nervous to set mine up. I've had 3 480s going in a X58-UD5 for a long while now, so I assume I am good. Would not be happy about my PSU taking a nosedive though, it's not cheap.

Thank you for your post. I may just return the UD5 and get a UD7 and not chance it. Knowing Gigabyte RMA, I don't stand a chance of getting them to replace my PSU if it fries out on me based on their email saying I will be okay. :sneaky:
 
I know I wouldn't chance it. I even checked the connectors on the motherboard and the power supply from time to time to make sure they were OK. Been fine and looked perfect for the past two weeks. Then yesterday when I ran PCMark Vantage the PC all of a sudden shut off. PCMark isn't even that GPU intensive, but it does a lot of polling between tests. That may have triggered the problem. I really don't think Intel took into account how much power todays GPUs could pull from the +12V lines when they came up with the 24-pin connection standard. It has only two +12V lines and they can get really hot with all 16X PCI-E slots populated. It really should be mandatory for any motherboard manufacturer to include an auxiliary PCI-E power connector with motherboards that support 3x GPU configurations.
 
I have a 6-pin PCI-E aux plug and a 4-pin DIMM aux plug on the Rampage M/B, so after reading your warning, I shall be plugging those in for sure. Only doing 2-way SLi right now, but better safe...
 
I'm glad I purchased an ASUS P9X79 WS. It has the molex plug on the mainboard. I should have the system put together Wednesday evening.
 
I took a look at the Asus P9X79 WS on the Asus website. The molex power connector in the upper right corner of the board is what they call an "EZ Plug" made to connect a power supply with an older ATX standard. It looks like it will work to also provide extra power for PCI-E.
 
Last edited:
Just finished my build and OC testing this weekend! Here are my results:

4.5ghz (100mhz x 45 multi) @ 1.296v (69C on max FFT Prime) with NH-D14

4.6 works OK too, but is pushing temps close to 80C, so I will keep this at 4.5ghz for now. The NH-D14 is so massive, I could not install the case fane I usually use, so I will work on getting a different case fan installed and that might help.

So far VERY happy with the results. Also - got all 8 DIMMS working with the cooler! 32GB RAM all working fantastically. 🙂
 
Update: After re-configuring my case fans, and adding a third NF-P12 Fan to the HSF (now it is Fan-HS-Fan-HS-Fan (with a case fan 8'' in front and 3-4'' behind) all flowing in the same direction, I have 4.6 GHz with LOWER temps that what I had with 4.5ghz. 🙂

67C max now on large FFTs with Prime.

I have my case all closed-up and will run 12 iterations on Prime95 for the next 48 hours to test stability.

Very happy!

Update: Core #12 is giving me issues (all the others appear to be fine); heat doesn't appear to be the problem though. Will need to tweak voltage and do some more in-depth memory testing as well.
 
Last edited by a moderator:
Update: After re-configuring my case fans, and adding a third NF-P12 Fan to the HSF (now it is Fan-HS-Fan-HS-Fan (with a case fan 8'' in front and 3-4'' behind) all flowing in the same direction, I have 4.6 GHz with LOWER temps that what I had with 4.5ghz. 🙂

67C max now on large FFTs with Prime.

I have my case all closed-up and will run 12 iterations on Prime95 for the next 48 hours to test stability.

Very happy!


What case ?
 
Back
Top