• 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 5 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
I use my system to game and crunch BOINC. My system is rock solid stable at the settings I've listed doing both for extended periods. I BOINC on the cpu 24/7 with no crashes @4.7ghz. That's stable to me.
 
DooKey said:
I use my system to game and crunch BOINC. My system is rock solid stable at the settings I've listed doing both for extended periods. I BOINC on the cpu 24/7 with no crashes @4.7ghz. That's stable to me.
Click to expand...

I've noticed with folding at home that the CPU temps are much much lower than when I test with AVX LinX. By over 10°C.
 
Still at 1.34v (underload) 1.36 set in bios (LLC is left on auto so not sure what settings its running adam)

ok, ran Linx AVX .. for 4hrs no problems temps the same under load @ 55-59deg

Left Heaven unigen looping for 2hrs no problems

Occt Small & large sets 2hrs each no problems

Ran LinX using 90% availabe ram no issues/no errors (ram @ 2133mhz 9-11-10-28-1.65v 16GB

Playing BF3 atm no issues so far.... lol uses 20% of cpu ... on my 4ghz 960 i7 was using 90%

found my 3930k to be 1.33 (underload 1.35v set in bios) to be stable .. but i always find stable point (2hr occt test) then give it 0.01v increase (so 1.36 in bios) just for extra stability peice of mind....

Now i found my 24/7 OC settings will post up some 5ghz attempts shortly just for the fun of it...
 
One thing i have noted, is watch the VRM temps , The rampage IV formula (not sure about the extreme) only has temps for only CPU-MB-PCH & OPT temp sensors.. not for the VRM 🙁

I used my IF Temp gun on the VRM heatsinks above the CPU 65deg C :| after few hours of LinX AVX running.... i wonder how hot they will get when i go for 5ghz runs @ 1.5v plus :hmm:

Put a thin sycthe 120mm fan blowing down on them with the speed turned right down (inaudible) 30deg now 😀

I can understand now why the small fan is on the R4E

Is 65deg too hot for these VRM to be running at??... on the R4F box it says there rated for 105deg so maybe 65deg is fine.... PCH /X79 tops out at 40deg under any test's ... good its not connected to the VRM like aother boards independent heatsink

What are your guys VRM + X79 chipset temps running at full loads after a while ?
 
On Amd's 5970 I know the VRMs were good up to about 120C. Looking at the specs of most VRMs you see they consume more power when they are hot so its more efficient to run them cool in the same way its more efficient to run silicon cool. 65C is nothing to worry about.
 
Couple of observations after having my 3930 running over the past week:

12 instances of prime (small FFT) + OCCT GPU stress puts my CPU temps right around 73-76. That's acceptable to me.

Did some temperature monitoring while playing Skyrim, and CPU stays pegged at about 44C with the GPU running at max. Very good temps gaming.

While Folding (on all CPU cores 100&#37😉 and GPU folding (still running the 4870 hack, so 50% stressed) CPU temps stay in the mid-high 50's.

Regarding stress testing, I hit 48hrs+ on prime95 small FFT without any errors and temps were <=70C which was my goal. Feeling good about this clock...Could maybe eke-out another couple hundred Mhz, but I really don't want to screw with the voltage too much. A very good WC setup could probably get 5ghz without a ton of trouble after increasing voltage.

Edit: My ambients are HIGH (~23-24C); office doesn't have a window outside, and seems to get suer cold in summer and nice and warm in the winter. Perfect for me! Temps will only go down in the summer...

After a couple days of 6 cores @ 100%, the temps jump a tad too!
 
Last edited by a moderator:
DooKey said:
What's unsafe? From what I've seen 1.4v and below is fine per Intel.
Click to expand...

Intel's max safe used to be listed as 1.35V for bloomfield, on the new process it should be lower but I can't find an official statement. 1.4V is what the community thought was probably OK and has been using as a guide for SB all year. But I have seen a report of death at 1.45V already with SB-E.

My previous i7 920 has run at 1.35V for nearly 3 years. In the last 4 months it lost the ability to run at 4Ghz and now wont go above 3.8 stable. I also can't turn HT on above 3.4, which when new worked at 3.8. Degradation happens even at moderate voltage increases within the potential life of the chip and most people considered the 920 was safe on 1.45V. All depends how long you would like your processor and motherboard to last.
 
BrightCandle said:
Intel's max safe used to be listed as 1.35V for bloomfield, on the new process it should be lower but I can't find an official statement. 1.4V is what the community thought was probably OK and has been using as a guide for SB all year. But I have seen a report of death at 1.45V already with SB-E.

My previous i7 920 has run at 1.35V for nearly 3 years. In the last 4 months it lost the ability to run at 4Ghz and now wont go above 3.8 stable. I also can't turn HT on above 3.4, which when new worked at 3.8. Degradation happens even at moderate voltage increases within the potential life of the chip and most people considered the 920 was safe on 1.45V. All depends how long you would like your processor and motherboard to last.
Click to expand...

Interesting. I have a 920 at 4ghz, 1.36v for the last 15 months. Still stable as the first day it was OC'd.

I had a 980X I ran at 4.2ghz, 1.4v for over a year. Still stable as the first day it was OC'd.

Guess ymmv.

If I can get 2 years out of my cpu at current clocks/voltage I'm happy. I'm going to upgrade to IB-E in 2013 anyway.
 
I don't know if anyone has noticed that when you take the multiplier past 32, turbo is set to on in the bios automatically, and the button is grayed out, so that it can't be shut off. Just like the i7 920's to 960's, when you would go past that 23rd multiplier, it's turbo. Yes folks, I'm sorry to say you didn't have a true 24th multiplier, or 4.5 GHz, or what ever your processor was set to. That one is only kicking on as a last resort so to speak, and if your stress test wasn't running the processor hard enough to keep it at the speed it was set to in bios, then it may not be stable either. This is also how your ridiculously low voltages are achieved. Try this on for size. First set the voltage on that 3930 to auto, so it doesn't bsod. Next set that multiplier to 32, which is max before it turns into turbo poo. Disable turbo it self, and leave what ever c states you want active. I personally only use c1e - lower processor idle state, because I don't use hibernate or any variation of. Then set bclock to 125. This will give you 4.00GHz. Almost forgot, turn off speed stepping, and you will see your true speed under windows experience index in the system properties. Now you will notice this processor requires quite a bit more voltage than the 1366 did, and there is very little tolerance. The lowest I could get my voltage while maintaining rock solid stability was 1.45 vcore. Yup, that's right. Still runs cooler than 1366 too. Same waterblock, and it runs cooler. At 4.00GHz with my i7 950 I had the vcore set to a very stable 1.28v. There is a considerable difference in voltage, but that is totally understandable, because we now have 2 more physical cores, and 2 more threaded, so this is a hungry processor. If you want to test my stability theory, download intel burntest V2 and set it to 128 threads with the memory at very high. Don't do max on memory because it will lock your system. Our goal is to tax the processor anyway. Burntest is preset to run for 5 cycles, and it will take maybe 45 minutes at the most. I guarantee this stress test will do a better and faster job than Prime 95. I would like some feed back on how some of your processors do with this stress test when you are clocked at 4.6GHZ etc... I hope I was able to give some helpfull advise. I am always open to suggestion, and still learning like all of you.

P.s. I would say you are unsafe, or beating on the processor, when you can't get those idle temps at 40c or below, and full load temps are consistently above 75c. It is the constant heat that damages the processors. If she is taking more juice, but staying nice and cool, the processor will be just fine and last for a long time.
 
Last edited:
Here's my idle. BTW - What's up with my 5th core? It always reports much lower than the others.

3960X-Idle.png
 
DooKey said:
New clocks for my SB-E.

4.8ghz@1.392v

48*100

Temps average 70C across all cores.
Click to expand...

Updated - but what are you using to cool now? A 212 EVO is going to have a tough time with that 3930K under load at 4.8Ghz. What load tests did you run at that speed & for how long?
 
Teufelshunde77 said:
... if your stress test wasn't running the processor hard enough to keep it at the speed it was set to in bios, then it may not be stable either. This is also how your ridiculously low voltages are achieved.
Click to expand...

I monitor my runs with CPU-Z; during load tests it is pegged to the max speed that I report. Since I use manual voltages, vcore doesn't change except with load line calibration, even when C1E state throttles it down.
 
4.719 GHZ @ 115 Bclk , 41x Multi , 1.395v (10hrs Linx w Avx stable)
4.615 GHZ @ 115 Bclk , 40x Multi , 1.360v (1.345 load) (5hrs Linx w Avx Stable)

Using 4.615ghz for my 24/7 OC atm

3930K

Koolance CPU370 Waterblock

Ran Linx Avx on full settings keeping cpu 100&#37; + ram at 98% (16gb of it lol) is that the best stability test for these OC ?

Adjustments Made

CPU LLC set to high
CPU Current capability = 130%

Temps seem good they hardly rise with increase voltage maxing out at 61deg and idle at 35-36deg..

Im seeing core #3 as stated before having temps 15-16deg Lower than anyother core maxes out at 49deg WTF .... maybe dodgy temp prob ??

is 1.4 safe for 24/7 SB-E ?? as long as it lives 3 years i will get every bit out of it lol .... at 1.4v maxes at 65deg idle 39deg (hotest core)

Did some tests at 1.49v 4.85ghz was Linx stable for 10min
and 1.56v for 5ghz was needed for linx stable for 10min

had to make some skew adjustements + LLC ultra + extra cpu current

(didnt run any longer than 10min) temps nudging 80deg

+ put a radiator in a bucked of ice for 5ghz run
 
Last edited:
mrjoltcola said:
Updated - but what are you using to cool now? A 212 EVO is going to have a tough time with that 3930K under load at 4.8Ghz. What load tests did you run at that speed & for how long?
Click to expand...

I'm sorry. I'm on full water cooling now. Apogee XT block.

I've run folding@home for 24 hours.
 
Got my R4E in today.

What a difference a motherboard can make. On my X79-UD5 I couldn't get past 4.5, I've gotten to 4.8 already on this board with 1.39. Just using the straight 100x48, no time to really play a lot or check much stability.

This R4E is rather daunting with all the options in the BIOS. I find the CPU vcore LLC setting helps a lot with an overclock. Putting it to extreme really seems to jack up vcore though.

For the moment until I have more time this weekend I am doing 4.6 with 100x46 @ 1.35. Only have done maybe 30 minutes of small ffts on prime and about 10 runs of linx, but no fails. BF3 also hasn't crashed and that was crashing on the UD5 @ 4.5 with 1.36 volts.

This board is a boss. My temps are still poor though, mid to high 30 idles. Perhaps I didn't mount my NH-D14 well. Couldn't find my usual NT-1 paste and had to use some AS5.
 
@GrooveRiding : Congrats on the R4E. Big difference isn't it? I'll update the thread. I love the board too (heck I'm a sucker for all ROG stuff, I have three total). The BIOS is the best to date. It is almost overkill, but leaves nothing to be desired. My only problem is my board is more capable than my 3930K. Sounds like you got a better 3930K than me. I may try again with the replacement board, but I'm not expecting any differences.
 
mrjoltcola said:
@GrooveRiding : Congrats on the R4E. Big difference isn't it? I'll update the thread. I love the board too (heck I'm a sucker for all ROG stuff, I have three total). The BIOS is the best to date. It is almost overkill, but leaves nothing to be desired. My only problem is my board is more capable than my 3930K. Sounds like you got a better 3930K than me. I may try again with the replacement board, but I'm not expecting any differences.
Click to expand...

I've done some more tweaking but still staying with the 100 BCLK, for some reason, maybe due to my inexperience - I get a lot of crashing when I try playing with gear ratio and BCLK. I think it may also be because I have 1600MhZ ram and I'm limited in the multiplier I can use.

I have 47x100 @ 1.36 20 passes of linx stable now. Going to let this go on Prime for a day and see if it holds. I had to play with the VTT, VTTCL and PLL. I also changed the clock generator to 20uf per something I read on the evga forums and that helped a lot in lowering voltages and maintaining stability.

My goal was 4.8, I think I can do it, but it's too much voltage for my liking. I can live with 1.36, maybe I can mess some more and get it to 1.34-1.35.

:awe: I love the board though, hope this does not set me down the path of buying expensive mobos now. Something I'd always avoided.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top