New to Overclocking on an Intel 2700k

I just got the OC bug and finally built a machine I've been dreaming of for years. Now, I"m trying to overclock it, but am getting pretty frustrated with the results. I'm following this guide to overclocking my i7 2700k, but I'm getting BSODs a lot and think I'm doing things wrong.

Using my Maximus V Extreme mobo's "CPU Level Up," I'm able to get great stability out of the box at 4.6GHz, but I want to go further. CPU temps are great, using an H100 as a liquid cooling solution and it's working AWESOMELY. Main problem I seem to be having is voltage.

First, I tried clocking to 4.8GHz using 1.35V and a 48x multiplier. I got some BSODs, but bumped the voltage. Is it really this painstaking? Should I only increment the voltage by 0.005V which is the increment on my motherboard? Usually, when I start Prime95 or Linpack, I get a BSOD within a few seconds. So nerve-wracking. Getting frustrated, I figured that I might as well try to break the 5GHz wall and set my multiplier to 50 and try to get my voltage up to where it'd be stable. I soon got to about 1.41V or so, when I gave up and came back down to 4.6GHz with the "CPU Level Up" method.

Am I doing things right? Are there other factors that can cause a BSOD other than the CPU not having enough voltage?

There's a lot more to overclocking than just raising the Vcore, especially when reaching for such high clocks. List the rest of your system spec's including the EXACT model of RAM installed, the quantity, the rated and current DRAM frequency, timings, and voltage. And what were the EXACT errors in the BSODs, including the OS you're running?

Intel Sandy Bridge i7 2700k
ASUS Maximus V Extreme LGA 1155
Corsair Vengeance 32GB (4x8GB) DDR3 1866 1.5V 9-10-9-27
EVGA GeForce GTX 670 4GB 256bit
1000W COOLER MASTER Silent Pro
Corsair H100 Liquid Cooling

My day-to-day OS is a Debian derivative (Linux), but I'm running most benchmarks in Windows 7 64bit, though I've ran extreme Linpack benchmarks in Linux and seen good stability too.

Before I start messing with CPU voltage, should I just run some basic stability tests to make sure that my memory is fine? Is that a good strategy? Could that be the problem? I'm not sure what the BSODs were from, but I can run more tests tonight to see what codes I'm getting.

Boot into Windows and open CPU-Z to the CPU and Memory tabs. There you can find the info I asked for, including the idle and loaded Vcore while running Prime, IBT, etc. As far as checking for RAM stability, boot to Memtest86+ installed to either a thumb drive or CD / DVD. Or if the Windows version is either Vista or Win7, you can run the built-in Windows Memory Diagnostic utility.

redduc900 said:

Boot into Windows and open CPU-Z to the CPU and Memory tabs. There you can find the info I asked for, including the idle and loaded Vcore while running Prime, IBT, etc. As far as checking for RAM stability, boot to Memtest86+ installed to either a thumb drive or CD / DVD. Or if the Windows version is either Vista or Win7, you can run the built-in Windows Memory Diagnostic utility.

Click to expand...

Will do. As far as the RAM test is concerned, how long should I run Memtest86? I've ran it in the past and it seems to run forever. Will it ever actually complete?

Here are the screenshots you requested. Thanks!

CPU-Z CPU Idle


CPU-Z CPU Load


CPU-Z Memory

*shameless bump*

Likely the BSODs are from the amount of memory your running. Try loosening the memory timings up a bit.

I doubt its the ram... its running underclocked at its factory timings...IMC however....

At that clockspeed though, you may want to make sure PLL Override voltage is enabled. Somewhere around 4.5Ghz+ you need to enable that. Some are low like 4.5GHz others need it much higher. Id try that first if raising vcore didnt change anything.

That said, you do have a ton of ram, so VCCio/SA may need bumped up as well.

After running Memtest86+ for 10 hours, I had around 10 errors by pass 2. Is this just a case of bad RAM? Do I need to RMA? I was running it at stock timings and stock speed (DDR 1866 and 9-10-9-27). Would giving the RAM more voltage help?

@prime81 How should I loosen the memory timings?

@EarthDog How should I bump VCCio/SA? I can enable PLL Override.

Raise the default VCCIO voltage from 1.056V to 1.12-1.20V (no higher than 1.20V), and leave the "2nd VCCIO voltage" option on Auto. Also bump the DRAM voltage from 1.5V to 1.55V, leaving the four primary timings at their rated 9-10-9-27-2T. You might also want to play w/ the "IMC-DRAM Offset" option.

Done. I'm still seeing BSODs on large IBT runs. I just ran 10 times @ the Standard (1024MB) with All Threads and no BSOD, but when I ran a "Very High" memory test, I hit this BSOD:



This is at my "stable" 4.6GHz clock as outlined by the screenshots above, albeit with the tweaks suggested by @redduc900 above.

Another thing to try maybe is to remove two of your DIMMs, try one set then try the other to see if it is your IMC or a bad module limiting your overclock. 32 GB is a lot more memory than most people use when overclocking, and will put a bit more of a strain on your IMC. But then again it may be fine.

A couple years ago I was pulling my hair out trying to figure out why I would get random BSODs and could never seem to get any overclock prime or linX stable, turned out 1 of my 4 DIMMs was bad and would only run reliably at 1066 speeds vs the 1600 advertised. @1600 Memtest would run a pass or two fine but would almost always start showing errors after pass 3. A quick RMA and all was well.

Edit: 0x101 BSOD is usually insufficient Vcore.

When you run the CPU at stock clocks does the memtest86 fail? If the memory does not fail I would look at the CPU V core.

georgelogy said:

Another thing to try maybe is to remove two of your DIMMs, try one set then try the other to see if it is your IMC or a bad module limiting your overclock. 32 GB is a lot more memory than most people use when overclocking, and will put a bit more of a strain on your IMC. But then again it may be fine.

A couple years ago I was pulling my hair out trying to figure out why I would get random BSODs and could never seem to get any overclock prime or linX stable, turned out 1 of my 4 DIMMs was bad and would only run reliably at 1066 speeds vs the 1600 advertised. @1600 Memtest would run a pass or two fine but would almost always start showing errors after pass 3. A quick RMA and all was well.

Edit: 0x101 BSOD is usually insufficient Vcore.

Click to expand...

I'm going to try wingman99's suggestion maybe on Saturday night and leave it running all night to see if MemTest86+ is failing on default CPU clock. If nothing conclusive comes up, I'll go through the long process of testing each DIMM at 1866 by itself. Should I test sets or should I test each DIMM separately?

I just need to keep telling myself that I'm learning a lot out of this, albeit the hard way

wingman99 said:

When you run the CPU at stock clocks does the memtest86 fail? If the memory does not fail I would look at the CPU V core.

Click to expand...

This I'll have to try. I'll try this over the weekend after I've installed my RAID card and hooked up my drives.

Should I also just try running my memory at a lower clock rate? Will I get better theoretical stability if I'm running my memory at 1333 or 1600 rather than 1866? I know it's rated for 1866, but could this be the problem?

It couldnt hurt... but one thing at a time so you know what change made the difference.

Good point, I'll be sure to test incrementally.

Okay, so for the last 12 hours or so, I've been running Memtest86+ here on individual DIMMs. So far, 3/4 have all shown errors in Memtest86+ running at stock timings (9-10-9-27) and stock frequencies (1866) at a DIMM voltage of 1.55V.

The final DIMM seems to have passed with flying colors at the same timings, frequency, and voltage. I haven't touched anything with the CPU.

Is it really possible that I got 3 defective DIMMs or is something more nefarious going on here?