• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Can you help with my Haswell-E 5930k overclock?

Caporegime
Joined
30 Jul 2013
Posts
30,187
I spent a few hours last night tinkering with my i7 5930k to try and improve the overclock but what I realised was it wasn't quite as stable as I thought.

Firstly, my system specs:

Cpu: Intel i7 5930k
Motherboard: Asus X99-S (latest BIOS 1103)
RAM: GeIL EVO Potenza 16GB (4x4GB) DDR4 PC4-24000C16 3000MHz
GPU: Sapphire 295x2 & Sapphire 290x (Trifire)

Cooler: Swiftech Open Loop H240-x AIO (280mm rad / 2x 140mm fans)
PSU: SuperFlower 1200W Platinum PSU


First time I overclocked I used the 5 way optimisation (all cores) and it came back with a recommendation of 4.4Ghz at something like 1.29 volts (adaptive)

I then manually adjusted it in the BIOS AI Extreme tweaker and got what I thought was a stable 4.5Ghz at 1.31 volts. (Will come back to this later)

However, I noticed my RAM was 'only' operating at 2400Mhz, so I attempted to activate the XMP profile.

This changed the CPU strap and BCLK speed from 100 to 125Mhz and the multiplier from 46 to 36, and the memory speed to 3000Mhz which is supposedly what the RAM is rated to.

After this, the PC won't even POST.

Now I don't know for sure, but could this be due to an incompatibility between the BCLK speed and my GPU's?

The reason I ask is that I was in a situation where my BCLK speed was set at 102, either by AI Tweaker or the BIOS Tuning Wizard and although windows would load, the GPU's weren't detected properly, and re-installing the drivers would cause the screens to flicker on and off. The only way to resolve that was to set the BCLK back to 100.

Note that neither of my GPU's are overclocked past factory defaults...

If it isn't the reason for not getting past POST, then why is the CPU (seemingly) happy to boot in to Windows at [email protected] and play games, but the minute I change it to [email protected] v, it won't even post?

This brings me to another issue. I thought I had better properly test my system stability at 4.5Ghz so I downloaded the Asus 'Realbench' stress test, set it to 16GB and left it running for 15 minutes.

After 5 minutes it BSOD

So I upped core voltage to 1.32 and ran the test again

After 8 minutes another BSOD

I again upped the core voltage to 1.33 and ran the test again

After 11 minutes another BSOD.


So obviously my 4.5Ghz overclock isn't as stable as I initially thought. I was monitoring the CPU using Core Temp and during the last stress test most of the cores were in the 75°C range, but core 2 did hit 82°C. I believe this is still well within TJ Max temps but considering the hefty cooling I have on the CPU it was a bit of a concern.

For now I've dropped the overclock back down to 4.4Ghz as the 5 way optimisation initially set. But really I'd like to be able to run the Ram at 3000Mhz as well, or try and get my CPU stable at 4.5Ghz.

I got lucky on my 4770k chip as that did 4.5Ghz all day long at 1.2v core, and that was a simple overclock but the Haswell-E looks like it will need a lot more tinkering.

What other settings should I be changing?
 
Last edited:
PCI e is tied to bclk. So for multi card configs it's recommended to leave bclk on defaults. System agent controls PCI e controller voltage. These are defo your candidates. System agent is non linear. It has sweet spots normally around 1v
 
Thanks guys

No mention of the BCLK or CPU strap affect on GPU's though?

Too much to cover and this is legacy really! I have trouble on my system running 2800 DRAM on 125 strap due to the divider taking PCI frequency over 101. Not grave trouble but very occasionally would encounter GPU instability. Some GPU won't even post with more than 101mhz.

As above I recommend you stick to 125BCLK when using 1.25 strap when possible, especially with more than one GPU
 
Well I tried it at stock but with the xmp profile enabled and its got through 15 minutes of the Asus real bench stress, so clearly the GPUs can run on a blck of 125.

Just need to see why it won't post once I up the multiplier.

Core #2 got to 69c at stock speeds under the stress test though. How normal is that with good water cooling? Most other cores were mid 60.
 
Looks good to me. Mine is 36 x 125 at 1.27v and it hits 79 on the hottest core on air. If it's stable and you're going above my temps you might want to check your block is seated correctly and that your rad is getting sufficient airflow.
 
Reseated the cooler for the CPU.

BLCK 125mhz, CPU 4375mhz and memory at 3374 MHz(?)

My FSB: DRAM ratio is apparently 1:14

Anyway, highest CPU core this time was 72c but most were mid 60's and that was with a bigger overcloxk than previous two runs of RealBench 2.41
 
Weird problem

System is stable as anything, plays games fine for hours, passes stress tests etc and turns on and off fine for 3-4 days.

Then I'll switch it on and all I hear is the fans on full. I have to hold down the power button, restart and then the BIOS screen comes up with a message saying Overclocking failed, press F1 to enter BIOS and I have to revert to default settings to get the PC to reboot.

Then I let Windows load up and then quit and go back to the previous overclock and then it works fine again for a few days.

What's going on there?

BLCK: 125Mhz
CPU Ratio: 35x
Memory: XMP 3000Mhz
 
I had something similar happen and it turned out to be my nVidia drivers having a display port problem. I reverted to an HDMI cable (and dealt with 50hz) and the problem went away. However, I recently updated to the latest nVidia drivers and given the DP another chance and had no problems.

It may not be your problem (I had the screen go black on me and then had to hold a cushion on the PCs face and then it exhibited your symptoms...) YMMV, but this may not be an overclocking issue...
 
I should really have checked the little display on the back of the motherboard.

I assume there is one actually. I had one on my Z87 ROG Impact VI, but not sure if the X99-s has one.

Anyway, I set my system agent voltage to 1.02 as recommended.

How much effect does that have on overall CPU temp? I was fiddling around trying to attain 4.5Ghz and 3000Mhz - fully stable.

Seemed to work fine at 1.32 Core voltage before I upped the System Agent.

When I did up the system agent voltage, I dropped CPU core to 1.31 and Windows blue screened about 13 minutes in to the Asus RealBench test.

Monitoring program gave me a 'CPU core temp' of 71°C - but I think that's the motherboard sensor?

It also shows Current/Max temperatures on each core and a 'CPU package' temperature and the maximum temp that 2 of 6 cores got to, was 80°C, which seems high.

I think they are the more accurate temps though as they are supplied by the Intel digital thermal sensor?

TL : DR

5930k - 4.5Ghz needs 1.32 Vcore
Hopefully system agent voltage at 1.02 will solve my reboot problem.
 
From what I've found the 22nm trigate process (Ivy Bridge onwards) doesn't really respond much to more voltage and generally once you have hit 'the wall' there isn't much you can do to go past it (unless you just ignore proper stability testing and pretend that running a game qualifies as stable).
 
No but I can run a good CPU overclock and RAM speed, perfect stability and have no problems booting unless I turn XMP on, and then I get problems with cold boots regardless of System Agent voltage, turning memory tuning off etc etc.
 
Back
Top Bottom