Cold boot "Overclocking failed"?

Associate
Joined
20 Nov 2009
Posts
2,050
Location
Haarby, DENMARK
Ever since I replaced my core i7 5820K with a core i7 6900K, I am beginning to experience cold boot problems, where the pc won't boot but shuts down and starts op with "Overclocking failed*. I then enter the bios and set my same OC settings again and reboots and then it runs fine.
I can run Prime95 and other stress tests for hours with no issues at all. I can also reboot the pc and windows with no issues.

Then the pc can run fine for some days and it doesn't complain when booting. Then suddenly it'll start complaining again and then after every shutdown then it will come with the error message a few times before running fine again.

I have no idea what is going on with it.

I normally run the OC at 4.1ghz, 1.252vcore voltage, Kingston HyperX Savage 2666mhz cl13 at 1.35v.
Dropping the cpu OC to stock doesn't help. I tried turning of xmp profile and set ram manually, but didn't help either. Also tried raising ram voltage to 1.36v to no avail.
But can it really be that the 6900k can't handle 2666mhz at all?, my old i7 [email protected] ran 100% fine with 2666mhz.

I am running the latest N. 80 bios for my MSI x99s krait motherboard.
It kinda annoying I can't figure out what is wrong, and I'm thinking about if trying another 2666mhz kit from Crucial or maybe get rid of the cpu and get a ci7 5960X instead.?

My wife is running 32gb corsair vengeance lpx CL 15 2666MHz ram, ci7 [email protected] on a MSI x99s gaming 7 without issues at all.
 
Last edited:
It is possible that the 6900K has a weaker IMC. Try it at 2400 for a while and see how you get on.
You also could try running a memtest to check the ram. Though I doubt that its a problem as you say it runs fine once booted.
 
It is possible that the 6900K has a weaker IMC. Try it at 2400 for a while and see how you get on.
You also could try running a memtest to check the ram. Though I doubt that its a problem as you say it runs fine once booted.

I will try 2400mhz at the same timings, and see if that changes anything. It just bothers me that I can't run it at those speeds and others can run 4.3ghz and 3000mhz ram.
If it won't do 2400mhz either, I'll try to get hold of a 5960x instead.
 
It could also be the board. You on latest bios?

Yes I upgraded to the latest bios before I installed the 6900K Version N. 80 from july 18th 2016. There are none newer.
I also suspected the board at once, but I never had issues for the 2 years i had the 5820K running. If the board wouldn't boot at all before doing a cmos reset, I would highly suspect it to have been the issue.
I'm so close to ditching the cpu and get a 5960X.
 
Yes I upgraded to the latest bios before I installed the 6900K Version N. 80 from july 18th 2016. There are none newer.
I also suspected the board at once, but I never had issues for the 2 years i had the 5820K running. If the board wouldn't boot at all before doing a cmos reset, I would highly suspect it to have been the issue.
I'm so close to ditching the cpu and get a 5960X.
I would get a j batch 5960x, much better cpu all round.
 
Trying to run the memory at 2400Mhz cl13 1.35v now and see what it does - and I placed an offer on a j batch 5960X also.
 
Did u try tuning memory sub timings and tuning io and sa voltages. I think this is just a training issue when cold.
 
Did u try tuning memory sub timings and tuning io and sa voltages. I think this is just a training issue when cold.
I haven't tried setting IO or SA voltages, I have just left them at auto since i'm not sure what the "safe values" would be :)
As for memory timing I did try running them slightly lower at CL14-16-16-39 2T 1.35v at 2666Mhz instead of their rated CL13-15-15-39 2T 1.35v 2666MHz, but that didn't seem to have much effect, since the 4th time I turned on the PC from cold boot I got the error again.
I am suspecting if it can have anything to do with the fact that my memory is double sided whereas I've heard Broadwell-E prefers single sided when running higher MHz ?
 
I can see that my motherboard as "Auto setting" sets the SA Voltage to +0.384v, so that is on all the time.

I have been running the ram at 2400MHz CL13-15-15-39 2T 1.35v and it (7-9-13 knocking on wood) seems to be stable when cold booting. Since I dropped in Ram frequency I upped the CPU frequency to 4.2GHz (same 1.252v) and Aida 64/OCCT runs stable at that speed and them temps average around 70 degrees celcius.
 
Last edited:
Seems ridiculously high offset on io and sa for that men speed. Msi going voltage crazy. +0.25 is the top of what is needed for 3200mhz+

For the speed your running surely +0.1 is enough.

I think you need to manually set up this memory correctly.

Matching your cache and core voltage may help with this cold boot issue or setting cache around 1.3v.
 
I tried setting the memory manually initially and at first I thought that was the solution (dropped the XMP preset), but on the second day it did the OC fail thing again. I haven't tried setting the SA Voltage manually to a lower setting, but wouldn't that make the system even more unstable if it cannot do 2666Mhz cl13 at 1.35v per default.

I might have a try setting the cache voltage to 1.25 to start with.
 
I noticed than in my wifes system a MSI X99S Gaming 7 with a i7 [email protected] 1.18v, and 32GB Corsair LPX 2666MHz CL15 ram using the XMP profile the SA Voltage is Auto set at +0.384v, but the VCCIO voltage is ALSO auto set at 1.198v, ! but on my rig that value isn't touched by the BIOS not even at Auto settings ?
 
Last edited:
Meh I haven't time to fiddle with my systems so much anymore, so I've settled for 2666@2400MHz CL13-15-15-39 at 1.35v, SA Voltage at Auto (+0.384v) and the CPU at 4.2GHz at Auto vcore (1.25v) now. It runs stable and 7-9-13 no cold boot issues.
 
Back
Top Bottom