Troubling instability OC'ing my 5820K :(

Goose

Goose

Goose

Soldato
Joined
18 Oct 2002
Posts
4,791
Location
Kent
I got said CPU a couple of weeks ago along with an Asus X99 A-II motherboard and OC'd it to 4.4ghz @ 1.25v. After about an hour of Realbench everything was fine and i've done some Aida64 tests that were also fine as well as Heaven whilst tweaking my GPU OC. I haven't had the time to game much but recently re-installed Skyrim with the intention of modding it heavily.

After playing Skyrim for about 10 minutes my PC reset itself which was rather alarming and initially i thought it was my GPU OC' so i backed it off and tried again. Again after about 10 minutes in Skyrim the screen went black and became unresponsive so had to turn the system off completely. When booting back up i got a message in POST saying my overclock had failed and I need to enter the UEFI and sort things out. I do so and reduced the clocks to 4.2ghz but still it crashes in Skyrim after 10 minutes or so but not after 30 minutes of Realbench. In the end i went back in to the UEFI and reset to system defaults and now Skyrim is fine with no crashing.

The troubling part is that when it crashes in Skyrim Riva OSD is only showing about 13-20% CPU load with temps of 60c and yet it's stable in RealBench with 100% load.

I'm really confused and since im a noob OC'er I really dont know how to troubleshoot the problem. FWIW I did not enable XMP and my PSU is an EVGA G2 850W that powered my previous system faultlessly.
 
Last edited:
I've just had the exact same issue on the same motherboard and it turns out one of my four G Skill Trident Z memory sticks has failed. Symptoms were exactly the same as yours, stable for hours then random resets and blue screens when not doing much.

Try each memory stick in slot A1 on its own and run memtest86. Three of mine were 100% stable with the fourth failing in 2 seconds. RMA time!
 
Just ran Memtest for an hour with all 16gb installed and there were no errors. How long is a full test meant to take?

Also, could someone explain this to me -



Is it a UEFI bug?
 
Last edited:
I just ran Realbench at 4.4hgz and 1.28v and it failed after 42 minutes. I guess that doesn't prove anything one way or the other except that I dont really know what i'm doing.

I'll start testing each RAM stick individually tomorrow and hope one of them fails. If it doesn't then I really wont know what the problem is besides maybe having a rubbish CPU (it's a J batch chip).

:(
 
Do you know bsod code?? Will be something like 50, 124 or 101 for example, try running a program called whocrashed.
What memory sticks do you have? List code/part number
Go into monitor section of your bios and list main voltages or post screen,
Vcore?
Cache/ring?
SA volts?
IO volts?
Dram volts?
 
Nickolp1974 said:
Do you know bsod code?? Will be something like 50, 124 or 101 for example, try running a program called whocrashed.
What memory sticks do you have? List code/part number
Go into monitor section of your bios and list main voltages or post screen,
Vcore?
Cache/ring?
SA volts?
IO volts?
Dram volts?
Click to expand...

There's no BSOD's just Skyrim either resetting or locking up the PC or Luxmark.exe failing in RealBench.

CPU vcore is 1.28 Fixed and every other voltage left to Auto. RAM is Corsair 2666mhz Vengeance LPX.

This is the only crash picked up by whocrashed.

Crash dump directory: C:\Windows\Minidump

Crash dumps are enabled on your computer.

On Sun 17/07/2016 15:26:11 GMT your computer crashed
crash dump file: C:\Windows\Minidump\071716-9828-01.dmp
This was probably caused by the following module: ntoskrnl.exe (nt+0x6756CD)
Bugcheck code: 0x124 (0x0, 0xFFFFE0018444D8F8, 0x0, 0x0)
Error: WHEA_UNCORRECTABLE_ERROR
file path: C:\Windows\system32\ntoskrnl.exe
product: Microsoft® Windows® Operating System
company: Microsoft Corporation
description: NT Kernel & System
Bug check description: This bug check indicates that a fatal hardware error has occurred. This bug check uses the error data that is provided by the Windows Hardware Error Architecture (WHEA).
This is likely to be caused by a hardware problem problem. This problem might also be caused because of overheating (thermal issue).
The crash took place in the Windows kernel. Possibly this problem is caused by another driver that cannot be identified at this time.
Click to expand...
 
I was on 1.28 when it failed Realbench and if it was vcore related I don't understand why it would crash in sky rim after 10 minutes under light load. I'm currently in the pain staking process of testing each stick of ram with memtest which seems to need a good 3 to 4 hours for each stick doing 4 passes.
 
I've always found although realbench stable, gaming stability always required a couple more notch's of vcore. Do you always have riva osd open when testing?? If so try without as the may be an issue there.
For 2666mhz mems you'll probably need around 1.05v for SA, whats yours reading at in bios?
 
Each stick of RAM passed 4 passes of Memtest and I decided today to go back to basics and do lengthy stability tests. I reset everything in the BIOS to get everything back to stock except that i downclocked my memory to 2100mhz. I ran Realbench for 4 hours and then Aida64 for another 4 hours and everything ran just fine. I'll do the same tomorrow but OC to 4ghz and put the vcore at 1.29ish and see how i get on and if it passes i'll up it to 4.1 etc etc. I dont really want to push the vcore much higher purely for longevity reasons and once i'm happy enough that i have a stable OC i'll put the memory back to stock clocks and see what happens.

Thanks to everyone who has replied so far.
 
Set your memory to xmp or whatever before testing your OC or you may find instabillity again as you may need more vcore with the higher mem frequency. Seriously no need to test for 8hrs!! 30 mins aida and a few runs of realbench, then just use your pc.
 
I'm just being extra careful.

I do have another query now though. I just set the vcore to offset mode with a value of +0.28 and when i started to run Aida64 @ 4.0ghz the vcore was running at 1.34(!) which is way higher than i was expecting. I thought with +0.28 the vcore would go to 1.29 at the very most. Have i misunderstood how Offset mode works or is my motherboard doing something whacky?

Thanks
 
Test your oc with fixed vcore first! I rarely use offset but iirc its just your VID +/- your offset value. Loadline calibration will play a part too.
At 1.29v you should be at 4.4-4.5ghz with ease unless your cpu is a bad sample
 
Try this

Cpu multi x44
Cache multi x40
Memory XMP
Vcore 1.29v(fixed)
Ring/cache voltage 1.25v
SA volts 1.08v
IO(D) volts 1.12v

If you try the above clear your cmos first

Oh and if you try just check that your still on 100strap after pressing xmp, if not change to 100 and set your mems to 2666mhz
 
You must log in or register to reply here.
Back
Top Bottom