2500k locks up when doing ... nothing

X

xorbe

Supreme [H]ardness
Joined
Sep 26, 2008
Messages
6,034
It's a 2500k oc'd to 4.5 on a Asus P67 mobo, Seasonic PSU, Crucial m4 SSD, Win7, and it has been doing this since day #1 (quite a while ago). About once a month, it generally freezes while idle overnight. Unresponsive, clock is frozen on screen, etc. I've been through a 570, 680, and Titan on that machine so it's not the video card. It never freezes while gaming or under load. It never blue-screens. Over time I lowered the oc from 4.8 to 4.7 to 4.6 to 4.5 then nudged the voltage up twice. No effect. I know it runs up to 5.1~5.2 because I played the entire game of Quake1 (software rendered) that way.

But recently it's started happening like once a week, and a bit more aggressively. I can surf the web for a while, step out for a few minutes, and when I come back it's frozen idle at the desktop. Nothing in the Windows event viewer, so it seems to be hardware related.

I did some Googling and thought it might be the Realtek driver due to some winsxs dll breach in the event logs, but even with Realtek uninstalled and the dll breach gone (due to an older buggy Realtek driver), it still happens.

Short of dropping oc's back to stock, is this a known symptom of something common I might be overlooking. My best guess is after idling a while and during a cooler state, the vdroop on the idle to load transition nails my particular cpu. Probably disabling idle state pegging the cpu at 4.5 would fix it, but my PC idles a lot. Hmm, I've talked myself into examining the load-line setting.
 
Last edited:
For testing purposes I would drop the CPU back to stock. I would suspect the CPU last as being completely faulty and running at stock is the best way to assume it's OK.

From there you're looking at any number of possible hardware causes.
 
I would also check power states, since your locks are happening at idle, perhaps sleep/hibernate/hybrid is broken. OR, you could join our Folding team, and never be idle again!
 
Sounds like an overclock thats not 100% stable.
I've had exactly the same thing you are experiencing and ended up using a lot higher vcore than I wanted to get it stable all the time.
If I were to leave the PC on for 8 hours a day every day, I might never have noticed or put it down to something else.
Running 24/7 it would crash occasionally at idle unless I used higher voltage.

I couldnt get it to fail using stability tests as these keep the CPU pegged with no change from idle to load state and back.
Its most likely due to the change from idle to load state and the voltage not ramping up fast enough by a hair.

Asus users have a setting that allows their system to change voltage quicker, cant remember what its called though.
I suggest you research this.
This allows you to use lower vcore for the same overclock 'when this is the problem'.
 
Yes, this is common when your vcore is set too low and your LLC set too high. LLC is really nothing more than increasing vcore anyway, only difference is that it only kicks in when the processor is under load, making it unstable when your vcore is too low.
 
Nenu said:
...Asus users have a setting that allows their system to change voltage quicker, cant remember what its called though.
I suggest you research this.
This allows you to use lower vcore for the same overclock 'when this is the problem'.
Click to expand...

Found the setting, its called VRM fixed frequency mode, I first saw it here on this Asus P67 motherboard guide, 4th image down.
http://hardforum.com/showthread.php?t=1578110
 
Thanks all, I updated some of those settings in the bios ... now to wait 1~8 weeks.
 
Had the same problem, ended up being the motherboard which was a Gigabyte UD4 P67. Cant remember if i tested with stock settings or not, only with high OC on 2500k. Microcenter let me exchange it for an Asus Z77 which now works great and the cpu is happily humming along at 5.0ghz.
 
You must log in or register to reply here.
Back
Top