Loosen timings ? Or not ?

Associate
Joined
12 May 2010
Posts
1,877
Location
West Bromwich
I'm trying to get a stable OC on a recently purchased 2500K with some vengance black 8-8-8-24 1.5V

If I heavily overclock the CPU (say up to 4.8Ghz @ 1.4V) do I need to loosen the timings on the memory like you had to on DDR2? Or perhaps up the voltage to 1.55?

I'm getting stability in prime but odd blue screens that occur just when I'm browsing the internet at home with no load on the CPU whatsoever.
 
It may be a factor, but usually its the Vcore being to low giving instability. The only time you probably need to give the ram and controller a bump is if you are running 4 sticks of ram.
Have you enabled PLL override?
There is an article I read giving error codes of blue screens and the cause (eg Vcore, mem controller etc)
Let me try and dig it out.
 
PLL overide is enabled yes. Based on your recommendation I wont touch the voltage, I'll loosen the timings up a bit and see if it makes any difference. Thanks for the link by the way.
 
TBH, it sounds like your clock isn't stable and your memory settings are unlikely to be the cause.

Just to check - have you clocked manually through the BIOS or are you using an OCing utility?

If it is manually i would consider dropping 100MHz and geting it stable at 4.7GHz and them move on from a stable foundation/settings.

If it is via utility - bin the clock and start again manually.
 
Then, personally, i would either play around with other settings - but increased vCore is the obvious choice but keep in mind that you're on the limit with this already. What is the vCore under load?

Or, ideally, drop the clock to 4.7GHz, keeping other settings as they are, and test for stability.

There are other settings you can try without touching the vCore but without knowing what you've changed/adjusted already it's difficult suggest what you could try.
 
I know it's a silicon lottery and all that but I find it hard to believe I am at absolutely maximum voltage and not even getting a stable 4.8. This is why I thought it might have been memory related. 4.6 is stable at 1.34.

Oh well, guess I bummed out on this chip.
 
IThis is why I thought it might have been memory related. 4.6 is stable at 1.34.

Unfortunately, that's not uncommon - there's lots of users who suddenly have to pump way more vCore (relative to their 4.6GHz clock) into the 2500K when trying to hit 4.8Ghz.

I wouldn't lose all hope there may be a combination of settings out there that may resolve your problem. Try a search in the overclocking section.

However, 4.6GHz at 1.34V is a respectable combo...
 
Ok so I found a Prime Stable frequency of 4.6Ghz that needed 1.37 volts. This ran for 12 hours yesterday with no errors.

I left the computer on this morning, went to work, then tried to access it remotely from work and it wouldn't respond. Got home tonight and it had blue screened, same error message. Interupt exception not handled.

Grrrrrrr!
 
My cousin had a similar problem - with his p67 mobo
he could prime all night but woulkd still get odd blue screens
He found a forum that suggested changing a pci setting and that seemed to help him

I will speak to him later and see if I can find out ehat he did / where he found it
 
my cousin said he read and has turned the c states off in bios

He has kept 1 on so that his cpu with throttle when not under load (using an offset oc)

let me know if that helps mate??
 
Yeah I'll give it a go when I get home, also going to update Bios to 0402.

I read last week that C1E should be kept on in Sandy setups so left them switched on.
 
Put your CPU back to stock speed and stress test your ram, an easy way to do this is to triple run Prime 95, Linx and Super Pi 32m. Super Pi would error right away with this much load on the ram, but if it completes a 32m run under that much system load, your ram is fine.

You always should overclock and stress your ram and CPU separately to make sure that each one is stable.
 
I am priming now so cant check my bios but I am sure I have only 1 CE States turned on - not the others
C1E throttles your cpu rather than run it at full volts all the time - or similar - keeop that 1 on try the others off

I also found things to be much more stable with cpu spread spectrum disabled in case that helps you

CPU PLL voltage at 1.700 rather than auto helps with stability of an oc also on p67
 
Last edited:
This is really starting to annoy me now....

I updated the BIOS and it 'froze' the Vertex 3. Have only just been able to gain access back to it.

I saved all my game files and did a re install after updating to BIOS 0402. Everything is currently at stock. I re installed Windows after backing up my save game files and then after the re install I tried to install the Nvidia drivers and the frickin thing BSOD'd on me at stock settings. So yeah, it was nothing to do with my original overclock.

On my previous install I ran memtest and it found no errors. I ran prime for 12 hours with no errors, its just that when the machine is ticking over it has the tendency to BSOD with the error I quoted in my earlier post.

All you get when you google is it bad drivers or hardware, and apparently there is very little information available on it because it is so rare.

Really loosing my temper now.

EDIT: I triple ran Super PI, Prime 64 bit, and Linx, I monitored memory usage throughout, max was 99% and not once did the machine crash.

Have just downloaded OCZ toolbox and it isn't even detecting I have a vertex 3 in my machine?!?!
 
Last edited:
Back
Top Bottom