Overclock
- Thread starter CSheldon
- Start date
jamesriley94
New member
I'd normally go low and work your way up until stable tbh.
You need to be stressing for around 12+ hours on Prime95 for stability. If for example you could get stable at 0.95v, you're gunna be there for days.
I'd start at about 0.95v, work your way up with 0.025v each time maybe? That should get you into the ballpark, then when you've got stable, bring it down by 0.005v and find the lowest voltage you can get stable at.
You need to be stressing for around 12+ hours on Prime95 for stability. If for example you could get stable at 0.95v, you're gunna be there for days.
I'd start at about 0.95v, work your way up with 0.025v each time maybe? That should get you into the ballpark, then when you've got stable, bring it down by 0.005v and find the lowest voltage you can get stable at.
Right at the moment, running OCCT, at 90% memory, using 4bits and AVX Capable Linpack on stock clocks (Turbo On)
Cores have hit 62/68/64/61, at max. Is this normal for an undervolt at 1.1v?
And after that, should i just bring it down to 1.095V, then 1.090V, or should i go down by 0.010V, and then once unstable, bring it up?
Cores have hit 62/68/64/61, at max. Is this normal for an undervolt at 1.1v?
And after that, should i just bring it down to 1.095V, then 1.090V, or should i go down by 0.010V, and then once unstable, bring it up?
jamesriley94
New member
Temperature wise, it depends what cooling you have obviously... But those temps are nothing to be worried about.
I was suggesting taking it down to 0.95v, then taking it up to 0.975, and going from there until stable.
I was suggesting taking it down to 0.95v, then taking it up to 0.975, and going from there until stable.
trapjaw72
New member
over-clock my i5 2500k to 4.5 ghz 1.164 volts.....http://valid.canardpc.com/2829745
KING_OF_SAND
New member
I got to 5ghz on my 2500k. look at sig.
Master&Puppet
New member
Don't worry about voltage fluctuations - that is entirely 'natural'. I'm guessing that you haven't fixed the LLC (which you don't need to touch on an undervolt) and that will mean that under load the voltage will fluctuate slightly. It's known as voltage droop, or Vdroop.
The easiest way is to do what James said. Personally I'd reduce the voltage from 1.000 to 0.975...0.950 etc booting into Windows each time until windows no longer loads. Then add 0.01 to the voltage and start up Prime95 on blend (and manually set the memory to 90% of the available). If it doesn't make 12hrs+ then add another 0.01v and retest etc.
Alternatively if you have an offset control on your motherboard that would be a better option. Set a negative voltage modifier (-0.025, 0.05, 0.075 etc) test the same way, adding 0.01 if you get instability. Just be aware that doing an negative offset can cause your clock to be stable under load but you might actually be unstable when the voltage drops at idle.
Another really easy way of setting an undervolt is leaving the volts on auto but setting the LLC to be really loose. That means that you'll still get the voltage dropping under load, like an offset, but you won't have idle undervoltage issues yet still manage to reduce the voltage at load which is the main thing you want to be doing.
Anyway, I'm blabbing on now. GLHF!
M&P
Netroxen
New member
I'm not going to sound arrogant in any way but I think stress testing Prime95 for 12-24 hours is completely unnecessary unless you require the highest level of stability. If you're watching movies, browsing the web and playing games, your CPU will never come close to what Prime95 stresses the CPU at. At such high load levels, P95 is intended to find even the smallest errors while maintaining a load-level that's very specific.
Therefore, under normal use these errors are unlikely to occur.
Run P95 for 3-10 minutes at a time and keep going until it starts readily displaying errors. Play some games and movies to see whether everything is running snug.
IMO, just my opinion.
Therefore, under normal use these errors are unlikely to occur.
Run P95 for 3-10 minutes at a time and keep going until it starts readily displaying errors. Play some games and movies to see whether everything is running snug.
IMO, just my opinion.
jamesriley94
New member
I don't agree with that.
Certainly not for 3-10 minutes anyway.
I've been stable on prime95 for 5+ hours, and still not been stable enough for gaming.
Also, 3-10 minutes doesn't give enough time for temperatures to properly settle - especially for watercooling where the coolant temperature won't have had significant time to stabilise at its load temperatures.
Fair enough if all you're doing is waterching movies - although then I'd probably argue is there any point in overclocking at all? A lot of us here Fold - ie, 90%(ish) stress 24/7 and so 10 minutes of prime for stability really won't cut it to be sure of stability.
I'd much rather have the peace of mind knowing my computer was stable and wasn't going to bluescreen regardless of what I did, and how long I did it for, whether it was gaming, folding, or watching a film - or something more intensive even like video rendering.
Netroxen
New member
I mean, that's your experience - But from my own I've never seen a need to... The temperatures stabilize on my H60 within a matter of minutes, probably because there isn't a huge amount of coolant in the system.
Right now I have my 3770k at 4.7Ghz, I ran P95 for 20 minutes and called it quits. I haven't had any stability issues in any games yet. So it depends, every PC is unique.
Right now I have my 3770k at 4.7Ghz, I ran P95 for 20 minutes and called it quits. I haven't had any stability issues in any games yet. So it depends, every PC is unique.
jamesriley94
New member
Can I ask what voltage you're using, and what temps you get please?
jamesriley94
New member
You'll degrade the processor at that - and pretty quickly too.
I'd consider turning it down to 4.5/4.6ghz if I were you, and get the volts below 1.3.
Netroxen
New member
Degrade, as in take 2 years off its lifespan? To be honest, I'll probably upgrade by the time its degraded to a point at which it can't sustain an overclock. Even then, I can just periodically reduce the clock speed. Honestly, when I hear people say a processor is going to degrade, I laugh - Well, because it simply isn't substantial enough to be noticeable. Maybe after a year or so, but really... 
Various sources state that at 1.31v the chip starts gradually degrading, although the effects are somewhat non-existent, so with a 0.04Mv I hardly see the silicon degrading "Pretty quickly..." as you said.
At 1.45v+ users on various forums have stated a noticeable reduction in stability over a period of many months, and that's at 1.45v!
Shouldn't get into overclocking if you're scared of pumping some volts in, it's like saying you shouldn't race a car because you're going to crash it one day... Well...
Various sources state that at 1.31v the chip starts gradually degrading, although the effects are somewhat non-existent, so with a 0.04Mv I hardly see the silicon degrading "Pretty quickly..." as you said.
At 1.45v+ users on various forums have stated a noticeable reduction in stability over a period of many months, and that's at 1.45v!
Shouldn't get into overclocking if you're scared of pumping some volts in, it's like saying you shouldn't race a car because you're going to crash it one day... Well...
Last edited:
jamesriley94
New member
Degrade, as in take 2 years off its lifespan? To be honest, I'll probably upgrade by the time its degraded to a point at which it can't sustain an overclock. Even then, I can just periodically reduce the clock speed. Honestly, when I hear people say a processor is going to degrade, I laugh - Well, because it simply isn't substantial enough to be noticeable. Maybe after a year or so, but really...
Various sources state that at 1.31v the chip starts gradually degrading, although the effects are somewhat non-existent, so with a 0.04Mv I hardly see the silicon degrading "Pretty quickly..." as you said.
At 1.45v+ users on various forums have stated a noticeable reduction in stability over a period of many months, and that's at 1.45v!
Shouldn't get into overclocking if you're scared of pumping some volts in, it's like saying you shouldn't race a car because you're going to crash it one day... Well...
Not scared at all about putting volts though my chips....but I had my 3770k at 4.7ghz (1.36v(ish)) and it noticeably started to degrade.
I was able to get 4.5ghz stable at 1.195v when I started, after only a couple of weeks I then needed 1.21v to get stable at the same clock.
It isn't about taking 2 years off a 15 year lifespan or whatever, it's about being unable to keep the same clock speed for volts after only a couple of weeks.
Netroxen
New member
I'll reiterate on what I said before, every setup is unique, whether it be the silicon lottery, the combination of hardware or the BIOS being used - Nothing is set in stone. It's been three months since I bumped my 3770k up-to 4.7Ghz and I haven't noticed any need to increase the voltage. However, my chip is a Cost Rica export - So it's actually part of a relatively good batch. Unfortunately my motherboard is an entry level overclocking board and thus requires more voltage to meet the needs of the clock speed.
There's always a trade-off - But, for now I'm happy. But hey, you should try bumping your chip up-to 4.7-4.8Ghz, feels good man! lol
Also, are you on water or still using the D-14?
There's always a trade-off - But, for now I'm happy. But hey, you should try bumping your chip up-to 4.7-4.8Ghz, feels good man! lol
Also, are you on water or still using the D-14?
jamesriley94
New member
I'm pretty sure all of them were made in Costa Rica? Mine was I know...
If you've only tested for 10 minutes on prime95 - you wouldn't know the lowest volts you'd actually need for it to remain stable. But I'd be pretty confident in saying they'd be higher now than they would have been.
Anyway - your chip - I just wouldn't like to be the one to buy it after haha.
