2600K Overclocking Question

[FRTW]

Hi all! I know I'm a bit late to the 2600K overclocking party, but better late than never. So I have a 2600K on an Asus P8Z68-V Pro motherboard with a Cooler Master Hyper 212+.

 

I'd like to get a moderate overclock of around 4.2-4.3Ghz on as close to the stock voltages as possible. I ran Prime95 for about 3.5 hours last night at stock and Cpu-Z was reporting voltages dancing around from about 1.264-1.288 for the 3.8Ghz turbo. The max temp that HW Monitor reported was 63C.   

 

So I'd like to know if anyone thinks that it is possible to get ~4.3Ghz on ~1.28V. I don't want to go above 1.3V, just cause I'm a bit of a freak about low temps. If I can get that overclock on stock voltage with the same(ish) temps, does that mean it is 100% safe to run 24/7?   

 

Thanks.

[Sauron]

It's safe and better if you can pull it off

[Speedbird]

You have to try and see. Go to the BIOS and set the CPU multiplier to 42 and run Prime95 for a day. Check the temps too and see if you like them. I understand you want to use the stock voltage, so if the temps go too high, lower the CPU multiplier.

[Windows98SE]

 If I can get that overclock on stock voltage with the same(ish) temps, does that mean it is 100% safe to run 24/7?   

 

 

Uh....yes?

 

And yes, you can probably get 4.2 on stock voltages. BUT no two chips are the same so you'll have to find out for yourself if it will work or not. 

[SkilledRebuilds]

4.2Ghz isn't too bad of an aim,.. 4.0-4.2Ghz instead of 3-5-3.9 is still quite nice to use for your Turbo Multi as the temps will stay low and no doubt I think easy to do on stock voltage & a nice extra bit of performance.

High performance power profile set in Windows, set your fan speeds and your good to go.

[FRTW]

Yeah. I'm still a bit of noob when it comes to overclocking and I'm probably too cautious, but I've got prime running right now at 4.3. Only like 45 minutes in, but so far so good. 

 

Thanks for the advice. 

[airdeano]

I ran Prime95 for about 3.5 hours last night at stock and Cpu-Z was reporting voltages dancing around from about 1.264-1.288 for the 3.8Ghz turbo.

 

look into the LLC (Load Line Calibration) from auto/normal to extreme or the

highest setting. the Z77 and older platforms didn't have near the digital control

granularity as today mobos have.

 

and really no need in monster time stressing, if it passes a 30-min test, then

proceed to more settings. then once stability is found, then use real-world appz

to test stability.

[FRTW]

look into the LLC (Load Line Calibration) from auto/normal to extreme or the

highest setting. 

 

Hmm. Correct me if I'm wrong with any of this, but as I understand LLC is to compensate for Vdroop, which adjusts the voltage to stay within Intel's TDP. Overclockers adjust it to stop the CPU from essentially throttling itself when it wants pull more power.

 

So if were to increase LLC above regular (I changed it from auto) in theory I would see the voltage dancing less? I'm going to assume that having a steady voltage at full load (prime95) is a good thing, but I know some overclockers swear by LLC and others say to avoid it. 

 

I'm open to trying it, but can anyone explain the pros/cons in a bit more depth?

[FellTheSky]

i have 2500k at 4.5ghz 1.34 v.  24/7 two years already, never one bluescreen. Just to give you some perspective

[FRTW]

What kind of cooling do you have and how are the temps? I know my Hyper 212+ is good, but that seems like it might a little too much for my cooling solution. 

[airdeano]

So if were to increase LLC above regular (I changed it from auto) in theory I would see the voltage dancing less?

 

yes, the voltage would be steady. as the load is applied, the LLC acts as a

power valve to supplement the voltage above noted algorithm on the range

of tolerance. most would dabble as the increased LLC did nothing to add more

heat or detracted from and stability. a 0.08v swing is usually harnessed. OC

users of the offset will not use LLC as then the overall voltage is too high.

manual OC is where LLC is needed.

[FRTW]

 OC users of the offset will not use LLC as then the overall voltage is too high.

manual OC is where LLC is needed.

 

I haven't played around much with the voltages (since I'm trying to get to 4.3(ish) on stock voltage) other than changing LLC to regular. I'll check in a little while, but I believe it is set to offset mode. 

 

look into the LLC (Load Line Calibration) from auto/normal to extreme or the

highest setting. 

 

 

Would you still recommend increasing LLC since I'm using offset. I don't want to use manual because I want the voltage to drop when speed step lowers the clock speed. 

[airdeano]

Would you still recommend increasing LLC since I'm using offset. I don't want to use manual because I want the voltage to drop when speed step lowers the clock speed. 

 

you'd have to watch your over-voltage as the application of voltage is offset from

a predetermined config (meaning this is it!) and then LLC adds on top.

manual voltage can be flexible.

 

voltage drop on idle is not from offset, but from speedstepping and CSTATES and

if the OS is using a balanced profile or performance profile.

 

due to the refresh rate of CPU-Z, voltage report can be late, a DMM and

measuring points are more accurate.

[FRTW]

 

voltage drop on idle is not from offset, but from speedstepping and CSTATES and

if the OS is using a balanced profile or performance profile.

 

Just tried using extreme LLC with the multiplier at 43 and a manual voltage of 1.280. It boots into windows fine, but even on a balanced or power saving profile the voltage never changes. All the CSTATES were enabled. What am I doing wrong?

 

---- EDIT ----

Gone back to offset mode and regular LCC, testing prime now with 1.256-1.264 volts. Temps are down on average about 5C from 1.280V, so I'm just going to keep on backing the voltage down until it isn't stable anymore.