LGA1156 Core i7 & i5 Overclocking Guide

[andy5174]

I have my voltage set in my Bios to 1.225V. Every time my computer loads cpu-z shows voltage to my processor at 1.232. The system is stable right now as i finished memory test and p95. But i don't understand why why the difference in voltage is there.

Second is temperatures. Running P95 my temps were upper 50ties / lower 60 ties with this OC. I'm using the 212+ and already resat the heatsink twice to try to lower these. When people are pushing 3.8-4.0 what temps are they experiencing. I know i can lower them a bit by turning up my case fans, but like to know what people are getting at various clocks.

1. It's the voltage protection feature called "Drooping voltage" which will make your full load voltage lower than idle/BIOS voltage. If there's a Load-Line Calibration(LLC) option in your BIOS, you can DISABLE it to have your voltage the same or very close to the BIOS voltage.

2. The core temperatures are dependent on the CPU heatsink, CPU fan, thermal compound, PC case, ambient temperature and motherboard. Hence, you can only get useful temp info from peoples who has the SAME CPU heatsink, CPU fan, thermal compound, PC case, ambient temperature and motherboard as you which is pretty hard to find.

 

[IamShakes]

1. It's the voltage protection feature called "Drooping voltage" which will make your full load voltage lower than idle/BIOS voltage. If there's a Load-Line Calibration(LLC) option in your BIOS, you can DISABLE it to have your voltage the same or very close to the BIOS voltage.

The odd thing is my idle voltage is reading 1.224V, while running Prime-95 voltage is reading 1.232.

The mobo is the new ASUS P55-E Pro. I'm not sure if the mobo is ignoring the setting and just providing the voltage of it cpu-z is reading incorrectly.

Its a mild OC so its understandable why its stable, but i like to understand what is going on.

 

[andy5174]

The odd thing is my idle voltage is reading 1.224V, while running Prime-95 voltage is reading 1.232.

What's the LLC status of your MB? Enabled, Disabled or Auto?

Some MBs enable LLC with AUTO when they detect OC. For example, my MB enables LEVEL 1 LLC automatically with AUTO.

It is normal for load voltage higher than that of the idle with LLC enabled.

BTW, it's good to be with LLC ON as it makes your load voltage extremely stable with minimal to none variation. You can run OCCT for 60 minutes with LLC ON and OFF, then monitor the voltage variation graphs generated by it at the end of the test if you don't trust me.

 

[andy5174]

In order to have extremely stable voltage at full load, set the LLC to the highest level if there is more than one levels like my Gigabyte MB.

You can run OCCT for 60 minutes with LLC ON and OFF, then monitor the voltage variation graphs generated by it at the end of the test if you don't trust me.

 

[Cattykit]

Before i ask, wanted to say thanks, been stalking here for months to make my decision on setup. System is all built and have a mild OC on it right now, but just a couple of questions i don't know the answer to.

I currently am OCed to 3.2 with EIST and C-states enabled. I have my voltage set in my Bios to 1.225V. Every time my computer loads cpu-z shows voltage to my processor at 1.232. The system is stable right now as i finished memory test and p95. But i don't understand why why the difference in voltage is there.

Second is temperatures. Running P95 my temps were upper 50ties / lower 60 ties with this OC. I'm using the 212+ and already resat the heatsink twice to try to lower these. When people are pushing 3.8-4.0 what temps are they experiencing. I know i can lower them a bit by turning up my case fans, but like to know what people are getting at various clocks.

My i7-860 with hyper 212+ @ 3.7ghz that eats up 1.288 vCore wil go all the way up to 72c. This is using Linx which I think stresses CPU more than Prime does.
Keep it mind HT makes so much difference. I've seen the difference between HT on and off was as large as 15c.

 

[Cattykit]

For someone that requested it. Here's a heatsink performance chart that I favor the most. I favor it because of following reasons:

1. He actually buys stuffs he review from different retail channels. So, nothing is cherry picked and provided by companies.
2. Tests are done using professional mesurement items.
3. He tests each heatsinks inside of a case with proper cable management as you and I would do.

http://down.playwares.com/bbs/data/mutiupload/image/f16d82ea7d5f4f286da78893c173eac7.GIF
http://down.playwares.com/bbs/data/mutiupload/image/01afe01bfde76a8be9fc1e35d4c607ef.GIF
http://down.playwares.com/bbs/data/mutiupload/image/57b745567ebe926e0befcea01c26ec49.GIF

 

[Gillbot]

In order to have extremely stable voltage, set the LLC to the highest level if there is more than one levels like my Gigabyte MB.

Be careful with this setting: http://forums.anandtech.com/showthread.php?t=271760

 

[imported_RJR]

Gillbot, this LLC debate has been ragging for some time now and everyone seems to have very logical and scientific theories for their side of the issue but I've yet to see any scope data (not illustrations) of the effect to prove it one way or another.

I've got an Asus P35 MB that doesn't have the option of LLC in the Bios, hence I did a pencil mod to take care of my massive vdroop. It has been running for over 2 years now that way and I haven't had any problems yet. That's not to say it won't die tomorrow but I would be interested in seeing actual scope shots of an actual processor run in different vdroop configurations. My newest computer (i7 860/UD4P), I did enable LLC because the vdroop/vdrop is about .12v combined, and that's just a little too much for me.

PLEASE, lets NOT start an LLC debate here, but if anyone has a scope and wants to map it out that would be great.

 

[Gillbot]

If I had access to a scope, I'd love to but the theory still stands and the phenomena is in fact real. I only want the less informed to be aware because many will follow a sticky to the letter without realizing the possible consequences of the settings they are adjusting.

 

[Munky]

I'd be careful with suggesting upping the voltages too much too soon. Higher voltage generates more heat, which may actually lead to more instability than lower voltage. In addition, it shortens the lifespan of the cpu. I would instead suggest gradually increasing clocks, testing, and increasing voltage only if you're well within safe temp limits, and never going over the specified max voltage if you plan on keeping this cpu for 24/7 use.

In my own experience with the i7 860, I initially changed/disables too many parameters, raced to 3.8ghz, only to find out all sorts of instability issues, even when I later backed down to DEFAULT clocks. So, I started from scratch, making only minimal changes and adjustments (vcore, vtt, mem timings, left turbo enabled, LLC on auto, PCIE spread spectrum on auto) and found that I could reach a stable 3.6ghz using a much lower voltage than I originally tried.

 

[andy5174]

I'd be careful with suggesting upping the voltages too much too soon. Higher voltage generates more heat, which may actually lead to more instability than lower voltage. In addition, it shortens the lifespan of the cpu. I would instead suggest gradually increasing clocks, testing, and increasing voltage only if you're well within safe temp limits, and never going over the specified max voltage if you plan on keeping this cpu for 24/7 use.

In my own experience with the i7 860, I initially changed/disables too many parameters, raced to 3.8ghz, only to find out all sorts of instability issues, even when I later backed down to DEFAULT clocks. So, I started from scratch, making only minimal changes and adjustments (vcore, vtt, mem timings, left turbo enabled, LLC on auto, PCIE spread spectrum on auto) and found that I could reach a stable 3.6ghz using a much lower voltage than I originally tried.

Although I never heard that high voltages could cause more instability, but it is an interesting idea and your idea makes sense.

Hmmmmm, what I believe is that the more voltages given, the more stable the systems are as long as those voltage values are within their MAX specification by Intel.

In addition, I suggested to start at the MAX voltage because noobs(who are the ones actually benefit from this guide) can avoid having to CLEAR THE CMOS due to system instability caused by IMMENSELY insufficient voltages while intending to achieve MAX overclock.

Thanks for your input which is very well related to the guide, interesting and makes sense!

 

[Gillbot]

Although I never heard that high voltages could cause more instability, but it is an interesting idea and your idea makes sense.

Hmmmmm, what I believe is that the more voltages given, the more stable the systems are as long as those voltage values are within their MAX specification by Intel.

In addition, I suggested to start at the MAX voltage because noobs(who are the ones actually benefit from this guide) can avoid having to CLEAR THE CMOS due to system instability caused by IMMENSELY insufficient voltages while intending to achieve MAX overclock.

Thanks for your input which is very well related to the guide, interesting and makes sense!

The thought is usually, the more voltage, the more excess heat that is generated and that is what causes instability.

 

[andy5174]

The thought is usually, the more voltage, the more excess heat that is generated and that is what causes instability.

OK.

I should had said that the more voltages given, the higher the frequency peoples can achieve instead. That's what I meant high voltages will make your system more stable.

Sorry for not being able to clearly express my idea with my bad English.

 

[Gillbot]

OK.

I should had said that the more voltages given, the higher the frequency peoples can achieve instead. That's what I meant high voltages will make your system more stable.

Sorry for not being able to clearly express my idea with my bad English.

More voltage doesn't always mean a higher clock, every chip has a threshold where more voltage only adds heat and not more clock.

 

[andy5174]

More voltage doesn't always mean a higher clock, every chip has a threshold where more voltage only adds heat and not more clock.

Of course! Otherwise, people can hit infinite OC and all OC guides are useless because all that you need is upping the voltage.

That's why I mentioned maximum OC in 1) Decision making on the OC of my OC guide.

That's why people want to find the minimum stable voltage.

 

[andy5174]

Are most members here in the U.S.?

EDIT: How come there are always so few people most of the time when I am awake?

 

[Markfw]

Gillbot, this LLC debate has been ragging for some time now and everyone seems to have very logical and scientific theories for their side of the issue but I've yet to see any scope data (not illustrations) of the effect to prove it one way or another.

I've got an Asus P35 MB that doesn't have the option of LLC in the Bios, hence I did a pencil mod to take care of my massive vdroop. It has been running for over 2 years now that way and I haven't had any problems yet. That's not to say it won't die tomorrow but I would be interested in seeing actual scope shots of an actual processor run in different vdroop configurations. My newest computer (i7 860/UD4P), I did enable LLC because the vdroop/vdrop is about .12v combined, and that's just a little too much for me.

PLEASE, lets NOT start an LLC debate here, but if anyone has a scope and wants to map it out that would be great.

I would really like to see some of these "discussions" left out, but I guess that isn't going to happen. I for one thing that have the most stable vcore, load or not (like my DQ6 boards) makes for the most stable overclock, even if it spikes. So long as you don't go crazy in bios, the spike should be find, if thats what you all are worried about.

 

[Gillbot]

I would really like to see some of these "discussions" left out, but I guess that isn't going to happen. I for one thing that have the most stable vcore, load or not (like my DQ6 boards) makes for the most stable overclock, even if it spikes. So long as you don't go crazy in bios, the spike should be find, if thats what you all are worried about.

Unfortunately, anytime you have a guide, there will always be "discussions". Just skim through my "Sysprep" guide and you'll see quite a few in there also.

IMHO, Anytime anyone posts a "how to", people always have to chime in with their settings to prove their way is superior to the OP's.

 

[GLeeM]

I would instead suggest gradually increasing clocks, testing, and increasing voltage only if you're well within safe temp limits

As a beginner OCer this is sort of what I did. I was too afraid to increase voltage to start off.
Actually I set the OC to a medium OC and checked how low I could get the voltage and be stable. Then I started increasing BCLK and only enough core voltage to keep it stable. When I got to where temps were too high (mid-upper 70s) I lowered OC to where I was comfortable with the temps and am happy.

Although I can see where your way of doing it would save a lot of time. But then I never went over 1.3 vcore and so felt safe.

 

[andy5174]

To achieve the absolute maximum OC, begin with applying the Maximum voltages(to be safe) or Absolute Maximum voltages(at your own risk) in BIOS.

I suggested that people start with maximum voltage IF they want to achieve max OC.

Since clearing CMOS will be required(which could be a tough job to noobs) if the voltages set are immensely insufficient to the OC they are trying to achieve, I made such an suggestion ASSUMING they have one of the top end cooler which is ESSENTIAL for this kind of insane OC.

In addition, I do mention the temperature monitoring in the torture test section for your concern. Besides, you will be fine even with high temperatures as your system WILL shut down the PC for you QUICKLY enough to avoid permanent damage to your components.

By the way, I prefer a mild OC with reasonable voltages as well which is the reason why I am at 3.6GHz instead of something above 4GHz.

 

[imported_RJR]

COMMON we need some VOTES, yes or no for sticky?

 

[lothar]

I vote no.

 

[Stefan Payne]

I voted no, 'cause it's missing some safety warnings and it's going too far.
And it can't be applied to any CPU, so a guide to overclock isn't bad, some warnings would be nice.

Remember, that there are (too many) users out there who don't think mutch but try everything...
So it could happen that there is someone out there, who sets 1,65V for the CPU, 2V for memory and just tries...

 

[Gikaseixas]

my 860 does 4.0 with only 1.295 vcore, i couldn't be happier

 

[andy5174]

my 860 does 4.0 with only 1.295 vcore, i couldn't be happier

LLC on, right? Still very good even if it is the case.