• |

How To Safely Overclock and Stress Test Your CPU

Posted on Saturday, June 14, 2014 | 10 Comments


As most PC enthusiasts and builders have noticed today, CPU makers have made overclocking a CPU pretty easy in comparison to days past. It is no longer strictly a "geek affair"  to overclock a CPU. For the most part, the days of messing with raising the FSB(Front Side Bus) which also messes with RAM timings, chipset timings, etc are over for casual overclocking.

Today, CPU overclocking is usually as simple as raising the multiplier and testing it. AMD and Intel both offer unlocked "K Editions" (Intel) and Black Editions (AMD) that make overclocking a relatively easy process. Infact, Pretty much every AMD CPU you buy nowadays are the unlocked variety, while Intel usually charges a slightly higher cost for their unlocked versions. However, if your were already going to buy the Intel CPU, your usually better off to get the unlocked version anyways.

That being said, below I will outline "my best practices" for overclocking and stress testing CPU's that I believe to reasonable and successful. This guide will go against the grain in the stress testing department simply due to my years experiences, trial and error, etc and what I have found works best for me. I hope this guides gives you a good baseline to work from.

Disclaimer: I am not responsible for anything that may happen to your PC when overclocking. If you choose to overclock, you do so at your own risk.. You are solely responsible for any damage to your CPU or other hardware due to user error, inadequate cooling, too high of voltages, and any other factors. Please remember to back up your computer before attempting this. Do not Overclock on the stock AMD or Intel CPU Heatsink and fans. Use Aftermarket cooling heat sinks of sufficient TDP or water cooling to ensure best chance of not having premature hardware failure.

Voltage And Temperature

Before we get started, lets talk about voltage. This is the single most important thing to understand before you begin overclocking, messing with the multiplier, or changing and adjusting the CPU Vcore(Voltage).

First thing to do is find out the max temp and recommended voltage levels for the CPU you are planning to overclock. The Intel Ark will give you all the relevant information concerning any Intel processor. The AMD Product Pages will give you the same information concerning AMD processors.

Its also a good idea to cross-reference your CPU on the technical specifications at CPU-World  which is a very reliable resource of processor information as sometimes the AMD product Pages is sometimes missing information on certain models.

DO NOT USE THE AUTO-OVERCLOCKING FEATURE OF YOUR MOTHERBOARD BIOS OR SOFTWARE.

These auto-overclocking features, while they sound great on paper, tend to supply more voltage then is necessary and can damage your CPU if your not careful.

Toms Hardware had a great write up on this in their Man Vs Machine Automatic Overclocking Article. where they found in their research that:
Our overclocking articles often mention a process called electromigration” where material is physically transferred from one part of a circuit to another. While the full description of this phenomenon is complex, it’s easy to understand that an insulator contaminated with conductive particles no longer insulates. Transistor gates function as either insulators or conductors depending on charge state and are particularly prone to this type of damage. And yet, many technology enthusiasts place the blame for a fried processor or GPU solely on heat, ignoring the fact that voltage is a measure of force.

Force causes electromigration, and colder silicon more easily resists that force by being less pliable. Colder temperatures also increase the insulation capabilities of transistor gates in the “off” phase, reducing the number of electrons that are forced through the closed gate. The  problem with blaming heat alone on a failure is that moderate increases in electromigration resistance usually require drastic temperature reductions. When it comes to protecting hundreds of dollars in equipment, we always make our recommendations to you erring on the side of caution.

We've learned through trial, error, and dead processors that voltage levels beyond 1.45 V at above-ambient temperatures can kill an Intel CPU etched at 32 nm (Sandy Bridge-based parts included) very quickly. Those same processors die a fairly slow death at voltage levels between 1.40 V and 1.45 V (somewhere between weeks and months on our test benches). And we're expecting more than a year of reliable service from the parts we've dutifully kept below 1.40 V. Not all motherboards are perfect however. Voltage instability on a particularly cheap motherboard fried one of our processors when it was set to only1.38 V. Subsequently, you've seen us use 1.35 V for the overclocking tests in older motherboard round-ups, embracing 1.38 V to 1.40 V in more recent pieces covering higher-end platforms.
The gist of the article from Tom's Hardware is that its possible to kill/damage your CPU with too much voltage, even if you have low temperatures. As was seen in their tests with Sandy Bridge. In other words, Voltage is a silent killer, and thats why is very important to ensure you don't use too much of it.

To give you an example, lets take the AMD FX 8350. According to CPU World, the AMD FX 8350 voltage is as follows:

Stock Voltage : - 1.35-1.37v @ 4.0 Ghz
Turbo State #1 -  1.4v @ 4.1 Ghz
Turbo State # 2 -  1.425v @ 4.2Ghz

So if I were to overclock this chip, I would not exceed 1.425V Vcore on this CPU. If I had Water cooling thats the highest volts I would go for an overclock...as raising Voltages higher then that could shorten the life of the CPU Dramatically.

In-fact, Your goal when overclocking should be to use the "least amount of volts(Vcore) possible at the highest clock speed you are stable" My FX 8320 under-volting article shows nicely how you can reduce voltage to extend the life of a chip and reduce temps.

For a good binned FX 8350 you should be able to hit 4.4-4.5 Ghz at somewhere around 1.39-1.4v which is a good 500Mhz overclock, that is well within the defined voltage range of the chip is rated for. However, not every chip is equal, some will go higher in Mhz and be stable and others will only be stable with lower Mhz...its the nature of the "silicon lottery".

When overclocking an Intel CPU, you use the same formula. Use the Intel Ark and CPU-World to find the stock and max rated voltages for the chip, and then stay inside that Vcore range.

Lastly for Vcore/Voltage, many motherboards have LLC Option(Load Line Calibration). Some overclockers swear by it, and others like myself avoid using it. Modern CPU's have what's called VDROOP as coined by Intel. VDROOP is when under heavy load your processor and motherboard will lower CPU Vcore. This is by design, however, it can cause stability issues with overclocking. LLC tries to compensate for this by reducing the amount of VDROOP under heavy loads. I would err on the side of caution messing around with LLC. In most cases your better off lower frequency and  getting stable at a slightly lower clock then messing around with LLC. Only hardcore overclockers that know exactly what they are doing, and understand the risks with CPU voltages should mess with the LLC setting. VDROOP is by design, and your best leaving it alone.

As for temperatures, this part is much simpler. Do not allow your CPU to exceed the max rated Temps its speced for.

AMD is usually around 61C and Intel is usually around 72C. However, Always Check the Intel Ark, AMD Product Pages, and CPU-World to verfiy the max temp for your CPU. and do not exceed it. Operating your CPU above the max temps the chip is rated for will damage your chip, so please ensure you have adequate cooling in place.

Overclocking

Overclocking itself is a fairly simple affair once you understand voltages and temp limits.

To overclock on Unlocked CPU's do the follow:

1. Go into BIOS and Set Multiplier to desired level.

2. Boot PC Into Windows.

3. If PC Boots sucessfully into Windows, Reboot PC back into BIOS.

4. Reduce CPU Vcore by 1 increment.

5. Reboot PC and Try to load Windows.

6. If Windows loads, Repeat Steps 3 & 4 above

7. At some point it will BSOD or fail to load Windows. At this point you have reached your undervolt limit.

8. Go back into BIOS and raise CPU Vcore by 1 Increment.

9. Try to boot into Windows, If Successful you have found a semi-stable overclock with undervolt.

 10. Now you can Test for Stability.

Note: During Stability testing, you may find the PC Boots into Windows, but crashes or freezes while running stability tests. If this happens Boot into BIOS and raise CPU Vcore by 1 increment. DO NOT exceed the max Vcore rated by the chip in the product specs. 

 

Stress Testing and Stability Testing

Now this is the part I go a bit against the grain. The overclocking community for the most part speaks as gospel that you should use programs like Prime95, Intel Burn Test, or LinX stress testing tool to Stress Test/Stability test your system. For the average PC Gamer i disagree with this premise for a variety of reasons.

In my opinion, these tests are too synthetic in nature, and unless your running and computing numbers for a distributed computing project such as Folding@Home, these tests offer very little value.

All Prime95, Intel Burn Test, and LinX do is cook your CPU. I have seen 72 hour Prime95 Stable, and Intel Burn Test Stable CPU's crash in games within seconds. Furthermore, there is circular logic involved in how long you run these programs. Most will say 48-72 hours of Prime95, but how did they reach this conclusion? Just because it doesn't crash in 72 hours doesn't mean it won't crash in 73 hours....you would have to run Prime95 forever to make sure it never crashes...I just don't think it makes any sense.

For the average gamer, if an overclock is unstable, it will freeze, error, or crash your system within the 1st few minutes in 9 out of 10 cases in my experience.

Instead this is "my stability/stress testing" method. This is what has worked for me. So how should the average gamer stress/stability test his overclock? Simple!

1. Download Asus Realbench.

2. Download CPUID Hardware Monitor, Core Temp, Open Hardware Monitor, or HWInfo and monitor your temps.

3. Run 5 passes of the Asus Realbench Benchmark test (Not the Stress Test)

4. Watch temps while running Realbench and ensure they don't exceed your max safe temps.

5. If you get through all 5 Real Bench Benchmark runs with no errors or crashes, move on to the next step, if not increase CPU Vcore in BIOS by 1 increment and run it again till it passes while watching temps.


6. Download Cinebench and run a CPU test with Cinebench.

7. If you get no errors or crashes while running the Cinebench CPU test your done, if you do error or crash, follow step 5 and ensure you don't exceed max voltage or lower CPU frequency slightly and try again.

8 Load up some of your favorite games now and play them for a bit and see if they crash while monitoring temps in-game

If you don't crash and temps are good, your done. Enjoy your increased performance.

Its been my experience that  if you can pass 5 runs of Real Bench your Overclock is usually 99% stable...I have yet to have a Real Bench Stable OC crash in a game or my PC before. Real Bench is a much less sythetic test and tests various aspects and different instruction sets on your CPU vs Prime95 and the others that cook your CPU for hours computing just numbers and only test a limited set of the CPU. Real Bench will also do a hardcore multi-tasking test, rendering a video, encoding with hand brake, and other programs to make sure your CPU OC is stable while using your GPU to render...its a very through test.

I know some of you may disagree with me on my stress testing methodologies, but its been my expereince that for the average gamer and novice, programs like Prime95, Intel Burn Test, and LinX just aren't necessary. However, for projects such as Folding@home and other distributed projects, it may be a good idea to use those applications, but for the average gamer, not so much.

Thanks for reading, and I hope this guide helps anyone out there new to overclocking. Have a great day! :)


Comments:10

  1. This is some reasonable explanation compare to my google search results. How much does excessive rendering would damage your cpu and gpu? If you do long period Seti like distributed computing job and excessive video rendering through your laptop cpu and gpu how much of it will get damaged? For example I did the above things with my core i5 laptop with minimum level gpu and it is now kind of freezing on some youtube videos. I clean my laptop cooling fan in every 10 months as it heats up by that time. Thank you.

    ReplyDelete
  2. As long as you keep your cpu and gpu within their rated temperature margins, they should last a very long time. Its only when using excessive volts and not keeping your temps below specifications that you should run into issues with shortening the life of your components. You could try updating the driver for the gpu on your laptop to see if that fixes the freezing issue.

    ReplyDelete
  3. Great guide. Worked perfect especially the testing methods. Have a stable overclock at x47 with 1.218 vcore i5-4690k on an Asus x97-a MB. Three weeks of gaming no issues at all. When pushing for x48 I am still not stable at 1.318 vcore. I am loathe to continue as I am on air cooling. Thanks so much for writing this.

    ReplyDelete
  4. Thank you for this guide super useful. Especially the stability testing. No issues with my overclock and I have three weeks of gaming down now. i5-4690k x47 multiplier 1.218vcore. I was trying to push a x48 but couldn't find stability at 1.316vcore so gave up the quest as I am on air cooling. What fun.

    ReplyDelete
  5. Best stress test guide I've come across used it to get my FX-8320E stable at 4.8 many thanks Ron.

    ReplyDelete
  6. No problem Brian, Glad my post was helpful for you!

    ReplyDelete
  7. Awesome Steve! Stability is the hardest thing to find sometimes, glad you got it sorted. :)

    ReplyDelete
  8. Cannot find the safe voltage ranges for the i5 4670k on the intel cite or cpu world. can you help me out? im trying to go for 4.4 for 1-2 active and 4.3 for 3-4 active and have it at 1.200v. it boots fine but i still need to stress test. so i should lower voltage .001 until windows fails to boot and then increase .001 until stable?

    ReplyDelete
  9. Hi Ron, apologies if this is a repeat...not sure my first post worked.

    I've just used your guide. Very clear and great advice...just one question though...if my core i7 4770k passed all tests but saw core 0 temp rise above the max temp advised by intel (72 degrees) to 78 degrees, but all the other cores stayed at or below 72, is this OK?

    I'm actually using a corsair h80i water cooler and reckon my temps are quite high, so plan to re-seat, but as a general rule, if 1 core is over the intel Tj.Max, is that OK?

    Cheers mate.

    Ali

    ReplyDelete
  10. Hello .. I'm new to OC'ing and just stumbled across your post. So far it's the clearest explaination/tutorial I've found. Though this post is a bit old I'm hopefully you're still around to answer a question I have.

    I just purchased an i7-4790k in the hopes of OC'ing, but when I look at CPU-world or ark.intel.com I don't see any information about max vcore ranges. Here are the specific pages for my chip.

    http://www.cpu-world.com/CPUs/Core_i7/Intel-Core%20i7-4790K.html
    http://ark.intel.com/products/80807/Intel-Core-i7-4790K-Processor-8M-Cache-up-to-4_40-GHz

    Can you provide any guidance around how I decipher these pages wrt max vcore?

    Thanks,

    Michael

    ReplyDelete

Copyright ronwoods.us. Powered by Blogger.

Get Updates By Email

Popular Posts

About Me

My photo

I know that you believe that you understood what you think I said, but I am not sure you realize that what you heard is not what I meant.