Jump to content

How do I tell my CPU is "stable"?

So I've spent a lot of time over the last few hours trying different settings and clocks. So far I've come to about 4.4GHz at 1.29V. Temperature is around 67-72 Celsius.

 

Despite only having air cooling (Hyper 212 EVO), is it safe to go higher voltage? Should I stop at a certain V, or just keep going until a certain temperature?

 

EDIT: In fact I've just noticed something. Despite setting it to 4.4GHz, it's only reading 4.19GHz in the stress test. I changed it to 4.5GHz, same result at 4.19GHz in the stress test. What's that about?

I use Intel® Extreme Tuning Utility for my stress tests. And I would say try not to go over 80-85 C for max temp 

Link to comment
Share on other sites

Link to post
Share on other sites

Intel Burn test intensive testing,OCCT,prime 95 and grab every single benchmark you know out there and do 2-3 tests like cinebench,3d mark and so on.

You want to make sure its stable in all sorts of loads,also ram testing.

Link to comment
Share on other sites

Link to post
Share on other sites

Intel Burn test intensive testing,OCCT,prime 95 and grab every single benchmark you know out there and do 2-3 tests like cinebench,3d mark and so on.

You want to make sure its stable in all sorts of loads,also ram testing.

 

What do I select?

 

jdydog.png

 

I think when I did Small FFTs, it went to 100 Celsius (while the others were much lower temperatures).

Link to comment
Share on other sites

Link to post
Share on other sites

What do I select?

 

jdydog.png

 

I think when I did Small FFTs, it went to 100 Celsius (while the others were much lower temperatures).

Don't run Prime95 without using manual voltage. The load line calibration over volts the CPU in offset or adaptive mode.

 

Try stressing in manual then turn on adaptive once stable.

Link to comment
Share on other sites

Link to post
Share on other sites

Don't run Prime95 without using manual voltage. The load line calibration over volts the CPU in offset or adaptive mode.

 

Try stressing in manual then turn on adaptive once stable.

 

I don't think my motherboard has the option for adaptive and such, I remember seeing that in Linus's video but not seeing it anywhere in my BIOS. It's a Gigabyte Z87-D3HP.

 

Also should I just use Intel Extreme Tuning Utility then? That's what I have been using.

Link to comment
Share on other sites

Link to post
Share on other sites

What do I select?

 

jdydog.png

 

I think when I did Small FFTs, it went to 100 Celsius (while the others were much lower temperatures).

Blend test will do just fine, but I definitely wouldn't recommend Prime, despite what others might say...especially without manual voltage as it has been known to damage the chips very quickly and isn't recommended by Intel. In fact, I would say that you shouldn't OC without manual voltage at all- you run the very real risk of damaging the chip otherwise. However, I would be very surprised if you MB can't do it as a quick Google says that it should. Intel Extreme Tuning is great, as is Aida (I would run both to be sure), and work with the parameters of 1.3V max voltage and 80C temp.   

i5 4690K | Asus Ranger VII | 8GB HyperX Fury | Asus GTX 780 | NZXT H440 | Samsung 850 Evo | Seagate Barracuda | Corsair RM 750W | Corsair H105 


 


E3-1246 v3 | Asus Gryphon Z97 | 8GB HyperX Fury | MSI GTX 970 | Enthoo Evolv mATX | Samsung 840 Evo | WD Red | EVGA SuperNova GS 650W | NZXT Kracken x41 

Link to comment
Share on other sites

Link to post
Share on other sites

Managed to get around 4.3GHz at 1.320V, did an hour stress test with Intel Extreme Tuning Utility. Any lower and I get blue screens and freezes. Temperatures are averaging around 72 to 84 Celsius.

 

Not only does it seem like I lost out on the lottery, but that I'm still doing something wrong. Also, surely my temps should be a lot lower? As said earlier, I'm using a Hyper 212 EVO, but I still feel like they should be lower.

 

Blend test will do just fine, but I definitely wouldn't recommend Prime, despite what others might say...especially without manual voltage as it has been known to damage the chips very quickly and isn't recommended by Intel. In fact, I would say that you shouldn't OC without manual voltage at all- you run the very real risk of damaging the chip otherwise. However, I would be very surprised if you MB can't do it as a quick Google says that it should. Intel Extreme Tuning is great, as is Aida (I would run both to be sure), and work with the parameters of 1.3V max voltage and 80C temp.   

 

Well I am setting the manual voltage, for example to 1.320V. Is that what you mean? I can't see any option for adaptive. There is an auto option though.

Link to comment
Share on other sites

Link to post
Share on other sites

Managed to get around 4.3GHz at 1.320V, did an hour stress test with Intel Extreme Tuning Utility. Any lower and I get blue screens and freezes. Temperatures are averaging around 72 to 84 Celsius.

 

Not only does it seem like I lost out on the lottery, but that I'm still doing something wrong. Also, surely my temps should be a lot lower? As said earlier, I'm using a Hyper 212 EVO, but I still feel like they should be lower.

 

 

Well I am setting the manual voltage, for example to 1.320V. Is that what you mean? I can't see any option for adaptive. There is an auto option though.

That sound right. From the look of the manual, under M.I.T, there is the option for "Advanced Voltage Settings". That should give you the options under "CPU Core Voltage Control", to swap from manual to adaptive once done with the OC. At least on my motherboard, adaptive and auto are different settings though I must confess that I have never been able to figure out the exact differences between them. I am assuming that your temps are that high under load? That seems about right given the high voltage and stress test you are using. For me at 1.3V, temps hit 65-70 max, but that is with a pretty hefty watercooling AIO. Makes sense for air cooling to be a bit higher. 

i5 4690K | Asus Ranger VII | 8GB HyperX Fury | Asus GTX 780 | NZXT H440 | Samsung 850 Evo | Seagate Barracuda | Corsair RM 750W | Corsair H105 


 


E3-1246 v3 | Asus Gryphon Z97 | 8GB HyperX Fury | MSI GTX 970 | Enthoo Evolv mATX | Samsung 840 Evo | WD Red | EVGA SuperNova GS 650W | NZXT Kracken x41 

Link to comment
Share on other sites

Link to post
Share on other sites

Anyone care to explain to me why prime is not used I still can't understand why?

Link to comment
Share on other sites

Link to post
Share on other sites

A lot of bad info in here.

Aida alone is not proof of stability.

Prime 95 custom with 80% ram usage for 12+ hours is stable.

Prime with 0 Windows kernal whea errors is legitimately stable. After that if you have 0 whea errors after a week or so of daily use your voltage and oc are perfectly stable.

Yo soy el hombre murciélago

Link to comment
Share on other sites

Link to post
Share on other sites

Also 1.4v is too high for haswell. 1.355 is upper echelon for long term reliability and life span.

Yo soy el hombre murciélago

Link to comment
Share on other sites

Link to post
Share on other sites

Managed to get around 4.3GHz at 1.320V, did an hour stress test with Intel Extreme Tuning Utility. Any lower and I get blue screens and freezes. Temperatures are averaging around 72 to 84 Celsius.

 

Not only does it seem like I lost out on the lottery, but that I'm still doing something wrong. Also, surely my temps should be a lot lower? As said earlier, I'm using a Hyper 212 EVO, but I still feel like they should be lower.

 

 

I wouldn't say you've lost the lottery, just gotten average. Mine is the same. My i5-4670k is at 4.4GHz, 1.325V manual voltage. 6+ hours stable stressed with the Intel Extreme Tuning utility. I believe temps were around 75C with a CM Hyper 212 Evo (stock fan), but it's been a while since I stress tested so I can't confirm off hand. In real life gaming/etc I don't usually see any higher than 55C (normally 45-50C) with this overclock.

Never say it's not broken. Everything is broken. Why? Because everything needs MOAR POWA!

Link to comment
Share on other sites

Link to post
Share on other sites

A lot of bad info in here.

Aida alone is not proof of stability.

Prime 95 custom with 80% ram usage for 12+ hours is stable.

Prime with 0 Windows kernal whea errors is legitimately stable. After that if you have 0 whea errors after a week or so of daily use your voltage and oc are perfectly stable.

Thank you sir, but since no one has yet confronted me, could you tell me why people have so many negative perspective of prime?
Link to comment
Share on other sites

Link to post
Share on other sites

They don't know better. Hop on ocn. True stability comes from several different tests. Windows kernal whea errors will tell the truth in the end!

Yo soy el hombre murciélago

Link to comment
Share on other sites

Link to post
Share on other sites

Remember to set ratio...etc. to "fixed mode" then run 10 cycles of IBT at least.

As long as it does not crash and/or go to 90 degrees, it will be fine.

Anyone who has a sister hates the fact that his sister isn't Kasugano Sora.
Anyone who does not have a sister hates the fact that Kasugano Sora isn't his sister.
I'm not insulting anyone; I'm just being condescending. There is a difference, you see...

Link to comment
Share on other sites

Link to post
Share on other sites

I'd stay below 1.25v with a budget air cooler. AIDA64/Intel XTU will show around same temps. IBT/PRIME95 push it higher even on manual voltage +10-20c 'ish.

 

~1.25V budget air.

1.25~1.30v Quality air.

1.25~1.35v Water.

Intel I5 4670k @4.6GHz // Corsair H100I // Asrock Fatal1ty Z97 Killer // Gigabyte R9 280x rev 2 // Sound Blaster Z

Evga 600b // 8G G.Skill Ripjaws X @2133Mhz // Corsair Spec-01 // Muskin Chronos 120G // Seagate Barracuda 1TB

Link to comment
Share on other sites

Link to post
Share on other sites

Prime95 can be used however if you are using the latest version it is going to seriosly limit the overclock because it stresses avx2 and fma3 to insane levels.

I prefer x264 and xtu(bench not stress) or adia64.

Link to comment
Share on other sites

Link to post
Share on other sites

Pretty sure version 27.9 (Prime95) was the last non super cooker version.

Intel I5 4670k @4.6GHz // Corsair H100I // Asrock Fatal1ty Z97 Killer // Gigabyte R9 280x rev 2 // Sound Blaster Z

Evga 600b // 8G G.Skill Ripjaws X @2133Mhz // Corsair Spec-01 // Muskin Chronos 120G // Seagate Barracuda 1TB

Link to comment
Share on other sites

Link to post
Share on other sites

Anyone care to explain to me why prime is not used I still can't understand why?

Prime95 will result in an overvolting of Haswell and Devil's Canyon CPUs on anything but manual voltage (it's something to do with how the cpu handles AVX instructions, from what I understand).  So most of the folks on this forum will recommend against it for newbies so that they won't damage their system.  On manual it's fine though.  

Also, Prime95 generates a hell of a lot of heat compared to other stress testers, which scares people.  AIDA64 seems to be the most recommended stress tester for Haswell and DC.  However, as EinZerstorer mentioned, AIDA64 alone isn't sufficient.  For instance I validated my 4690k OC with AIDA64, but I needed an additional 0.01V to make it stable for normal usage.  To each their own "stability".

Isopropyl alcohol is all you need for cleaning CPU's and motherboard components.  No, you don't need [insert cleaning solution here].  -Source: PhD Student, Chemistry


Why overclockers should understand Load-Line Calibration.


ASUS Rampage IV Black Edition || i7 3930k @ 4.5 GHz || 32 GB Corsair Vengeance CL8 || ASUS GTX 780 DCuII || ASUS Xonar Essence STX || XFX PRO 1000W

Link to comment
Share on other sites

Link to post
Share on other sites

Test for 24 hours to be absolutely certain. 

I've seen ocs fail after 40 hours. 48 hours is what I recommend. 

Link to comment
Share on other sites

Link to post
Share on other sites

Prime95 will result in an overvolting of Haswell and Devil's Canyon CPUs on anything but manual voltage (it's something to do with how the cpu handles AVX instructions, from what I understand). So most of the folks on this forum will recommend against it for newbies so that they won't damage their system. On manual it's fine though.

Also, Prime95 generates a hell of a lot of heat compared to other stress testers, which scares people. AIDA64 seems to be the most recommended stress tester for Haswell and DC. However, as EinZerstorer mentioned, AIDA64 alone isn't sufficient. For instance I validated my 4690k OC with AIDA64, but I needed an additional 0.01V to make it stable for normal usage. To each their own "stability".

Ohh I think I get the picture. So basically most people don't even manually set their vcore and such which the system maybe give some insane variables. I didn't even notice people are just doing adaptive mode letting the system tune their volts. No wonder why I see complains. Oh well that means I am good for prime as I had used it for many years.

Thanks for you insight.

Link to comment
Share on other sites

Link to post
Share on other sites

Ohh I think I get the picture. So basically most people don't even manually set their vcore and such which the system maybe give some insane variables. I didn't even notice people are just doing adaptive mode letting the system tune their volts. No wonder why I see complains. Oh well that means I am good for prime as I had used it for many years.

Thanks for you insight.

Just to clarify, adaptive voltage is great to use once you have a stable overclock.  The issue is people validating overclocks (running stress tests) with adaptive voltage on.  That results in some really unpredictable voltage response.  

Isopropyl alcohol is all you need for cleaning CPU's and motherboard components.  No, you don't need [insert cleaning solution here].  -Source: PhD Student, Chemistry


Why overclockers should understand Load-Line Calibration.


ASUS Rampage IV Black Edition || i7 3930k @ 4.5 GHz || 32 GB Corsair Vengeance CL8 || ASUS GTX 780 DCuII || ASUS Xonar Essence STX || XFX PRO 1000W

Link to comment
Share on other sites

Link to post
Share on other sites

Just to clarify, adaptive voltage is great to use once you have a stable overclock. The issue is people validating overclocks (running stress tests) with adaptive voltage on. That results in some really unpredictable voltage response.

Well that feature might be great, but the fact that it goes crazy raising the volts on a somewhat unrealistic extreme Street test scares you. I mean the test might be impractical but it's possible and at least for me it shows the level of execution that the features introduces. I still would rather manually find the lowest threshold while maintaining optimal performance. Most safe and I have control of it :). If so then people shouldn't blame the software as much.

So I thought it had something to do with haswells silicon's heat distribution or such but am I afraid not I guess.

Link to comment
Share on other sites

Link to post
Share on other sites

If you dial in offset properly no stress test will over volt your pc. Your offset setting is directly in control of voltage levels. Period. Llc plus offset. In normal use you will see less voltage.

Yo soy el hombre murciélago

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×