Jump to content

Prime95 should be just fine with Devil's Canyon (DMM testing)

DISCLAIMER:

If you're new to overclocking, go watch a tutorial video.  If you want to use synthetic tests (Prime95, AIDA64, etc.), use manual voltage only (not adaptive).

 

So I kept hearing over and over that you shouldn't use Prime95 on Haswell because there are crazy voltage jumps (or similar statements), but I never really saw the evidence for this.  I decided to have a look for myself.  

 

I picked up a 4690K and an ASUS Maximus VII Gene, and did some voltage testing under load (with a digital multimeter reading off the voltage checkpoints), comparing the voltages seen with AIDA64 to Prime95 small FFTs and Blend tests.  

 

Setting a multiplier of 39, and testing voltages at 1.05, 1.10, 1.15,1.20, and 1.25V, there was no appreciable differences between the VCores of AIDA64, Prime95 small FFTs, and Prime95 Blend.  The VCores differed by about +/- 0.002V, which is certainly nothing to write home about.  

 

When upping the multiplier to 42 and 44, there was no difference between AIDA64 and Prime95 Blend voltages, and Prime95 small FFTs resulted in a VCore boost of +0.005V.  Again, this is nothing to write home about.  There is a small voltage boost from using Prime95 small FFTs at higher (overclocking range) multipliers, but this is a one-step jump in VCore, which is not concerning.  

 

After validating an overclock (4.6 GHz with 1.29V Vcore), I tested this again, and found the same.  There was no difference between AIDA64 and Prime95 Blend, and a +0.005V jump using Prime95 small FFTS.  

 

In terms of temperatures, AIDA64 and Prime95 Blend had similar maximums and averages (~71C max, 60C average), but Prime95 Small FFTs resulted in maximums of 85C, with an average of 75C.  (These were taken using RealTemp, and the cooler was an H100 with two NF-F12s.)  

 

So to sum this all up:

- You should be able to use Prime95 just fine on Devil's Canyon processors (Prime95 will NOT result in a significant voltage jump).  

- If you do use Prime95, use blend (not small FFTs, because you'll be looking at some pretty crazy temperatures)

- If you are using synthetic stress tests: use MANUAL VOLTAGE ONLY (NOT ADAPTIVE)

 

 

An additional note:

Remember to set your Load-Line Calibration (read more here and here).  For Haswell chips, LLC won't apply too much voltage to VCore, but it does affect the VIn (the voltage supplied to the CPU).  With lower LLC settings, there was a pretty significant droop to VIn, and the read values were very unstable (varying by almost 10 mV over a couple seconds).  These lower and much less stable voltages would probably destabilize an overclock quite easily, and I found that the auto LLC setting of my motherboard chose 2/9, which resulted in some pretty awesome VIn droops.  With higher LLC settings, the VIn remained much more stable at the value actually set across the VCore voltage range tested.  

I ended up going with a LLC setting of 7/9 on this board (I might bump up to 8 for higher overclocks), but anyone overclocking Devil's Canyon (or Haswell), should test and set their own LLC settings for more stable overclocks.  

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

I always thought if you just turn off adaptive voltage you'll be fine using prime.

Maybe that's what it is, but I think some folks took that to town and said that you shouldn't use Prime95 on Haswell.  

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

@Queek,

 

i do not suggest using P95 on Haswell or Devils Canyon as i do not know their

aptitude on using this tool. telling them to use blend without adaptive mode, with

adequate cooling.. someone is going to FUBAR and point fingers. so easier to

use is the Intel Extreme Tuning Utility (XTU) since it is an Intel tool, has monitoring

embedded, and has simple benchmark/stress test modes. no guessing.

 

P95 was great for ivy bridge and back CPUs, but this newer architecture and mobo

functions can really throw a new experience into a shutdown and they do not know

what they have done.

Link to comment
Share on other sites

Link to post
Share on other sites

@Queek,

 

i do not suggest using P95 on Haswell or Devils Canyon as i do not know their

aptitude on using this tool. telling them to use blend without adaptive mode, with

adequate cooling.. someone is going to FUBAR and point fingers. so easier to

use is the Intel Extreme Tuning Utility (XTU) since it is an Intel tool, has monitoring

embedded, and has simple benchmark/stress test modes. no guessing.

 

P95 was great for ivy bridge and back CPUs, but this newer architecture and mobo

functions can really throw a new experience into a shutdown and they do not know

what they have done.

Are you saying that you don't recommend that new overclockers use Prime95 on Devil's Canyon?  For what reason?  I just showed that if you're doing it right, you won't get huge voltage jumps, and I've said that you should use manual voltage for synthetic tests.  

This post was directed to the moderately experienced overclockers which populate this forum, not brand new users.  I'm trying to show that if you still rely on synthetic tests to validate your overclock (instead of general usage), you should still be able to use the old tools.  

 

I would love for someone to tell me why exactly I can't use P95 or AIDA64 on Devil's Canyon.  Will it boost the voltage?  Will it increase temperatures?  

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

@airdeano

 

You do know that XTU uses Prime95 as its benchmark utility.

 

Prime95 is perfectly fine for overclocking, so long as you have adequate cooling and voltage is set to manual.

"I genuinely dislike the promulgation of false information, especially to people who are asking for help selecting new parts."

Link to comment
Share on other sites

Link to post
Share on other sites

Are you saying that you don't recommend that new overclockers use Prime95 on Devil's Canyon?  For what reason?

 

as explained earlier, you get newer code application, simple use and with

load/temperature monitoring. not having 3 apps to do the same with one app.

 

new builders or veterans, the new tech does have a lot of peeps chasing their

tail with unreliable temperatures, inexperience to RAM tuning (unring technology)

and shear old tech relationship to new tech applications. the newer CPUs have

different instructions than the older platforms and work entirely different

including voltage increases in pro-long stressing environments.

 

 

@airdeano

 

You do know that XTU uses Prime95 as its benchmark utility.

 

Prime95 is perfectly fine for overclocking, so long as you have adequate cooling and voltage is set to manual.

 

yes i know XTU and IBT use P95 algorithms, but not P95 proper.

Link to comment
Share on other sites

Link to post
Share on other sites

as explained earlier, you get newer code application, simple use and with

load/temperature monitoring. not having 3 apps to do the same with one app.

 

new builders or veterans, the new tech does have a lot of peeps chasing their

tail with unreliable temperatures, inexperience to RAM tuning (unring technology)

and shear old tech relationship to new tech applications. the newer CPUs have

different instructions than the older platforms and work entirely different

including voltage increases in pro-long stressing environments.

Ok, if I'm understanding you correctly, you're saying that the reason to use XTU is that it has everything in one utility?

Or are you saying that the new instructions aren't adequately tested with older programs?

I'm honestly having difficulty following what you're trying to say.  

Are the core temperatures from RealTemp unreliable?  How about AIDA64's temperatures?

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

xtu uses linpackxeonx64 to stress test(non avx version), the linpack avx version is a gruesome torture test . xtu is not the best for stressing avx2 instruction sets. most users will not run many real world applications that stress avx2 anyways., so xtu is suitable for 75% of users on this platform. i personally enjoy the xtu interface for both tuning and monitoring vitals plus it can cope with adaptive voltage so thats one variable eliminated...highly recommended.

 

note- the xtu benchmark uses p95 and will overvolt if benched on adaptive.

 

prime95 v27.9 blend is a good test when looking for avx stability. i would suggest it over xtu if finding stability is an issue.

 

plain and simple do not run prime, orthos, aida64 or ibt on adaptive voltage. set to manual mode if your keen on using  these software suites.

Mainboard Asrock Z170 OCF CPU 6700k RAM Tridentz 3600 HDD Intel 730 240gb GPU GTX 780ti sc acx PSU Silverstone Strider 1200W  Case Antec 900 Laptop Lenovo Thinkpad T520 build log-   http://linustechtips.com/main/topic/35809-antec-900-the-re-birth-of-a-legend/ Check out the Tech Center https://www.youtube.com/user/prokon24/videos LTT's Unicore King

 

Link to comment
Share on other sites

Link to post
Share on other sites

  • 1 year later...

I OC my 4690k to 4.2Ghz with 1.1v and my load line calibr.. went to 8/9 in ai suite...is this ok?
i would like to push it to 4.6Ghz.. am i ok to continue? 

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

×