OC'ing Guide for the Z97 Extreme6/4790K?

[Ken429]

Now that I have cleared up the Mouse, Keyboard and UPS USB sleep wake problems and the system is stable running with all the default BIOS settings I'm looking for an OC'ing Guide for the Z97 Extreme6 MB. Seems there are more than a few different BIOS options on this MB/BIOS compared to the Z77 OC Formula MB. I must be looking in all the wrong places because I can't find much information on this issue. I'm not trying to set any world record just a reasonable OC to see what it will do compared the Z77 OC Formula/3770K running at 4.5GHZ.

[godto]

Re: OC'ing Guide for the Z97 Extreme6/4790K?

Ok i can help you start .. this is my OC to 4.5 stable with same setup as your

OC Tweaker in Bios

- All core
- Cpu ratio = x45
-cache ratio = x45
-cpu Vcore Voltage mode = adaptive
- off set voltage = +0.400
-cpu load line = level 1 ( this same as LLC off)

that all i do..

idle temp = 31-34C - Idle voltage = 1.20V
full load = 84-88C ( average range 62-78C ) Voltage = 1.248V ( this is on Aida, OCT,Prime95 stress test)

i just oc little .. let the board set my voltage and turn out good ... i was surprise how good the Adaptive Voltage work ..
I back down to 4.4 for everyday .. I'm on Air 212Evo ( all i change on it is X44 .. that all)
very stable , if i have better cooler i would go for more. but at 4.4 my max temp only 81C

this was the easiest OC i ever done.

here is the link to the pict https://www.flickr.com/photos/125579...6/14708444651/

[Ken429]

Re: OC'ing Guide for the Z97 Extreme6/4790K?

Thanks for sharing your settings.

Sorry it took me so long to get back but I was doing more searching and testing. I found this "old" video from Asus which was a very good tutorial.

https://www.youtube.com/watch?v=Z7zPu9255ZI

I did exactly what he said to get a 46 multiplier and it seems to work. The ASUS BIOS terminology is very similar to the ASRock BIOS terminology.

The system idles at ~24C at .805V and runs at ~1.239V and maxis out at ~62C at 100% utilization using the ROGRealbench Version 2.2 program. The bulk of the time the Vcore under load is 1.229V and the max temperature for any core is < 50C. Everything I read says if your using Adaptive Voltage Control don't run Prime95 or IBT or a bunch of other stuff because it overrides the Adaptive logic and could cause problems. Hopefully, someone can shed more light on this subject.

The settings I used were:
Multi Core Enhancement = All Core
CPU Ration = 46x
CPU Cache Ratio = Auto
Load XMP Setting = XMP 1.3 Profile1
CPU Vcore Mode = Adaptive
Vcore Adaptive Voltage = 1.20V
Load Line Calibration = Auto

Sure wish ASRock had some OC tutorials for their stuff but then that would make it too easy?! The system appears to run everything I throw at it with no BSOD or Event Log entries. Based on the result I've had so far it appears I may have a good processor that can run at a higher multiplier. We'll see.

[godto]

Re: OC'ing Guide for the Z97 Extreme6/4790K?

yeah ,,, i think you have a good chip .. i'm redo my OC now .. 4.5 at 2.3Volt :(
trying to tweak it now ... also i'm changing my cooler too.
from 121Evo to NH-D14 ( have it by tomorrow .. hope i can OC better)

[parsec]

Re: OC'ing Guide for the Z97 Extreme6/4790K?

Thanks for sharing your settings.

Sorry it took me so long to get back but I was doing more searching and testing. I found this "old" video from Asus which was a very good tutorial.

https://www.youtube.com/watch?v=Z7zPu9255ZI

I did exactly what he said to get a 46 multiplier and it seems to work. The ASUS BIOS terminology is very similar to the ASRock BIOS terminology.

The system idles at ~24C at .805V and runs at ~1.239V and maxis out at ~62C at 100% utilization using the ROGRealbench Version 2.2 program. The bulk of the time the Vcore under load is 1.229V and the max temperature for any core is < 50C. Everything I read says if your using Adaptive Voltage Control don't run Prime95 or IBT or a bunch of other stuff because it overrides the Adaptive logic and could cause problems. Hopefully, someone can shed more light on this subject.

The settings I used were:
Multi Core Enhancement = All Core
CPU Ration = 46x
CPU Cache Ratio = Auto
Load XMP Setting = XMP 1.3 Profile1
CPU Vcore Mode = Adaptive
Vcore Adaptive Voltage = 1.20V
Load Line Calibration = Auto

Sure wish ASRock had some OC tutorials for their stuff but then that would make it too easy?! The system appears to run everything I throw at it with no BSOD or Event Log entries. Based on the result I've had so far it appears I may have a good processor that can run at a higher multiplier. We'll see.

Your OC at that voltage and CPU temperature is very good, although I don't know how stressful the ROGRealbench program is. Your CPU cooler is really keeping the CPU temperature low, relative to Haswell processors, that run hot. Even the best air and CLC liquid coolers won't keep 'K' Haswell processors from approaching or reaching their thermal throttling temperature when stress testing with some programs.

AIDA is another good test, and the Intel Extreme Tuning Utility includes a stress test, as well as a good way to set and test OCs in the Windows environment:

https://downloadcenter.intel.com/Det...ng&ProdId=3483

The "issues" with Prime95, IBT, or AIDA happen because those tests use, or have the option to enable, AVX 2.0 CPU instructions during the test. When AVX instructions are sent to Haswell processors, they will increase the VID/CPU voltage beyond your settings when using Adaptive voltage, particularly when running stress tests at 100% CPU usage. That is normal behavior for these processors, but in real world usage you would never encounter the combination of AVX instructions at 100% CPU usage uninterrupted for long periods of time. When I ran AIDA on my i5-4670K PC with AVX enabled, I hit 100C on a core or two in a minute, using a small CLC cooler. That was the first time I ever had a CPU thermal throttle itself, although I'm not a hardcore OC enthusiast.

The stock Intel CPU cooler could not stop thermal throttling within 30 seconds with a 4.6GHz OC running a stress test with AVX instructions. Some people insist an OC must survive long term stress tests with AVX instructions enabled at 100% usage, but these (or all) processors are simply not meant to deal with that type of load. You see the lengths some people go to, to achieve high OC's with AVX instructions, like de-lidding the CPU.

As an exercise in pure over clocking capability, I understand that anything goes, but to worry about not achieving that level of stress testing capability for real world use is ridiculous IMO.

[godto]

Re: OC'ing Guide for the Z97 Extreme6/4790K?

actually i run IBT, AIDA, OCC , Intel Extreme tuning and i run these at least 1 to 2 hours .... the only thing i haven't run pis Prime95 ... haven't have time yet ... i know if i pass these 4 then i can run prime85 latter ... i'm sure it will be fine ... on 4.4 i did run prime for 6 hours.... i find the intel extreme tuning is very helpful ... they give you hint and also if some thing wrong they just stop ( actually they frozzz on me few time). Tomorrow I will have Noctua D14 ... I will run IBT more because right now i hit 91 and that too hot for me ... what ever i can OC to with the max temp is under 85 is a sweet spot for me... still teaking now and i go .005 at the time .. very very small increment to get the most out of it.

[Ken429]

Re: OC'ing Guide for the Z97 Extreme6/4790K?

parsec,
I ran the AIDA64 V4.50 stress test with FPU turned off and the Max Voltage and Core Temps were very close to the ones I got in RealBench V2.2. I also ran the stress test included in RealBench an again I got similar results. I'm kind of obsessed when it comes to holding the core temps to reasonable levels and am very impressed with the 4790K so far. My 3770K system with custom cooling hits 80C at 4.5GHZ under 100% load. I haven't tried the Intel Extreme Tuning Utility but I'll download it and give it a try. Does it mess with the AVX instructions when stress testing using Adaptive Voltage?

One thing that I don't understand is the max voltage under stress of 1.229-1.239V. I have the max set to 1.20V. In the video when he put the ASUS system under stress it held at the 1.20V. I have the LLC set to Level 1 (Off?) and yet the CPU is getting additional voltage. Is this the difference the ASUS BIOS vs the ASR BIOS or the Z87 Chip vs the Z97 Chip or the 4770K vs 4790K? Whatever I assume it's the BIOS logic, maybe I should lower the max Adaptive Voltage to account for the "extra" being supplied?

goto,
your idle voltage is insanely high. Probably because you have +.40V in the Offset Voltage (.805V+.40). The ASUS video says don't screw around with the Offset Voltage setting just plug a voltage in the Vcore Adaptive Voltage. I'm surprised that doesn't show up in the max voltage under stress testing? I'm also surprised your Vcore and Core Temps don't go out of sight if your running IBT using Adaptive Voltage?

[godto]

Re: OC'ing Guide for the Z97 Extreme6/4790K?

thanks Ken429

Yeah i notice my volt still to high. I'm starting my OC over and set manually most everything. I'm at 4.5 now and the only weird thing is ... i pass all the test AIDA,Prime, IBT, Intel Extreme, OCCT large,
but the kicker is in OCCT Small I Crash. ( all these test i ran for 1/2 hours just to see if they stable only, haven't done 2 hours full test).
I have to lower my Cache all the way down to 35x ( been on the net reading everything i could about 4790k and OC and compare to other resulted) I see most people with stable OC ran their Cache at 32x .
I read that they don't make that much different in performance of OC.

so far my OC are 4.5 at 1.165v , Input voltage to 1.920v , cache 1.17v ....... and Very stable OCCT Small only 75c max. and IBT Standard only 81c max.

BTW. this all done on my new NH-D14 cooler :)

[parsec]

Re: OC'ing Guide for the Z97 Extreme6/4790K?

parsec,
I ran the AIDA64 V4.50 stress test with FPU turned off and the Max Voltage and Core Temps were very close to the ones I got in RealBench V2.2. I also ran the stress test included in RealBench an again I got similar results. I'm kind of obsessed when it comes to holding the core temps to reasonable levels and am very impressed with the 4790K so far. My 3770K system with custom cooling hits 80C at 4.5GHZ under 100% load. I haven't tried the Intel Extreme Tuning Utility but I'll download it and give it a try. Does it mess with the AVX instructions when stress testing using Adaptive Voltage?

One thing that I don't understand is the max voltage under stress of 1.229-1.239V. I have the max set to 1.20V. In the video when he put the ASUS system under stress it held at the 1.20V. I have the LLC set to Level 1 (Off?) and yet the CPU is getting additional voltage. Is this the difference the ASUS BIOS vs the ASR BIOS or the Z87 Chip vs the Z97 Chip or the 4770K vs 4790K? Whatever I assume it's the BIOS logic, maybe I should lower the max Adaptive Voltage to account for the "extra" being supplied?

ASRock has been setting their LLC as level 1 is the highest/most LLC, level 5 is the lowest. When the LLC option is highlighted, you should see a small graph displayed on the right, which shows the LLC correction for each number. Check the graph, you'll see what I mean.

Adaptive is infamous for raising VCore under heavy load, but try lowering your LLC. Also, note where the LLC setting is, right by the CPU Input Voltage. I know LLC affects CPU Input Voltage, my question is does it also affect VCore, which is determined by the CPU's voltage regulators (FIVR.) We can safely assume (?) that a more stable CPU Input Voltage will also affect VCore, but does the LLC setting also change the FIVR's response. I'm simply not sure.

godto's idle voltage might be affected by the CPU power saving options he has enabled, we don't know what those are.

A little trick I use with Adaptive voltage, is I use a negative offset voltage. I'll set the Adaptive voltage to say 1.200V, and then set the Adaptive Offset to -0.080V. I end up with a max VCore/VID of 1.12V. The VCore won't go above that in the IXTU stress test, but I haven't run the others lately. I find setting a lower Adaptive VCore and no offset causes BSODs. Your experience may be different, I'm not using an i7-4970K.

I'm also finding that a lower CPU Input voltage does not cause problems. I'm able to do a small OC of 4.0GHz on both my G3258 and i5-4670K with CPU Input voltage at a bit over 1.6V. I actually have the G3258 at a CPU Input of 1.568V, VCore of 1.12V max, and it's fine, and no doubt easier on the VRMs.

IMO, you may enjoy IXTU now, it works fine on my Z87 and Z97 ASRock systems.

[Ken429]

Re: OC'ing Guide for the Z97 Extreme6/4790K?

ASRock has been setting their LLC as level 1 is the highest/most LLC, level 5 is the lowest. When the LLC option is highlighted, you should see a small graph displayed on the right, which shows the LLC correction for each number. Check the graph, you'll see what I mean.

Adaptive is infamous for raising VCore under heavy load, but try lowering your LLC. Also, note where the LLC setting is, right by the CPU Input Voltage. I know LLC affects CPU Input Voltage, my question is does it also affect VCore, which is determined by the CPU's voltage regulators (FIVR.) We can safely assume (?) that a more stable CPU Input Voltage will also affect VCore, but does the LLC setting also change the FIVR's response. I'm simply not sure.

godto's idle voltage might be affected by the CPU power saving options he has enabled, we don't know what those are.

A little trick I use with Adaptive voltage, is I use a negative offset voltage. I'll set the Adaptive voltage to say 1.200V, and then set the Adaptive Offset to -0.080V. I end up with a max VCore/VID of 1.12V. The VCore won't go above that in the IXTU stress test, but I haven't run the others lately. I find setting a lower Adaptive VCore and no offset causes BSODs. Your experience may be different, I'm not using an i7-4970K.

That's why I had the "?" around "LLC = Level 1. I looked and re-looked at the graph in the BIOS and it said use Level 1 to "Disable". However, it appears it don't make any difference what LLC is set to when you are using Adaptive Voltage. I tried Level 3 and Level 5 and the Vcore under load (AIDA64 Stress Test) still came up with 1.229-1.239V. So I'll play with the Vcore Adaptive Voltage and/or negative Voltage Offset and see what happens.

[Ken429]

Re: OC'ing Guide for the Z97 Extreme6/4790K?

Things are going down hill! I tried using the Vcore Voltage Additional Offset of -.02V and that worked EXCEPT it lowered the idle voltage from .805V to .785V. I don't know where the 4790K stalls but... Next I tried putting the Offset back to Auto and changed the CPU Vcore Adaptive Voltage to 1.18V and that had no effect - the Vcore under load was still 1.229-1.239V. So I set the Vcore Adaptive Voltage to Auto and guess what - the Vcore under load was 1.229-1.239V! Time to buy an ASUS MB?!

Or...is it P1.33? I can't find it on the ASRock site. Did they pull it because of problems?

[godto]

Re: OC'ing Guide for the Z97 Extreme6/4790K?

the same here with adaptive voltage. i use fix voltage and set lower limit .. seem to work great. i do use adaptive on GT clock(i think it video mhz). but one thing i do is set adaptive voltage and go into MB to see what they bump it up to .. then i get an idea of my max voltage ever be. it a guide for me to not over voltage. hehehe

BTW I'm on 4.6 now ... the temp little high on IBT 92-93c .... will try little more if can't get the temp under 90c , if not then i go back down to 4.5

[parsec]

Re: OC'ing Guide for the Z97 Extreme6/4790K?

Things are going down hill! I tried using the Vcore Voltage Additional Offset of -.02V and that worked EXCEPT it lowered the idle voltage from .805V to .785V. I don't know where the 4790K stalls but... Next I tried putting the Offset back to Auto and changed the CPU Vcore Adaptive Voltage to 1.18V and that had no effect - the Vcore under load was still 1.229-1.239V. So I set the Vcore Adaptive Voltage to Auto and guess what - the Vcore under load was 1.229-1.239V! Time to buy an ASUS MB?!

Or...is it P1.33? I can't find it on the ASRock site. Did they pull it because of problems?

Not to worry, the reduction in VCore at idle when using Adaptive voltage is normal, that is one of the features of Adaptive voltage. With CPU power saving options enabled (C-States, EIST) using Adaptive voltage, the CPU frequency is 800MHz, and I've had my VCore/VID below 0.650V. On boards that derive a true VCore (or attempt to, who knows if it is right) can show a CPU voltage at idle using Adaptive of 0.200V or less!

Haswell processors are very different than earlier Intel processors, don't think everything that was true in the past about Intel processors still applies, it doesn't. I see Haswell CPU Package power (the entire CPU) use at idle below one Watt.

I wanted to upload a picture of HWiNFO readings of my Z87/G3258 PC's voltage and power readings using Adaptive voltage, but once again the forum's picture uploading capability is broken (tells me a jpg and png file is an invalid image type) so I can't.

UEFI/BIOS P1.33 is history, removed for officially unknown reasons, unofficially many users complained about it. But I guarantee that using 1.30 or 1.20 won't change the idle Adaptive voltage, it is supposed to be that low, completely normal and fine.

[Ken429]

Re: OC'ing Guide for the Z97 Extreme6/4790K?

parsec,
OK, I'll use the Offset to lower the Vcore. I suppose ~1.20V under load is OK for 24-7. We'll see what happens when I try to push the multiplier to 48x.

The other question I had was...it appears that the "CPU Vcore Adaptive Voltage" has no purpose! Changing it from 1.20V to 1.18V had no effect? I'll go back to P1.20 and see if that changes the BIOS behavior.

I'm still looking for the flaw(s) in P1.33. Normally I stumble into every issue that exists. Since I got my wrist slapped for asking why P1.33 was pulled it would be nice to know who was having what kind of problems that were posted on this site. Just curious, is there another Forum for ASRock MB users?

[parsec]

Re: OC'ing Guide for the Z97 Extreme6/4790K?

parsec,
OK, I'll use the Offset to lower the Vcore. I suppose ~1.20V under load is OK for 24-7. We'll see what happens when I try to push the multiplier to 48x.

The other question I had was...it appears that the "CPU Vcore Adaptive Voltage" has no purpose! Changing it from 1.20V to 1.18V had no effect? I'll go back to P1.20 and see if that changes the BIOS behavior.

I'm still looking for the flaw(s) in P1.33. Normally I stumble into every issue that exists. Since I got my wrist slapped for asking why P1.33 was pulled it would be nice to know who was having what kind of problems that were posted on this site. Just curious, is there another Forum for ASRock MB users?

I hope you're not referring to anything that I said about your question about that. If so, that was not my intent at all. All I tried to say about that topic in general was, no one does it. That does not mean I think it's the right thing to do, but I can understand some of the implications that a manufacture may want to avoid if they did. But as the disclaimer associated with BIOS updates states:



Other user issues with 1.33? As I recall, random/spontaneous Windows Sleep occurring (wait, that was you), unable to enter Sleep, PC starting from shutdown by moving mouse or touching any key on keyboard.

Apparently you don't believe me, but Adaptive voltage is unique to Haswell processors, and its behavior is different that any other Intel processor voltage setting. Adaptive is a good name for it, it adapts the CPU voltage to the situation the processor finds itself in. At idle it decreases the CPU voltage lower than any other Intel CPU can continue running on. Did you ever hear about how some PSUs are not compatible with Haswell processors, since they can use so little power that the load on the PSU is so low, the 12V rail will go out of ATX spec, and cause the PSU's OVP to activate and shut it off? During certain stressful situations (AVX 2 instructions) it increases the CPU voltage beyond the settings in the BIOS. If you don't like it, don't use it, many OC enthusiasts don't like it. Personally, I use it and like it. You should have seen Adaptive voltage before we had negative Adaptive offsets, that was not good IMO.

You seem to think that Adaptive voltage behaves as it does due to ASRock's BIOS having bugs. You can try another manufacture's board, but I predict you'll find it is no different. Perhaps BIOS 1.33 is buggy with Adaptive voltage, I've never used 1.33.

Things can change in the way a processor generation behaves, and how it is over clocked. We don't (can't) OC like we did in the Nehalem and Core 2 processor days, and Haswell has changed that a bit again with Adaptive voltage.

Don't forget that a Haswell compatible mother board is not providing/creating the VCore voltage for a Haswell processor, or most of the other voltages a mobo provided to a processor in the past. VCore and other voltages are created by the FIVR voltage regulators in a Haswell processor. The only voltages provided by a Haswell compatible mother board for the CPU are CPU Input Voltage, and DDR Memory voltage.

The vast majority of people OCing Haswell processors above ~4.4GHz use more than 1.2V for their VCore. You might want to check some of the threads in this forum about Haswell over clocking: Intel CPUs

Another ASRock forum? Yes there is another, check this page: ASRock > Forum