having overclock trouble

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.

steedsofwar

Distinguished
Aug 22, 2015
133
2
18,695
TLDR: Why is the multiplier showing 44 in CPUZ when it shows 46-47 in bios? Seems the overclock isn't being applied.

Okay guys... My old setup suffered water damage to the MITX motherboard, which I've since replaced with a larger MATX ASRock Z97M OC Formula board and a midi case to fit it in. This is due to the lack of availability of good overclocking boards in the z97 chipset at the MITX size here in the UK. I used it as an excuse to factor in a future SLI setup and more RAM.

So my problem is that I'm a noob. It's really a repair of my first build and my first overclock. I am stuck on 4.4ghz under prime95 loads despite setting a higher multiplier in bios at 1.2v. Temps are definitely higher than it should be (maxing out at 80C at high ambient temps) but that's for later. I have watched tonnes of videos online and read tutorials from popular sources, of a confusing variety of z97 boards. I can't figure it out. I have read Tom's review of this board several times too. It was not very flattering for an OC labeled board.

Why is the multiplier showing 44 in CPUZ when it shows 46-47 in bios? Please see what I've changed in BIOS and what CPUZ is showing: https://www.dropbox.com/sh/xuna4x9u87gmueh/AADroxm7Zi0sO6n8VgZTW8gWa?dl=0


My system consists of:

Core i7 4790k
16GB Kingston Hyper X 18xx RAM
ASRock Z97M OC Forumula
MSI GTX 980Ti
H100i GTX in push pull exhausting
Two 140mm Phanteks x series intake fans
1 Phanteks 140mm case fan at exhaust
Phanteks enthoo evolve midi gun metal case
Corsair RM650 80+ Gold PSU I believe
 
Solution
Only use your Bios to OC your system and not from software within Windows.
If you make an incorrect setting ie: Bad OC multiplier ratio, incorrect voltage or bad Primary Timings, the OC would not be accepted and would revert back to default.
Do not make arbitrary changes in Bios without knowing the consequence.
Do not try to go to an unrealistic OC in one hit, rather follow the OC guide and methodology.
Get to know your Bios, there are many vids on youtube that can help. Also specific help is available here if you are having issues.


So, under OC Tweaker, it says Nick Shih's OC Profile set at: Default.
OC Tweaking is set to: Enabled.
Load optimised CPU OC Setting is set to: Disabled instead of Turbo 45. Turbo 4.6 and Turbo 4.7ghz.

Then under CPU Config

CPU Ratio field is set by me to value: All Core
All Core field is set to: 45
CPU Non-Turbo Ratio: Auto
CPU Cache Ratio: 45
BCLK Boot Frequency: 100.0
BCLK Eventual Freq: 100.0
BCLK/PCIE Ratio: Auto
Spread Spectrum: Disabled
CPU OC Fixed Mode: Enabled
Every other field in this section is set to auto as default.

DRAM Timing Config

All three fields here are set to auto

FIVR Config

All four fields here are set to Auto as per the default loadout

VOLTAGE CONFIG

CPU Vcore Voltage Mode: Override Mode instead of Auto or Adaptive
Vcore Override Voltage: 1.250
Vcore Voltage Additional Offset: Auto
CPU Cache Voltage Mode: Auto
CPU Cache Adaptive Voltage: Auto
CPU Cache Voltage Offset: Auto
System Agent Voltage Offset: Auto
CPU Analog IO Voltage Offset: Auto
CPU Digital IO Voltage Offset: Auto

CPU Input Voltage: Fixed Mode
Fixed Voltage: 1.900v
CPU LLC: Enabled
Every other field here is set to Auto.
 
I don't have intimate knowledge of your Bios and there is little info or guides available to refer to for your MB. Some of it will be trial and error.
Nick Shih is an experienced OCer and provides these profiles for inexperienced OCers who don't want to OC for themselves and don't wish to learn. Don't use default.

CPU Input Voltage: Fixed Mode - Not sure but if you need to change this to enable manual then do so.
Fixed Voltage: 1.900v - change to 1.250V
CPU LLC: Enabled
Every other field here is set to Auto.

Leave other values on Auto for now. If the OC proves stable then go back to your Bios and look at all fields that were set to Auto. If there is a value in the left column then enter that value in the field instead of Auto. This will prevent the value from changing as it can. They are known as static values. Then save the profile.Save a profile.
 


When you get a chance, MM. Take a look at that dropbox link where I've tidied up and updated with more screenshots. 4.6ghz is causing very high temps when fpu is selected in AIDA64... Something like 87'C max although mostly wondering between 79-84. I'm gonna try and lower the voltages a little and see if it is even stable at all as it seems a stretch but no way to know without trying. Courage!
 


https://www.dropbox.com/sh/50mfi5lyuosimzr/AACW97uCZGhq4cNXbgcAXPl_a?dl=0

1. Really not sure what's going on here, MM41. Is it even possible to get a stable OC at 4.6ghz using 1.16v? I read that it isn't, as most people are achieving this clock speed using 1.3v and some very good cooling solutions.

2. Kindly go to the trouble of scouring through the numbers in HWMonitor as I believe the system is compensating for the low vcore voltage by increasing something else. I noticed LLC in HWMonitor at ~1.38v? Or is it the BCLK/Cache Ratio...i'm quite confused. So many sensors and readings.

3. I'm wary of setting it to adaptive from here on as I don't want sudden spikes of voltages to meet load demand. Is it very unhealthy for the VCORE to be fixed at 1.160v for a daily driver machine like mine? I'm currently on a fixed vcore of 1.160v.

4. If point 1 and 2 are non issues (miraculously), would I be reckless to attempt an even higher multiplier?

5. Will keeping the cache ratio lower than cpu ratio allow me to keep temps manageable while pushing a higher clock speed?

When I get up tomorrow, I'm gonna share images of my BIOS setup so you can see if there's certain values somewhere that can help explain the above questions. I hope you don't mind my hassling you over this.
 
Hi steedsofwar :) There's good news and there's bad news. The bad news first.

Your misunderstanding the difference between Core Voltage and VID and please leave Cache ratio the same as Core ratio.

The VID is what Intel rates the chip at for stock speed. It's what your motherboard uses to know what core voltage to set. It's pretty much the stock voltage of the chip. It is not always accurately reported and its not a figure you can change. Do not concern yourself with VID. Its Vcore or Core Voltage which are the same thing and is the actual voltage the chip is receiving.

The Good News.
I know you are a little confused at the moment, but believe me you will look back later and realize how good your readings are. You have run AIDA64 at 4.6GHz for 1hr on the CPU and FPU and that tells me your system is stable and your temps are within the expected range. Those temps on all cores did not exceed 80C so you are good to go my friend. Max temps are over 100C before throttling would start however you have a good chip and should be happy at your current OC.

There is little more to gain and you are very close to your MAX OC. You may be able to achieve more however each % increase in voltage has an exponential increase in Temperature and once you pass 80C then you risk the life of your chip and the difference in performance whilst gaming would be negligible.

Leave adaptive alone as your system at 4.6GHz is good, adaptive is OK for higher OCs which I don't think you have the need to chase and would require Water Cooling to get more.

Have you set your XMP profile for DIMMs performance yet.

Buy now you should have a decent understanding of what you can do in Bios and can be fun if you are realistic on what you can achieve. Try Realbench for a realworld stress tester. It is considered the ultimate test and if you pass the benchmark you can be assured you have a stable 24/7 OC.

By all means share your Bios images for others who will read your thread and its no hassle at all and a pleasure to assist.

 


https://www.dropbox.com/sh/50mfi5lyuosimzr/AACW97uCZGhq4cNXbgcAXPl_a?dl=0

First and foremost, thanks for assisting me along my journey navigating problems and exploring solutions. I wasn't able to update sooner as I was busy tweaking different setups, software and even some minor hardware changes. Then testing them.

Please see the link above again to see some of the results. I wanna say that what you said is very true. After 4.6ghz:

1. The temps go higher by a much larger margin and also require a lot more voltage for stability.
2. My previous 4.6ghz OC that you commented on was on a FIXED voltage and it passed aida64 fpu tests as you saw in the screenshots but it failed Prime95 v26.6 within the first 10 minutes. This is despite temperatures being lower on Prime95! It seems Prime found potential instabilities better that AIDA64 missed. TomsHardware has a dedicated article by a former US vet on overclocking that emphatically states that Prime95 is the best standard.
4. I then took a lot of time trying to learn what adaptive is and how it works and whether LLC was causing the instability, on my fixed voltage OC. I turned LLC off and it seems more stable now and I managed to set adaptive even though I still can't say that I fully understood it. I definitely wanted lower voltage during idle and low load tasks. I tested this at 4.6gh and it passed the 1 hour tests in Prime95.
5. Got brave (or reckless) and switched up a notch to 4.7ghz, 1.2v adaptive -0.001. Temps were easily over 80'C so I knew I needed better cooling. So I got rid of the Noctua NF F12s from the H100i GTX and replaced them with the stock H100i fans that are very noisy but run at a 1000 RPMs faster, hence noisier under load and as quiet as the noctuas when on a good curve in idle. Dunno why I ever believed the hype and went with the Noctuas in the first place. I added two more corsair fans onto the rad to create push pull. I also replaced my intake case fans (expensive 2x Phanteks x series 140mm fans, that look v attractive, are quiet and very smooth) with Noctua 3000RPM fans that are unbelievably noisy! I mean OMG! It sounds like a hurricane, it's unreal.
6. I believe the stock H100i fans made the biggest difference as I'm currently passing the 1 hour mark of both AIDA64 and Prime95 tests just barely within the parameters of safety, by the skin of the teeth. The max they kiss is 81-82'C, although they float in the mid to high 70s. I'm now able to achieve a 4.7ghz OC with 1.2v and a -0.001 adaptive setting. See the screenshots in the link. I remember that JayzTwoCents got similar temps despite that monster setup of his (clearly a more silent setup). Under load, my machine is literally a vacuum cleaner. No joke. I guess this is why the better but more expensive and more complicated solution to the noise is a more beautiful custom water cooling loop, which is in my list of to do's but not anytime this year. My fans will never get that noisy anyway, unless i'm stress testing synthetic benchmark programs. At gaming, they do spin up but not at 100%, so noisy but not insane. This is because my CPU stays between 50 and 60c, often staying at the 40's depending on the game.

I HAVE noticed some strange behaviour but then I was receiving some display driver updates and some nvidia updates too. I will report back if I reckon it has something to do with the OC, instead.
 
Quite an impressive set of results steedsofwar and I can see you'r enjoying your newly acquired knowledge and interpretation of the results. It takes time and patience to get where you are, I'm sure you agree and glad to see your achieving good results.

Thanks for the feed back and be careful of being too enthusiastic as you are in danger of degrading your chip chasing higher than you have. You have an above average chip so take care of it.

A little trick.

UsJW2Ft.jpg