i5 4690k Overclocking: ASRock Z97 Extreme3 Mobo, BIOS Settings & HW Monitor Voltages

Hello!
As the topic I want to crank up the i5 4690k. The motherboard is Asrock Z97 extreme3. Below I am giving a link with photos to my bios settings. The last photo is a screenshot from HW monitor:

MY SETTINGS

Maybe a stupid question, but should the HW VCORE CPU monitor be around 1.32V and not 0.940V. I thought it should be the sum of vcore adaptive voltage and offset voltage. I might be wrong In any case, with these settings, the system crashes at the beginning of the intel burn test. Previously, the clock was set to 44 at vcore voltage 1.200 and offset voltage + 0.010V. It was a 2-year-old girl and it was OK but I would like to get something more out of him. Can be? Maybe I'm doing something wrong?

I recommend taking a look.
https://www.google.pl/?gws_rd=ssl#q=i5+4690k+overclocking+guide&*

And when it comes to tensions, nothing is ever 100% accurate. There is always a measurement error, and in the worst case the power supply needs to be replaced.

But if I did not set it, the HW monitor shows a similar value. Therefore, I wonder if the right amount of V goes to the proc, i.e. the one I set in the bios. Unless there is still something in the settings that I do not know, but I have not read it anywhere.

You should start with the fact that you can see that you have no idea what is inside your computer.

Your RAM is most likely clocked at 2400MHz.
In one of the photos there is an XMP profile for 2400MHz while you are set to 1333MHz.

Therefore, I would ask for a complete computer configuration.
Exactly about the power supply, frame bones (preferably screenshots from CPU-Z, memory tab and SPD for each slot) and CPU cooling, the rest is less important.

Secondly. Filming the Cache gives little tangible results. Don't tire your CPU with turning the cache. Turn to factory and leave it there.

Next case. Adaptive voltage on haswell is not a good idea. Problem rolled over many times. The voltage jumps too aggressively in this mode when loaded. For this value you have set to AUTO. It's asking for trouble.
Mode for fixed / manual and then you start working on the OC. Adaptive voltage can only be introduced when you have an idea about the capabilities of the processor. You do tests and watch how the voltage behaves.

CPU input voltage should not be increased unnecessarily. When you hit a barrier, you only increase it. It is worth noting that passing over 2V at this point sometimes "magically" removes the clocking barrier which cannot be broken by increasing Vcore (on my 5820K 4.6GHz it was not stable even at 1.4V / 1.95Vccin but 1.34V / 2.05Vccin was already stable).

Thanks for showing interest

I provide the full specification (copied from the order history in apricots ). Everything is here:

Crucial MX100 512GB SATA3 2.5 "SSD 550/500 MB / s 7mm (CT512MX100SSD1)

Asus Xonar DG sound card - PCI card, System 5.1 - Xonar DG

Asus GeForce GTX 780 graphics card, 3GB DDR5 (384 Bit), HDMI, DP, DVI, BOX (GTX780-DC2OC-3GD5)

Asus DRW-24F1ST E-Green 24x SATA drive (DRW-24F1ST / BLK / B / AS) OEM new model DRW-24F1MT / BLK / B / AS

Fractal Design Define R4 Black Pearl (FD-CA-DEF-R4-BL) case

Team Group Vulcan DDR3 2400 2x8GB TLAD38G2400HC10TDC01 Memory

ASRock Z97 EXTREME3, Z97, DualDDR3-1600, SATA3, RAID, HDMI, DVI, D-Sub, ATX motherboard (Z97 EXTREME3)

Intel Core i5-4690K, 3.5GHz, 6MB, BOX processor (BX80646I54690K)

Corsair CS Series 650W Modular 80+ GOLD power supply - CP-9020077-EU

Asus PCE-N10 network card, wireless N150 WiFi adapter, IEEE 802.11b / g / n, PCI Express, transfers up to 150Mbit / s, 64/128-bit WEP, WPA / WPA2-PSK, WPS [Low profile] support

Thermalright HR-02 CPU cooler - Macho - Black White FM1 AM2 AM3 AM3 + LGA775 LGA1150 LGA1155 LGA1156 LGA1366 LGA2011 - TR131

Microsoft Windows 8.1 PL 32/64-bit DVD BOX (WN7-00934) operating system

Eizo FS2333 monitor (FS2333-BK)

I completed the components thanks to the suggestions of forum users less than 3 years ago. It was OK, so I never really went into specific settings

Perhaps, let's start with the basics, i.e. temperatures.

Factory settings. You run a program like OCCT (prime95 with newer processors is not really). You write down the temperatures during the stability test.

Then you give Vcore at exactly 1.275V in manual mode (don't touch the clocks). You repeat the procedure. You have to see what the temperature reserve looks like. Processors from Ivy Bridge (except LGA2011 / 2011-3 where the core is soldered to the diffuser, reason why I have 5820K) upwards have poor thermal paste between the core and the diffuser. The temperature then depends on how it decomposes. Therefore, it should be established individually for each copy.

Only after determining what is the stock, you can start the ticking of the clocks.

I was gone for a while, I am already taking it

I have reset the bios settings to the factory settings. Test done in IntelBurn Test at maximum load. Max temperature: 58 degrees. Below is a screenshot:



Then:
CPU Vcore Voltage Mode changed from Auto to Override Mode
Vcore Override Voltage changed from Auto to 1.275
Vcore Voltage Additional Offset left at Auto

Test done in IntelBurn Test at maximum load. Max temperature: 73 degrees. Below is a screenshot:



PS. I changed the paste about 2 years ago because the temperatures did not suit me. I put on GC-Extreme.

I think that you will get to 1.35Vcore on this cooling (temperatures will probably be around 85 ° C under full load).

You're giving 1.275Vcore for now and trying to max out.
Input Voltage (default 1.8V) give 1.9V which would not limit it yet.

Vcore Votlage additional offset, if I remember correctly, increased the voltage under load.
You can give 0.05V auxiliary.

You can use HWinfo to see exactly what is going on with Vcore during load tests.

With this CPU you aim to around 4.5GHz. You start at 4GHz and go all the way to 4.5GHz testing 10-15min OCCT each time.

When it crashes, slowly increase the voltage up to (for now) 1.325V.
Monitor temperatures. When it starts to reach 85-90 ° C (in stress test) it means that you can't do anymore. During normal operation the temperatures will be lower.
You go down to the last stable (also if he doesn't want to go higher at 1.325V) and give up the OCCT test for a few hours.

As a last resort, you can try Input Voltage at a little over 2V (e.g. 2.05V), but be careful that the LLC does not break it much higher. Therefore, you should monitor both Vcore and Input Voltage (Vccin) all the time.
However, this is a last resort as it does not want to move higher even with a very high Vcore (e.g. It does not go up 100MHz even when increasing from 1.32V to 1.4V). Generally, the processor usually "ends" in these areas, so it is only for squeezing the last MHz.

Ok. So the BIOS restored to factory settings I am writing step by step what I change:

Vcore Override Voltage: 1.275
Input Voltage: 1.9
Vcore Voltage Additional Offset: 0.050
Multi Core Enhancement: Enabled
Cpu Ratio: All Core
All Core: 40

First test and ........ temperature 90 ° C. Below is a screenshot from OCCT, Intel Burn Test and Hw monitor:



Another test at All Core: 41 and 92 ° C



Another test at All Core: 42 and 94 ° C



Another All Core test: 43 and 94 ° C again, except that I changed the power of the fans from 5V to 12V (I accidentally switched yesterday when vacuuming). Stable of course, but temperatures .... !!



Another test with All Core: 44 and ... 98 ° C




And what's interesting with the following changes:

All Core left at 44
Vcore Override Voltage: 1.200
Input Voltage: Auto
Vcore Voltage Additional Offset: Auto

max temperature 79 ° C



And steadily all the time. All tests are done after about 12 minutes. at max settings.

Neither the OCCT nor the HWmonitor is reading the Vcore voltage correctly.

Both show less than 1V. With this voltage, there is no question of such temperatures.

Use HWinfo and not HWmonitor.
https://www.hwinfo.com/
If there will be a defeat, Intel XTU will be.

Disables the colleague options such as Intel Speedstep, C1E, C3 / C6 states, Spread Spectrum.

They make life difficult with third party liability insurance and can make the readings false.

Judging by the fact that at only 1.2V it is stable 4.4GHz, I suspect that there is a significant increase in voltage under load. He will turn his colleague Vcore voltage additional offset to 0 and then check what is happening with the temperature. 1.275 + 0.05 = 1.325V. The only question is whether the album sticks to it.

There may also be a question that simply 1.325V is too much for this cooling.

With the Vcore reading I had it all the time and I didn't know what was going on either.
Intel Speedstep and Spread Spectrum disabled. I can't see C1E, C3 / C6 states.

Left 1.2V, and Vcore Voltage Additional Offset set to 0.001 (I can't give 0, because it goes into auto). Test about 25 minutes. and the result:



Clocks all the time at 44. It is stable and the temperature is also OK.

Another test approx. 25 minutes. with the same settings with HWinfo enabled.
Indeed 1.2V And what now?



All the time 1.2V and clocks at 45. Still stable and only the temperatures jumped to 83 ° C (I do not know if it matters but I changed the RAM clock to 2400 because after restoring to factory settings it jumped back to 1333)



When changing to 46, the computer freezes while booting

pawelr98 ...?

Vccin set to 1.9V?

Show what the input voltage looks like in hwinfo (look for in the line describing the motherboard) with and without load.

4.5GHz at 1.2V is not a bad result. The temperature reserve is slowly running out. I think 4.6 GHz will be the end of the possibilities on this cooling.

4.6GHz can be stabilized with a slightly higher Vcore, depending on the input voltage level.

Fixed Voltage was tested above 1.9V. After changing to 1.8V stability problem.
Then Vcore Voltage Additional Offset set to 0.010, Vcore Override Voltage left 1.2V, and Fixed Voltage 1.8V and all during the test OK. this is how the computer went for a week and it was OK.

No load:


After the test:

I guess it will stay that way