[SOLVED] Overclocking I7 6700k

[Kurome]

Hello forum members,

I've had time on my hands recently with the COVID-19 pandemic so I decided to overclock my processor i'm at 4.2 Ghz currently but i'm a little concerned about the temperatures during stress testing

I am using a combination of stress test programs (Prime95 version 26.6 and intel's XTU software since I don't trust myself with the bios settings.

I'm monitoring the state of the processor using RealTemp. CPU-Z and the montoring utility inn XTU. i'm doing that because i want to see that the numbers more or less agree with each other particularly for the temperature.

at 4.2 GHz I stress test for 2 hours using each of prime95 to see whether the temperature was a problem ( I know people say to do this for a day but the temperature flattens out affter around 1 to 2 hours so I do not test any further)

for the 2 hours I tested for I look for the tempreature spikes in the XTU software because the graph shows me every spike and I have something very close to this for the maximum points under the small FFT stress test

68, 70, 71, 73, 71, 72, 70, 71, 73, 75, 76, 79, 80, 87, 71, 76, 73, 75

RealTemp also said that there was a maximum of 87 so they agree

I then did a stress test using the XTU software and got something similar to these readings again for 2 hours

64, 63, 65,67,68,67,65,61,63,64,67,66,69,70,
71,70,68,70,71,69,67

the other numbers are accurate to around 1 or 2 degrees the maximum was exactly 71 according to XTU and 70 according to RealTemp

I suppose I should mention the small FFT test was carried out 5 or so minutes after the XTU software CPU stress test

My question is:

Should I be concerned about any of these temperatures?

I've read on multiple overclocking guides that temperatures of 80+ are not good but I don't know whether that means a wall of 80+ during a stress test or just one single reading of larger than 80?

i'm further confused and possibly slightly worrried because some guides say "meh 80 is normal but a little warm" others say i'm about to cook my chip and what the hell am I doing so I thought i'd ask here and try to get a definitive answer

One more thing i'd like to say is that my processor reaches temperatures of a maximum of around 70 to 72 degrees if i play an intensive game, its current idle temp is around 30 to 35 degrees and I never use my computer at anything beyond 60% load really

Thank you for taking the time to read this and any help will be greatly appreciated!

 

[Darkbreeze]

Overclocking to 4.2Ghz on the 6700k is pointless because at the default configuration it can already run all four cores at 4.2Ghz.

Plus, using ANY kind of utility to overclock the CPU is plain discouraged. If you can't do it in the BIOS, you shouldn't do it at all. That's really all there is to it. And, it's a LOT easier to get things RIGHT by using the BIOS to overclock anyhow. It's very easy to unnecessarily limit your overclock or get something wrong by allowing a utility to make some of the decisions.

87°C is too high, by about 7°C.

What is your full hardware configuration including motherboard, memory kit model, CPU cooler, case, number of case fans AND the exactly configuration/orientation of those fans, graphics card, power supply (Exact model)?

 

[CompuTronix]

Kurome,

You might want to read these:

CPU overclocking guide and tutorial for beginners

Intel Temperature Guide

 

[geofelt]

I do not see any reason for concern.
Stress testers do just that... stress.
Each method uses different instructions to reach their results.
Those instructions likely bear no resemblance to what you use during a game, for instance.
The key thing to monitor is the vcore.
You do not really want to go past 1.4v.
The higher the multiplier, the higher the voltage needed to support the speed.
Some chip samples do better than others.
As of 12/04/2016
What percent can get an overclock at a somewhat sane 1.4v Vcore.

I7-6700K
4.9 5%
4.8 21%
4.7 64%
4.6 96%

At higher overclocks, the voltage increases.
It is the voltage that increases the temperature.
A stress test running at 85c. is considered as ok.

Do not worry too much.
The processor monitors it's temperature.
If it detects a dangerous temperature, it will thropple or even shut down.
That is around 100c.

 

[Darkbreeze]

Do not worry too much.
The processor monitors it's temperature.
If it detects a dangerous temperature, it will thropple or even shut down.
That is around 100c.

I think this is a bit misleading, because the uninitiated will not understand that at some point, thermally, probably around 85-90°C, you start running the risk of damage from electromigration if you are running a configuration that is consistently seeing those kinds of temperatures, and most certainly at temperatures between 90-100°C you are likely to be irreversibly degrading the CPU. So while the CPU might "shut down" or throttle at 100°C, that doesn't mean you are fine all the way up to that temperature.

I know that, you likely know that, but many of our less experienced members might see that comment as "I am fine up to 100°C".

 

[Kurome]

Thank you to anyone who helped me

 

[Darkbreeze]

Kurome, in MY opinion, 87°C is too high, and I see that it did peak at that temp during your testing. There might be a reason for that which hasn't been accounted for yet.

When you ran the Small FFT test, did you have AVX and AVX2 disabled?

 

[CompuTronix]

He ran 26.6.

 

[Darkbreeze]

I missed that. Thanks Comp.

I guess some are just too shy to answer the questions that are put to them. Like the what their hardware and cooling configuration consists of.

 

[Kurome]

Kurome, in MY opinion, 87°C is too high, and I see that it did peak at that temp during your testing. There might be a reason for that which hasn't been accounted for yet.

When you ran the Small FFT test, did you have AVX and AVX2 disabled?

Apologies for the late reply, i went into my bios last night and the voltage is 1.26V, for some reason only one of the cores is set to turbo boost to 4.2 GHz. If i go into performance tuning it lists 4 turbo boost multipliers (42, 40, 40 and 40) surely all cores should be 42 be default?

Further to that the cooling system being used is quite old (Corsair H80i) its configuration is the radiator is attached to the back of the chasis where all of the motherboard sockets are, there are two fans, one is attached to the back of the case then the radiator is attached to that fan then the second fan is attached to the radiator itself.

inn addition to that there are two stock case fans, the graphic card is a GTX 980 Gigabyte version (it has three fans attached to it) the motherboard is a gigabyte Gaming 5 EU

The power supply is an XTR 750W

memory kit is corsair vengeance 3200mhz DDR4 2 x 16GB

for the AVX settings i'm not sure I know I ran 26.6 because many places said that later versions are not reliable for testing.

 

[Darkbreeze]

Later versions are reliable for testing, but you must DISABLE the setting for AVX2 and AVX, and those options are right there on the main page. I would try the testing again with the newer version.

 

[Kurome]

Later versions are reliable for testing, but you must DISABLE the setting for AVX2 and AVX, and those options are right there on the main page. I would try the testing again with the newer version.

Do I use small FFTs or smallest FFTs in the new version?
Update: Temperature is still very high with Prime95 small FFT even with AVX disabled, might be that I need a better cooling system or new thermal paste,it's not throttling or shutting off but I don't feel like nearly 90 degrees is good even with a torture test.

 

[Darkbreeze]

Small FFT. Smallest FFT's are for specific use cases, not general thermal testing.

Did you disable BOTH types of AVX, including AVX and AVX2? Note that the option to disable AVX doesn't "un-gray" until you've disabled the option for AVX2 in the main options window. 90 degrees is absolutely too high. Anything above 80°C would be considered undesirable. Anything above 85°C would be considered bad. Anything 90°C or higher would be considered "stop and fix the damn problem". Some will tell you that it's fine because "you'll never see those temperatures in real world use".

I say, "Really"? Because that's nonsense. MANY applications and games use AVX instructions these days and while you of course CAN use an offset in the BIOS to account for that, your CPU was designed to be able to be run within thermal spec, which recommends no higher than 80°C under normal conditions, "normal conditions" meaning full 100% TDP at a steady state load. If you can't run your CPU, ANY CPU, at full TDP loads then there is a problem if it's at the stock configuration. If it's overclocked, then there is either a configuration issue such as too high of overclock, too high of voltage, cooling not good enough, motherboard not good enough, too many DIMMs for that overclock configuration, really high ambient temperature, or something.