i7-8700k runs insanely hot even at stock speeds ?

[cucker tarlson]

Correct, but many people perceive that statement as 'so why would I run P95 or IBT, its not a guarantee of stability either' which is just plain wrong

It's plain right, IBT is no guarantee of stability at all. Both on 3570K in the past and on 5775c now I found CPU intensive games crash or freeze too high OC/too low voltage even though IBT max. preset could go on forever without an error

 

[EarthDog]

This makes no sense on so many levels.
Let's say you passed 10 hrs of p95. Why then do p95 regularly afterwards, you just found out it's stable. And no stress test really reflects what our normal uses tend to be.

Right, stress tests are generally WORST CASE scenarios. In many cases if you pass a stress test for x amount of time you are stable. The trick is finding the right program and length of time for your needs.

 

[Jetster]

I think the one thing to remember is that taking your hardware into the danger zone with temps can damage it or shorten it's life. If you can keep the temps in a safe place then stress it all you want. The point is to keep it stable right?

 

[Vayra86]

I think the one thing to remember is that taking your hardware into the danger zone with temps can damage it or shorten it's life. If you can keep the temps in a safe place then stress it all you want. The point is to keep it stable right?

The point of stressing is to find limitations, and force it over the edge so you can fine tune your OC.

If that is not your cup of tea, slap an AUTO voltage on there, select a Turbo for each core load, and you're done, with a guarantee of wasting power AND having no grasp on real overclocking or the workings of your CPU.

Its really that simple, either do it proper, or let the board do the work for you. Anything in between? You're a fool in my book, wasting time on nothing substantial and trying to sell that as truth on a forum is a BIG no-no.

 

[EarthDog]

I think the one thing to remember is that taking your hardware into the danger zone with temps can damage it or shorten it's life. If you can keep the temps in a safe place then stress it all you want. The point is to keep it stable right?

sort of...a safe place is anything under its tjmax at which point it will throttle.

It's plain right, IBT is no guarantee of stability at all. Both on 3570K in the past and on 5775c now I found CPU intensive games crash or freeze too high OC/too low voltage even though IBT max. preset could go on forever without an error

Then IBT isnt the program for you. Not all stress tests are created equal. Its just that simple.

I run realbench for an hour or so and prime95 small fft for an hour or so... no issues for my uses using this methodology. As always, YMMV.

Let's say you passed 10 hrs of p95. Why then do p95 regularly afterwards

just caught this... you are misinterpreting what i am saying i think(?).

Once im stable, im stable and dont run tests anymore (who does??). I use a two part testing method as it covers my usage model the best and helps me to find a stable clock at the lowest voltage possible best. For many a single app is fine in a lot of cases. Sometimes it isnt. This is why every person here has a different method which works for them. They arent the end all, but its better than not testing or having too much fat on the bone (higher voltage than needed).

 

[Bund]

TechSpot https://www.techspot.com/downloads/4965-intelburntest.html

My idle temps CPU Core Voltage @ 1.080v (stock frequencies), ram Voltage @ 1.35v, 3333Mhz CL-16-18-18-36 2x8GB, CPU fan speed at stock, case fans 700, 700, 620rpms. Thermal paste Arctic Cooling MX-4

Can you help me? What settings did you change on BIOS to get that? Take a look on my case below.

 

[Bill_Bright]

torturing your CPU with stress tests for hours days and weeks. You proved absolutely nothing.
Sure, they're helpful, mainly to see if you're not hitting some sort of power limit

You proved your case and HSF assembly are providing adequate cooling - a good thing. But that could have been proven in less than an hour.

I think the one thing to remember is that taking your hardware into the danger zone with temps can damage it or shorten it's life.

I agree. It's like taking your car to the race track and red lining it. Even if it doesn't blow up, you took a few miles off its life.

Frankly, I think all those torture tests get you is bragging rights. They sure don't increase performance or give better game play.

 

[cucker tarlson]

sort of...a safe place is anything under its tjmax at which point it will throttle.
Then IBT isnt the program for you. Not all stress tests are created equal. Its just that simple.

I run realbench for an hour or so and prime95 small fft for an hour or so... no issues for my uses using this methodology. As always, YMMV.

just caught this... you are misinterpreting what i am saying i think(?).

Once im stable, im stable and dont run tests anymore (who does??). I use a two part testing method as it covers my usage model the best and helps me to find a stable clock at the lowest voltage possible best. For many a single app is fine in a lot of cases. Sometimes it isnt. This is why every person here has a different method which works for them. They arent the end all, but its better than not testing or having too much fat on the bone (higher voltage than needed).

I know overclocking involves determining the right voltage, but people do not need to overreact this much on a slight voltage excess, as long as it does not influence the temperatures.

 

[EarthDog]

Who's overreacting?? You overclock, you stress test. If you dont want to do that, have at it...

I dont condone several to 24 hours hours of testing for anyone unless they are overclocking and need their system for productivity. That said, it surely has a point and a need for those that dont.

Again, it simply comes down to knowing the right program or combo for your uses.

just play games and run stuff as usual,

i disagree with this assertion is all. While it works for you, i simply prefer to run tests to find instability and resolve it, way way more often than not, in one 'session'. To me, it beats a freeze a day or few, a week or couple, later and have to again adjust settings - perhaps multiple times even. Get my testing done at once.

 

[sneekypeet]

Can we all please get back to helping rather than debating the validity of stress testing? Pretty sure we have a bazillion threads with this argument in it already!

 

[notb]

Can we all please get back to helping rather than debating the validity of stress testing? Pretty sure we have a bazillion threads with this argument in it already!

And yet it seems there are some fairly different ideas and opinions, often contradicting. So why not write an article about it on TPU? You know... some proper tutoring on OC. Many pages do that. You know... add a "best OC stability testing procedure of 2017" to the year-end summary. That would be really something...

 

[sneekypeet]

And yet it seems there are some fairly different ideas and opinions, often contradicting. So why not write an article about it on TPU? You know... some proper tutoring on OC. Many pages do that. You know... add a "best OC stability testing procedure of 2017" to the year-end summary. That would be really something...

So do it, but this is not the place for said discussion. Also it will be hard to do an end all be all thread on the topic, as the previous comments have shown, opinions on the matter differ. Move along now please.

 

[Vayra86]

Can we all please get back to helping rather than debating the validity of stress testing? Pretty sure we have a bazillion threads with this argument in it already!

The cause and solution to OPs problem have already been identified (cpu auto voltage too high), which was then followed by a back and forth about why you should or should not touch your bios at all. In the end we are still on topic: OP should stress test his rig at lower, manually set vCore and check temps again.

 

[sneekypeet]

Since post 129 all I saw was a bunch of winging about who knows how to stress test better. Not the topic of the thread.

 

[cucker tarlson]

Did he specify his cooling solution ? I know 1.32v is too damn high for stock, and I know nothing about Coffee Lake, but should he be getting +95 degrees in IBT at 1.317v ? My CPU barely grazed 70s in IBT at 1.416v on air.

 

[Vayra86]

Can you help me? What settings did you change on BIOS to get that? Take a look on my case below.

Not enough info.

We need your CPU-Z in there *while you run a stress test* at the very least. Also, cooling solution, what have you done so far, etc.

Same goes for @Verbatim who is now down to giving us idle voltages and asking what's wrong Its real simple: you both probably need to go into BIOS, manually set a good vCore, and test if you are stable. This topic is full of hints as to what could be a good voltage.

There is another solution: sell that K-CPU, buy a non-K and call it a day. Or, as @notb points out, the i5 8600k is a much cooler CPU.

 

[Cocio21]

Both the MSI Z370 SLI Plus and ASRock Z370M Pro4 does OC the 8700K, even with the respective "Enchanced Turbo" (aka Asus MCE) function turned off in bios !

Do check that your mobo dont do this on auto, i got vCore up to 1,47v from this on auto.

Had to punch in the Turbo bins manualy to get it running stock on both mobos

Btw, use Realtemp and run the I7GT exe, works well for getting true core clocks in real time use.

And for those wanting to do stability testing, do Prime/IBT or whatever you feel like first, and then follow up with OCCT´s PSU test, this will load both Cpu and Gpu hard, making it a better overall gaming stresstest than IBT or Prime

 

[Krit]

Is thats true that Thermalright Macho Rev. B performs the same as Be Quiet! Dark Rock Pro 3 ?

 

[trog100]

if the 4 core 7700k chip runs hot at higher clock speeds.. its petty obvious the 6 core 8700K will do at least the same..

at least its pretty obvious to me..

trog

 

[Cocio21]

update to my first post.

Realtemp falsely shows 4,7Ghz when running at 4,3Ghz for some reason, use HWinfo64 instead its much better for getting sensor readout and so on coffee lake.

I reconfirmed my MSI board still does OC to 4,7Ghz/1,47v with auto settings/MCE off with HWinfo....

Verbatim set this for :

SA volt = 1,050v
IO volt = 0.950v
and then bump from there if needed

Most 8700K will do DDR3600 on stock IMC related volts, so no need to add the extra heat to the Cpu, my MSI board set 1,36v/1,25v on auto just using DDR3000 ram, way way too much and totally unneeded.

I locked in most of my volt settings (see pic), disregard SA/IO is higher then the selected value, took the pic before i saved bios/restarted

/E

Trog100

I would argue that this is caused by board mfg´s auto settings in most cases, Coffee lake 8700K IS cooler running than 6700K/7700K, but in any case its dependant on the board settings (mainly being MCE functions and SA/IO volts when set to auto)

For me 4,7Ghz is no problem at 1,47v auto settings (i have custom water with external 280 Rad) its just again totally unneeded and wastefull with 1,47v @4,7Ghz

 

[Hood]

IBT is way too much stress for any CPU newer than Ivy Bridge. I noticed this after my upgrade to Haswell Refresh 4790K, 3 or 4 years ago. Instead, try using AIDA64 System Stability Test.

 

[trparky]

IBT is way too much stress for any CPU newer than Ivy Bridge.

Whaaaa...

 

[eidairaman1]

update to my first post.

Realtemp falsely shows 4,7Ghz when running at 4,3Ghz for some reason, use HWinfo64 instead its much better for getting sensor readout and so on coffee lake.

I reconfirmed my MSI board still does OC to 4,7Ghz/1,47v with auto settings/MCE off with HWinfo....

Verbatim set this for :

SA volt = 1,050v
IO volt = 0.950v
and then bump from there if needed

Most 8700K will do DDR3600 on stock IMC related volts, so no need to add the extra heat to the Cpu, my MSI board set 1,36v/1,25v on auto just using DDR3000 ram, way way too much and totally unneeded.

I locked in most of my volt settings (see pic), disregard SA/IO is higher then the selected value, took the pic before i saved bios/restarted

/E

Trog100

I would argue that this is caused by board mfg´s auto settings in most cases, Coffee lake 8700K IS cooler running than 6700K/7700K, but in any case its dependant on the board settings (mainly being MCE functions and SA/IO volts when set to auto)

For me 4,7Ghz is no problem at 1,47v auto settings (i have custom water with external 280 Rad) its just again totally unneeded and wastefull with 1,47v @4,7Ghz

Try Hardware Monitor from CPUID or try openhardwaremonitor

 

[jboydgolfer]

Whaaaa...

This is a guess, but I'm thinking that's a comment on the switch from solder to lesser capable thermal paste ?

I can remember running IBT on my 2500K overclocked to almost 5GHz and it handled it like a champ under a simple 120 mm aio, but I haven't run it in years so it's just a stab in the dark

 

[TheoneandonlyMrK]

He set all cores at 4.3.

There's something wrong here?? Its a bit warm for the clocks, but i see nothing remotely "insane" about these temps.

Edit: i see... hes at stock... that is where all cores boost is 4.3. Since hes stock, voltage at 1.25 is likely too much and is on auto.

I don't know about these chips but could the Imc be to blame with that memory speed, try stock memory clocks.