Prime95 crashes everytime on same core

[chiro]

Hi,

I'm trying to use asus 5ghz OC profile on my 7700k and when I try to use prime95 to stress test cpu my second core fails immediately after start with FATAL ERROR, while others work fine.

What's the problem ?

Note: I've never done any OC before that's why I'm trying out that profile instead of normal overclocking.

[fivemack]

Originally Posted by chiro

Hi,

I'm trying to use asus 5ghz OC profile on my 7700k and when I try to use prime95 to stress test cpu my second core fails immediately after start with FATAL ERROR, while others work fine.

What's the problem ?

Note: I've never done any OC before that's why I'm trying out that profile instead of normal overclocking.

There is no problem. Your second core is the least capable one, and it cannot run at 5GHz. Try overclocking less.

[tServo]

Originally Posted by chiro

Hi,

I'm trying to use asus 5ghz OC profile on my 7700k and when I try to use prime95 to stress test cpu my second core fails immediately after start with FATAL ERROR, while others work fine.

What's the problem ?

Note: I've never done any OC before that's why I'm trying out that profile instead of normal overclocking.

According to my sources & experience, you won't be able to do 5Ghz unless you:
use liquid cooling & delid & replace the Intel's TIM with something better than the toothpaste they use.
Even then, only 3/4 of 7700K processors will get to 5Ghz; 25% will fail due to the manufacturing variances.
Don't believe all those youtube videos or posts in gamer's blogs where they blow smoke about the speeds they reached.
A lot of that is pure posturing & bragging. Also, they boot once at that speed and that's all.

The system's memory has a lot to say about stability at those speeds also. It had better be top-notch.

BTW, what is "normal overclocking" as opposed to other kinds of overclocking?

[chiro]

Originally Posted by tServo

According to my sources & experience, you won't be able to do 5Ghz unless you:
use liquid cooling & delid & replace the Intel's TIM with something better than the toothpaste they use.
Even then, only 3/4 of 7700K processors will get to 5Ghz; 25% will fail due to the manufacturing variances.
Don't believe all those youtube videos or posts in gamer's blogs where they blow smoke about the speeds they reached.
A lot of that is pure posturing & bragging. Also, they boot once at that speed and that's all.

The system's memory has a lot to say about stability at those speeds also. It had better be top-notch.

BTW, what is "normal overclocking" as opposed to other kinds of overclocking?

Well, I have Kraken x62 cooler. After OC my temps were 90-91°C on all cores so that's pretty bad. But I'm planning on doing delid but I'm still trying to find some retailer in Europe that has it in stock or ships it into my country.

I tried prime without OC and my temps were all 80+°C exept one core which was 90°C. That's even worse I guess.

As for memory I have G.SKILL 16 GB KIT DDR4 3200 MHz CL16 Trident Z RGB.

By "normal" I meant I didn't change clocking or voltage of CPU manually but via my mobo's 5Ghz profile.

[mackerel]

In addition to what's been said, there's a difference between reaching 5.0 GHz, and having 5.0 GHz Prime95 stable. I wouldn't count on it at all outside of extreme cooling.

I think you have AVX offset in that CPU generation, you can use that to reduce clocks while Prime95 is running. Of course, if you do so then Prime95 wont get the 5 GHz treatment, but other tasks still can. They can be stability tested separately.

Back a step, is the system stable at stock (no overclock)?

[chalsall]

Originally Posted by chiro

I tried prime without OC and my temps were all 80+°C exept one core which was 90°C. That's even worse I guess.

Please don't guess. Know.

One of the most important uses of Prime95/mprime is to ensure the sanity of hardware. Some use CPUs for very important analysis, and they would like to ensure their kit is sane.

Have you tried running any Double Checks with your new system?

If not, it would be highly recommended. If you did, did the residuals match?

[M344587487]

Originally Posted by tServo

...
Don't believe all those youtube videos or posts in gamer's blogs where they blow smoke about the speeds they reached.
A lot of that is pure posturing & bragging. Also, they boot once at that speed and that's all.
...

Reviews from big players also need to be taken with a pinch of salt, sending out golden review samples is probably common practice: https://www.youtube.com/watch?v=pl0we6-ZiQY

[chiro]

Yeah, at stock system is totally stable. No crashes. Well, I have pc for gaming only, I don't do some render/graphic or some data heavy projects.

What do you mean by Double Check ?

[chalsall]

Originally Posted by chiro

What do you mean by Double Check?

Sign up to get an account at https://www.mersenne.org/

Install the client appropriate for your system, available at this page.

Launch the client and answer the questions.

I hope that helps.

[science_man_88]

Originally Posted by chiro

What do you mean by Double Check ?

https://www.mersenne.org/various/mat...ouble_checking is what was being referred to.

[CRGreathouse]

Originally Posted by M344587487

Reviews from big players also need to be taken with a pinch of salt, sending out golden review samples is probably common practice: https://www.youtube.com/watch?v=pl0we6-ZiQY

It wouldn't surprise me. But I must disagree with the video's claim that "a sample size of 2 is absolutely meaningless from a statistics perspective". If Linus bought a random CPU against which to compare his review CPU, you can conclude, with 95% likelihood, that the difference between the two is less than 4.5 standard deviations in either direction. This holds regardless of what the compared quality is, as long as it can be given as some real number, and it holds regardless of the particular distribution of values (although the distribution will, of course, determine what the standard deviation is, and so determine th scale of the result). You can use Chebyshev's inequality to increase or decrease the confidence and Cantelli's inequality to make it one-tailed (making it 4.36 standard deviations instead of 4.48 for 95%).