CPU testing & delidding

[Phizix]

Hi all, 

 

I've recently just built my dream machine and it's my intention to overclock and de-lid her in the next week or two. 

 

I've got prime95 set up and I'm ready to start stress testing but I'm wondering whats a good time frame I should do this for to get an accurate temperature whilst under full load. I've heard different figures being banded about so wanted to get your take on this. What's a good temp for the 8700k as well. 

 

Thanks in advance.

[iRileyx]

Air cooled, AIO, Custom loop cooling?

 

If its an air cooler only about 15 minutes should be enough, AIO give it 30 or so minutes(for a 120/140/240/280mm) and for a custom loop then it depends how many rads you've got.

[Mira Yurizaki]

I would say at least half an hour for a quick and dirty test. Otherwise several hours just to make sure the system is able to handle it.

 

However, Prime95 is a worst-case scenario test. You almost never will encounter something that will give it that bad of a load outside of it.

[Phizix]

15 hours ago, RKRiley said:

Air cooled, AIO, Custom loop cooling?

 

If its an air cooler only about 15 minutes should be enough, AIO give it 30 or so minutes(for a 120/140/240/280mm) and for a custom loop then it depends how many rads you've got.

 

15 hours ago, M.Yurizaki said:

I would say at least half an hour for a quick and dirty test. Otherwise several hours just to make sure the system is able to handle it.

 

However, Prime95 is a worst-case scenario test. You almost never will encounter something that will give it that bad of a load outside of it.

So I ran her today for 30 minutes and these are my results:

 

The snip was taken 30 seconds to finishing. 

[Phizix]

In terms of the temps are they good, bad or ugly? 

 

Like I say it's my intention to delid eventually. 

[Christophe Corazza]

9 hours ago, Phizix said:

So I ran her today for 30 minutes and these are my results:

- snip -

The snip was taken 30 seconds to finishing. 

 

When testing a system for its stability, you should run it a few hours and not 30 minutes.

[Phizix]

7 minutes ago, Christophe Corazza said:

 

When testing a system for its stability, you should run it a few hours and not 30 minutes.

I did clarify this in the above comments and was told that generally 30 minutes is fine. 

 

[Christophe Corazza]

2 minutes ago, Phizix said:

I did clarify this in the above comments and was told that generally 30 minutes is fine. 

 

 

To quickly test for stability, 30 minutes is fine. However such a quick test should always be accompanied afterwards with a thorough stability test that takes longer.

 

If instabilities would occur, they will do so during those few hours. And even then you’re not 100% sure the system is stable. (Therefore real enthusiasts run such a test for a whole day.)

30 minutes is a too narrow time window for system errors to occur.

[Phizix]

1 minute ago, Christophe Corazza said:

 

To quickly test for stability, 30 minutes is fine. However such a quick test should always be accompanied afterwards with a thorough stability test that takes longer.

 

If instabilities would occur, they will do so during those few hours. And even then you’re not 100% sure the system is stable. (Therefore real enthusiasts run such a test for a whole day.)

30 minutes is a too narrow time window for system errors to occur.

Thanks for the clarification, would you also recommend I run Prime95 or another program?

[Christophe Corazza]

I would run Prime95 to stress test the system to the fullest. These are indeed workloads that are not often encountered for a regular PC, but you are testing for stability after all (and less for temperature in this case).