Seperate stress tests for CPU and RAM

[Costas Athan]

Prime95 blend stress test tests along with the CPU the RAM too, right?

Isn't it better to run separate tests for the CPU and the RAM DIMMs like OCCT Linpack and Memtest86+ respectively?

 

[Idontcare]

Correct that Prime95 blend tests both. But you can run "small FFT" and focus the stress test on the just the cores, or run "large FFT" and focus the stress test on the IMC and memory.

 

[Costas Athan]

Correct that Prime95 blend tests both. But you can run "small FFT" and focus the stress test on the just the cores, or run "large FFT" and focus the stress test on the IMC and memory.

Yes, that's what I'm talking about. I didn't mean that you have to use OCCT Linpack test necessarily. I suggested it because for what I have read it stresses the CPU to its limits.

But anyway, isn't it better to run separate tests for the CPU (Prime95 Small FFTs if you like) and the RAM (Memtest86+). Or a Prime95 blend test is more reliable? In other words my question is: Which is the best way to go in order to test system stability? A combined test like Prime95 blend test or two separate stress tests like the ones I mentioned?

 

[Idontcare]

What you say is true. Generally we want our stress tests to focus on specific hardware components in isolation of stressing secondary hardware because then we know what component is the problem when a stress test fails.

However, some people like to go that one extra step, after testing each component separately, of testing the components together in case there are interactions that create instabilities of their own.

That is what blend is good for.

 

[Smoblikat]

What you say is true. Generally we want our stress tests to focus on specific hardware components in isolation of stressing secondary hardware because then we know what component is the problem when a stress test fails.

However, some people like to go that one extra step, after testing each component separately, of testing the components together in case there are interactions that create instabilities of their own.

That is what blend is good for.

^^ 100% correct.

Back in the day I ran OCCT on my CPU, and memtest86 on my RAM. If they both passed I would do a superPI run, and maybe a P95 blend/IBT on high.