What type of CPU stress testing is sufficient for my needs -> CPU rendering?

[paw4444]

From what I have read so far I see two different approaches in CPU stress testing to determine a stable or sufficiently stable overclock.

One uses extreme stress tests like Prime95 (For many hours) or IntelBurnTest (For a quicker alternative to Prime95) and the other uses something less demanding like the x264 Stability Test overnight or other options which do not stress the CPU to a degree which would not be met anywhere outside the synthetic tests.

The type of work I would do is CPU rendering.
What type of stress testing would be sufficient to determine a reliable overclock that would use such load as the Corona Benchmark or Cinebench for a prolonged time?

 

[bicycle_repair_man]

A stress test that pushes all the CPU threads to 100% is the best bet as it shows that the CPU can cope in a worst-case scenario. The Prime95 Small FFTs test does this.

The problem with emulating a typical workload is that there's no such thing as typical. A CPU overclock may work just fine in Cinebench but may crash during another application that's more CPU intensive.

 

[paw4444]

Thanks. I will stick with the extreme stress tests to make sure the system is running reliably as I am already pleased with the performance gains of the overclock at the current voltage.

 

[Aeradom]

Thanks. I will stick with the extreme stress tests to make sure the system is running reliably as I am already pleased with the performance gains of the overclock at the current voltage.

Just to piggy back here, I'd run the test over night just to make sure everything is stable as sometimes an hour or two test is insufficent. I ran mine over night after verifying it appeared stable with an Intel Burn Test. In fact, I'd run the Intel Burn Test as if the computer is going to crash, it'll do it then. Afterwards, go to the Prime 95 for 8 hours.

 

[chenuki]

A 8k min/max FFT size is good stability test for cores at full load. You can also use prime95 version 26.6 to test without AVX for higher stable OC: http://overclocking.guide/stability-testing-with-prime-95/

 

[paw4444]

Thank you! This helps clarify why temps are so high in Prime95 if compared to all other tests I have done.
Now that I have a better understanding I can go back and reevaluate the OC of the CPU.