[SOLVED] Which stress test is more "realistic"?

[Hmb556]

So I recently upgraded to a 3900x from my old i7 4790k. Did some stress tests and it ran pretty hot on prime95 at 81C with some spikes to 85 (I liked my 4790k temps better at like 65 max). I tried the CPUZ multicore stress test and that got a much better result of about 72C. Im using a corsair h115i I think, the 280mm one. I dont know the differences in how they stress test, which one would be closer to real world temperatures if I'm gaming and streaming? Thanks

 

[drea.drechsler]

RealBench

Keep in mind the whole idea behind a stress test is to put an unrealistic load on the system, so even that's not 'realistic'. The processing loads it uses are 'real world' applications: running a POVRay rendering, a Handbrake H.264 video encoding and a Luxmark ray-tracing and a few other things too; but it's highly unrealistic because it's doing it all simultaneously. It's very demanding on memory and drives too.

Another 'real-world' processing stressor is Cinebench 20; you can set it up to run like 120 minutes (2 hours) of re-rendering the same scene continuously.

Or run some Folding@Home tasks overnight and at least put the power you're using to some good use.

 

[Hmb556]

Ok thanks I will check those out as well. Would you say those temperatures are fine or should I try something like reseating the heat sink?

 

[drea.drechsler]

Ok thanks I will check those out as well. Would you say those temperatures are fine or should I try something like reseating the heat sink?

Looks very good to me. 80-85C isn't unexpected for a 3900X, and just as you suspect, Prime95 is about the most unrealistic of processing loads possible.

You're on an AIO cooler so you'll want to know when the water in it thermally saturates. Run Prime95 continuously for several hours keeping an eye on temperature. It should climb then stabilize and then start climbing again when the liquid saturates. If it does it will be maybe one to two hours into the test.

That is IF it saturates: the radiator may be able to shed the heat to the air stream as fast as the processor generates it.

 

[RodroX]

Besides what drea.drechsler wrote, keep in mind that the Ryzen 9 3900X has 12 physical cores, while the Core i7 only have 4, thats 3 times more real cores than need to be cooled down.


Probably some stupid questions but:

Did you clean and changed the thermal paste after swaping Motherboard and CPU?

Did you did a clean Windows install?

 

[Hmb556]

Besides what drea.drechsler wrote, keep in mind that the Ryzen 9 3900X has 12 physical cores, while the Core i7 only have 4, thats 3 times more real cores than need to be cooled down.


Probably some stupid questions but:

Did you clean and changed the thermal paste after swaping Motherboard and CPU?

Did you did a clean Windows install?

Yeah this was basically an upgrade of everything except the cooler, I cleaned the old stuff off with some alcohol before putting the new paste on. My only concern was maybe I did a bad application and could lower temps by trying again, but if those aren't crazy high results I won't bother. Thanks guys.