stability AIDA Stability test VS others like Prime,OCCT etc

[dmitriy.p]

Hey guys,

First of all, i love AIDA, as this a thing from my childhood, which i was use with my Athlon 900 and i don't even remember how much years ago it was... I would like to thanks it's creators for this pure masterpiece  

Testing my 6600k OC modes i was running few popular stability tests: Prime, OCCT, Linx -- all of them burn my CPU quite well and using top clocks i always get an errors or reached temperature limits (i have only performa 10).

When i'm doing AIDA Stability Test, my CPU heat is far lower, than with "guys" from above, somewhere around 60-65 C* and without any errors! So it is possible to add more MHz's and make voltages lower a bit, then i was needed for OCCT and others.

so, my question is: can i rely on AIDA stress test and keep my OC boundaries higher, than i'd reached with those hellish apps? If you gonna ask me, i use my PC just for games and browsing.

I understand, that software from my example designed for ultimate testing purposes, with abnormal CPU overloading which gonna never happen in real life. BUT... if my CPU is ok with AIDA stress test, may be this means i could play any games being OC'ed without risk to receive BSOD or having some of my cores "lagging" at the background?

ps: i think AIDA is not using AVX instructions when stress testing, no,?

 

Please, help me figure out with this one.

Thanks!

please, don't kill me if this thread already exist somewhere 

[Fiery]

It depends on what subtest(s) you enable. If you only enable FPU, then you can reach the maximum CPU temperatures.  It's because the FPU subtest uses heavy AVX and FMA floating-point code, and that is the method to use on modern processors to get them heat up the most.

However, that may not be the best method to hunt down computing errors or RAM CRC errors.  If your intention is error-checking, then it's best to enable CPU+FPU+RAM+Cache substests altogether. That way your processor will not run as hot as with the FPU subtest alone, but the workload will be much more diverse, and so potential computing glitches will be revealed much easier.

In most cases we recommend at least a few hours of FPU stressing, in order to check if the CPU and chassis cooling solution are both adequate. Then, if it passes, it's best to run the CPU+FPU+RAM+Cache subtests as well, for a few more hours. Optionally you can also enable HDD and GPU stressing in this second round, although in most cases it will take precious CPU cycles away from the other 4 subtests.  Of course if you have time and you want to be 100% thorough, in a 3rd round you can check all subtests and put a few more hours into yourstress testing session  

[dmitriy.p]

thank you, Fiery! i wasn't thinking, that enabling separate features brings some focused tests! i've just enabled all pc + ram ones.

however, i will have to include more test hours indeed, as 1h was not enough -- i had bsod after playing one of my games for ~40 minutes. 

 

thanks again!