5.0 ghz aida64 stability ???

[Phillyflyer]

Just did a little tweaking with this board and the 3930k.....Wondering what the concensus is as to running AIDA64 for determining 100 % stability.

System Specs:

Asus Rampage IV Extreme (Bios 1404),3930K OC'd 5.0GHZ 1.424V,Water Cooled XSPC RX360 Rad With (6)Scythe Gentle Typhoon 1850 RPM Fans In Push/Pull Config Running @ 1100 RPMS,XSPC RX120 Rad With (2) Scythe Gentle Typhoon 1850 RPM Fans In Push/Pull Config Running @ 1100 RPM,Swiftech Apogee HD CPU Block,Swiftech MCP35X Pump,IC Diamond 24 Carat Thermal Compound,NZXT Sentry LX Fan Controller,OCZ ZX Series 1000W Fully Modular 80 Plus Gold Power Supply,32GB Corsair Dominiator GT 2133 MHZ 9-11-10-27-2T @ 1.5V,(2)Samsung 830 128 GB In Raid 0,Intel RSTe Version 3.5.0.1092,(2)ASUS GTX 670 DC II 2GD5 SLI ,Coolermaster HAF X Case,Asus VE247 LED Monitor,Windows 7 Ultimate 64 Bit.

[Nodens]

Haven't tried its stability test at all. Do we know which engine it's using? Or is it their own proprietary engine?

[Arne_Saknussemm]

I think it is OK....maybe a bit gentler on your CPU than the latest version of Prime. So the "consensus" may well be that it's not hardcore enough 😉

[Amb669]

to determine "100% stability" is not possible using only one program alone, every piece of software tests your computer in a different way.
although AIDA64 does lot's of different kinds of tests, it still can't test everything in every cpu, it just can't happen with the infinite amounts of code your cpu can handle
your pc can 100% pass all Linx, AIDA64, memtest86, intel burn test, prime95 testing and still fail at stability during a game .
my general rule of thumb, after using programs to test stability and everything passes with flying colors your 99% done testing for stability, the rest of stability testing is just in general use of your daily computing, things you do day to day on your pc, gaming ect... i'v seen pc's pass every kinda test thrown at them and fail on a game due to bad insufficient voltage, quick voltage fluctuations ect.

/edit by the way, nice overclock on that 1.35v :cool:

[Zka17]

I'm using AIDA64 for while now... I have to agree, it's not so hard on the hardware as the prime applications are... and it runs a lower temps... - however, I do love it's temp monitoring feature! :cool:

One thing based on OP's picture... stop using more than one temp assessment application!

[Dartmaul]

I don't think that 3930k can be stable on 5Ghz with only 1.425v Vcore.
AIDA has very poor CPU stress test so passing it doesn't says about any stability at all.

While overclocking the CPU I use both LinX and prime95.
LinX is useful while looking for some average frequency\voltage. If you can pass 8GB LinX stress that means you are somewhere near from stable spot. When reached some nasty frequency I'm using prime to find out optimal voltage.

[Zka17]

Dartmaul wrote:
AIDA has very poor CPU stress test so passing it doesn't says about any stability at all.

Are you sure about this? How long and what settings are you using AIDA64? For me, if it passes for 1 hour (selected CPU, FPU, cache and memory), then it will pass everything! Only thing is that the wPrime 1024M will produce higher max temps...

[Dartmaul]

Zka17 wrote:
Are you sure about this? How long and what settings are you using AIDA64? For me, if it passes for 1 hour (selected CPU, FPU, cache and memory), then it will pass everything! Only thing is that the wPrime 1024M will produce higher max temps...

As I've said before, I believe that 5ghz cannot be stable on 1.425v.
There are a lot of software (stress tests and not only) they could give 100% CPU load, but it doesn't means real stressing and stability.
Also yes, I've seen many times when CPU can pass AIDA stress but fails after few seconds of prime.
I really can't explain why is that (maybe Nodens can) but I guess that AIDA loads CPU cores only and come cache with FPU enabled) but there's something else that affects stability and isn't stressed by this test.

BTW my chip handles 16GB of mem 2400MHz CL10 with 1.1v VCCSA

[klepp0906]

I get that this is an oldie but since it popped up top in google and I keep seeing the same similar "omfg" in other current threads regarding prime and stability - I figure I'll start attacking the misinformation where I'm currently at.

Their is really only one way to say it. For you guys currently having a conniption over his or anyone's 3 hour prime stability test, you do realize their is a reason we as a community suggest 24 hours of prime testing for stability right? Even the newest CPUs with the highest overclocks take over 14 hours to finish ONE ROUND of prime95?

That being said, we will use my CPU as an indicator. I have a golden chip. First one I've ever been lucky on and it's on a overkill custom loop. Running my 4790k at 5.0 took 1.39v. This is until I reached hour 10-11 of prime (much further along than the sandy would be so I would imagine 14hr in maybe) but the point is at that point it took me another .03v up to 1.42 to not lock at that point.

I don't have the exact numbers next to me but at that point the most stressful FFTs are tested which is why many people are stable up to that point and get stuck.

That being said, with prime in it's current incarnation (28.5) it leaves me shaking my head. It was overkill all it's life, but has reached the point that it's no longer worth stressing with. (Assuming you want to reach an ideal overclock in which you aren't held back by something synthetic that tests levels of stress which are IMPOSSIBLE to reach outside said program)

Fwiw I recommend IBT for a quick n dirty and for those of you who want to make sure your cooling solution is up to the task. Hyper PI is ideal for dialing in memory timings, and for your all out stressor, using x264 v2 is ideal. It stresses using one of the most intense methods you will ever run into in real world use (encoding) which is why a lot of people simply test with handbrake (however x264 takes it a few steps further).

If you can pass that (without causing your cpu unrealistic heat) for a few hours and your PC ever crashes on you, let me know - I'll take you out for some Ruth's Chris. Until then....

[Nodens]

While I can't comment on the AIDA64 stress test itself due to lack of experience, The OP's screenshot.. is a bit hard to believe that it's from a stable system indeed. 5Ghz + 32GB of 2133 RAM @ 9 CAS stable @ 1.425v? Although we can't see VCSSA, if that is stable then it's the most impressive 3930k sample I've seen!
If it's not stable I won't have high hopes for the test as not-stable at that voltage should be visible in seconds on Prime/IBT/LinX. But then again it could be a matter of configuration as Zkal7 said. If I can afford some downtime on my RIVE rig, I'll see if I can make it to catch borderline unstable settings that are easily detectable with Prime and IBT.