Which stress test should I trust?

[trikat]

I am currently OCed to 3.9GHz on a temporary GA-EP45-UD3R v1.1 mobo w/ F11 bios (latest), but my issue is with possible instability. (Other comp components are in the signature.)
I can run Intelburntest v2 (20 passes using custom 2700 MB out of 3200 available ram) without any errors, but the Prime95 (p64v2511) Small FFTs torture test errors out fairly quickly.

To get an idea of the stress levels Intelburntestv2 pushes the CPU (dual core) temps to 57 C / 57 C while Prime95 pushes the temps to 52 C / 48 C).
Temperatures were measured using Real Temp 3.00.

I also re-downloaded the programs to make sure they were not corrupt.

Which test should I trust in this scenario?

[whovous]

Re: Which stress test should I trust?

If either test errors out, you have a risk of instability.

The gold standard, I think, is 24 hours of Prime95 without error.

[nate42nd]

Re: Which stress test should I trust?

If you REALLY want to go all the way, use IntelBurn Test. It stresses your CPU quite a bit worse (or better) the Prime95. However the couple of times I used it....it scared me.

Prime95 should be fine unless you in a competition or something.

[whovous]

Re: Which stress test should I trust?

IBT gets the core hotter, faster, which means you can find out very quickly if something is wrong. Nonetheless, if your machine cannot handle 20 minutes of Prime95, something is wrong. It passed the harder test but flunked the easy one. That does not strike me as a good sign.

[trikat]

Re: Which stress test should I trust?

O well, thank for the replies.
I'll put more emphasis on Prime95 (which has been THE test for years and years) to get absolute stability.
It does come close to IBT in terms of generating heat, but it seems to find it's own unique weak points in stability.
I just never had this issue before so it kind of puzzled me a bit.

[Chike]

Re: Which stress test should I trust?

At 3.9GHz you temperatures seems a bit low, unless you have water cooling or low ambiant temperature.
IBT results should be 26GFlops or above, if not it usually means more vcore.

[trikat]

Re: Which stress test should I trust?

At 3.9GHz you temperatures seems a bit low, unless you have water cooling or low ambiant temperature.
IBT results should be 26GFlops or above, if not it usually means more vcore.

My ambient temp might be lower than most. Thermalstat is set to 73 F, but in this small apartment the temps can go up rapidly if i have my door closed (such as 57 C / 54 C w/ Prime95). Right now it is opened.
I am using a HDT cooler, specifically the Xigmatek S1283 with Gelid extreme paste. I also did not lap my CPU.
I kind of think my temps are just right or maybe a bit higher (considering the ambient temps) as this E8400 E0 isn't one of the golden picks.
I have to give it more voltage to increase stability. It saddens me I did not get a chip that can run near 4 GHz at low/mid voltage, but i'll take when I can get.

[Chike]

Re: Which stress test should I trust?

IBT is a very good indication. One step increase in core may result in 6c+ and 2+GFlops difference.