Ultra stability test

[Chilly]

Just tested it on my SR2 system, it only stress 8 out of 24 threads

[makizeka]

Let me explain:

My stress test focuses only on the stress of the CPU and not of the memory ... And for this reason it has low temperatures compared to other programs. The objective is to implement a testing program of stablity very fast which does not produce much heat.

Regarding the experiments:

Switching the CPU in a state of semi-instability, among the tests carried out, only my test program failed while other programs passed.

Ex1: cpu voltage rock solid with 1.25:
I decrease the vcore a little bit (1.24375).Then I test with OCCT IBT ... ... and I pass the test, while my program detects instability.

Ex2: cpu voltage rock solid with 1.25:
I decrease the vcore a little bit (same as before 1.24375), I try with IBT OCCT ... ... they give me a failure after a time much bigger than the time of my program.

@JD
I think that if you lower the vcore a little bit or you increase a little bit the frequency...the tests will give always "fail"
Could you confirm these results?

I hope I explained myself :-):-)

@Chilly
My test was created for 8 threads....for istance like the platform intel i7.

And thank you for your willingness

[JD]

Originally Posted by makizeka

@JD
I think that if you lower the vcore a little bit or you increase a little bit the frequency...the tests will give always "fail"
Could you confirm these results?

I will give it a try when I have some time to spare. I surely can't increase frequency anymore though as it results in my computer not being able to POST at all.

[Arinoth]

How long is this test actually suppose to take?

[Chilly]

Originally Posted by makizeka

@Chilly
My test was created for 8 threads....for istance like the platform intel i7.

And thank you for your willingness

My Xeon's in my SR2 are the one and the same platform as the i7 LGA1366 your testing on. My specific CPU's are Westmere(i7 970, i7 980x, i7 990x, Xeon 5600 series) which are a 32nm shrink of Nehalem(i7 920 and etc, Xeon 5500 series) with minor improvements(AES). Only real difference between the two platforms, 2nd QPI is enabled and price. The silicon is the same.

Just add either auto detection of threads or a field we can enter/select how many threads we want the program to spawn.

[Coffaholic]

Originally Posted by Arinoth

How long is this test actually suppose to take?

No idea. Just started running it on my s775 quad and it seems to be working. More information would be nice on the screen as mentioned above. ie. how long the test has been running, or how long until the test is done if its not an infinite thing.

[Arinoth]

Uh coffaholic, the test is designed for the i7 LGA1366 platform, not any other. I wouldn't be surprised if it gives you a false positive, crashes or just doesn't work.

[Coffaholic]

Guess i misunderstood the poster. Oh well, i know im stable anyways :)

[makizeka]

Originally Posted by Arinoth

How long is this test actually suppose to take?

The test takes little time ... in my i7 950 max 5 minutes if the system is stable...but less if it is unstable ...

[makizeka]

Developed new release!

Download:
http://www.cad.polito.it/research/Ev...es/M&G_v2a.zip

Changes:
- Optimized code and fixed some bugs.
- Addition of a progress bar.
- Running time.

On this page you will find the next versions of the tool:
Overclocking

Comments and criticisms are welcome!