Stress testing base clock

[BababooeyHTJ]

I am playing around with my new Lynfield. I'm just stress testing the base clock and ram right now. I'll play around with the actual core clocks tomorrow once I'm sure that everything else is stable.

I am currently have all four dimms occupied on my board and would like to keep the VTT as low as possible. I'm using 1.24v vtt, 200 bclk, and 800mhz cas 9 on the memory and the cpu at a low 2.6ghz. I ran about an hour of Prime large fft with run fft in place unchecked and 7200MB of ram set to be used. That ran with no issue. I'm also about 10 passes in with Linx.

What would you guys use to test stability? I would like to run one test over night. Should it be prime blend? Some custom blend? Linpack?

[trickson]

Originally Posted by BababooeyHTJ

I am playing around with my new Lynfield. I'm just stress testing the base clock and ram right now. I'll play around with the actual core clocks tomorrow once I'm sure that everything else is stable.

I am currently have all four dimms occupied on my board and would like to keep the VTT as low as possible. I'm using 1.24v vtt, 200 bclk, and 800mhz cas 9 on the memory and the cpu at a low 2.6ghz. I ran about an hour of Prime large fft with run fft in place unchecked and 7200MB of ram set to be used. That ran with no issue. I'm also about 10 passes in with Linx.

What would you guys use to test stability? I would like to run one test over night. Should it be prime blend? Some custom blend? Linpack?

Intel burn test it and be done . All this linpack and over night runs well there all good and fine but don't you want to have some fun with it ? I mean you could do nothing but test it for ever and still have some say it is not good enough . I would just Intel burn in test it and be done . Now if you are looking for absolute stability then Linpack it OCCT for 8 hours .

[Undead46]

Prime95 is good, but I hear Intel Burn Test is more efficient and a lot faster.
I'd say 50 Intel Burn Test passes and you should be 99% stable.

[BababooeyHTJ]

I want to be absolutly sure that the ICH is stable with my current bclk, low vtt, and four occupied dimms. Running all four dimms at their to rated speed makes it a little more difficult to detect errors. I've run into problems with this on a couple of 775 boards. Linpack would run fine but I would run into the occasional bsod and would fail the custom prime run that I mention earlier. ICH/MCH stability issues=data corruption

Once I know that the bclk and memory is 100% stable all that I have to do is play with the multi and vcc. It makes things easy.

That custom prime run that I mentioned heats up the memory much more than Linpack. I've also had Linpack run fine with max memory and then fail prime small ffts in under an hour. Linpack is a great test but isn't the end all be all.