burn in a HT processor

[#18]

I need some advices here. Usually Prime95 can keep any CPU running at 100% full load. With Hyper-threading technology, it shows as two processors on MBM5, and none of them ever close to be 100% load, the combine number is always around 120%. Does it matter? I try to burn in this system, do I really need both show 100% to get burn in effect? If so, then turn off HT should work right? If it does not matter, then I will keep it this way.

[Xymurgy]

Each processor should show up as being 50% utilized. What I do is copy the Prime95 directory, paste and rename it, then run two instances of it (the .exe file in both directories).

[NookieN]

You don't really need both virtual processors to show 100% load. One instance of Prime on an HT-enabled CPU stresses the chip just as well as one instance on a non-HT CPU. You might get slightly more load (and slightly higher temps) with two instances, but don't expect there to be a huge difference.

[stan03]

yea i actually tried this with seti and folding, the temps only go up my around 1 degree for me comparing one instance and two instances.

[#18]

thank you for the information.

[oc_newbee]

Question for NookieN,
I had two instance prime95 running all weekend long and one fail at 28 hours and the other kept going until I stop it at 35 hours. Do you consider that as prime95 stable?

[NookieN]

Technically you don't want Prime95 to ever fail. But if you're running two instances and it takes 28 hours for one to fail, I would consider that very stable. I've had a second instance fail on me after only 4 hours, and while it bothered me a little, a single instance would run indefinitely.

As long as you can run a single Prime95 in torture tests without error, that's good. If you can get two instances to run without error that's great, but probably not absolutely necessary.

[Lancelot]

I test with both one instance of Prime95 and one instance of SETI running...

[markodude]

I test with both one instance of Prime95 and one instance of SETI running...

Me too, and I see about an extra 2-3c with them both running compared to a single instance of Prime95.....I alsouse good old memtest86 on the RAM, its annoying how one bad chip on a stick can muck up your overclocks....

[alcky]

I don't know what the big deal is with prime testing and all that... HOWEVER

I do know a way to stress 'both' procs, a friend was doing a maya render and needed to get it done fast so I offered to do an animation for him. Since maya is multi threaded I set it up to use ALL available processing power and let it run. The render took about 20 hours and all that time I checked the status every so often, utililization would only dip when it was changing frames to render any other time it was pegged at 100% for both utilization meters.