Running prime95 for stress test

[Blackninja543]

Are there any tutorials out there on how to run prime for a stress test. I know its a good idea to run prime after OCing even 10 mhz on but how many hours should i do. I was thinking about running it for around 2 hours for a general stress test between OCing checks, then once i have my system the way i like it, to run it for 24 hours as a stress test for longevity.

[tkrow21]

24 hours is great, though I've heard 8 hours is enough.

[Blackninja543]

What kind of settings should i use in order to thoroughly test both the cores on the CPU along with the ram and entire FSB?

[Verran]

Honestly, I would not recommend Prime for the "in between" testing.

 

What I usually do is start with short tests and scale upwards as I get closer to my final speed. I start by just testing with SuperPi 1M and 4M. Those are real quick, but they're not very accurate. So I get those, and then when I get a little closer, I start using the OCCT 1 hour test. I use that until I get VERY close, and then once I think I've got my final settings, I use Prime for 24 hours.

 

Running long tests for every minor change is just a major pain, so I don't I only do it when I'm almost done.

[ccokeman]

I only move to Prime 95 when I am about sure of my permanent overclock. A boot into windows and a quick pass of Cinebench means I can move upwards. What you should do is run the blend test as it tests not only the CPU but the memory and subsystems.

[HD4850]

Here's how I test for stability:

 

1: Extract a few large 1-2gb files from Winrar. If your northbridge is unstable, Winrar will return a "CRC Failed" error.

 

2: Run OCCT Large Data Set for 10hrs.

 

3: Run Prime95 Small FFT for 12hrs.

[Blackninja543]

Ran OCCT 1h and it only got through the first 20 then stoped and printed the info on the test. No errors where reported on any of the tests however it obviouly didnt run the full hour anyone have any ideas?

[Verran]

Ran OCCT 1h and it only got through the first 20 then stoped and printed the info on the test. No errors where reported on any of the tests however it obviouly didnt run the full hour anyone have any ideas?

It's either stable or it's not. There's nothing tricky to it. If it didn't get through the whole test, your OC is unstable.

[Blackninja543]

Any ideas as to what it was, I never changed any of the voltages so it was probably that. All i ended up doing was bumping the bus up from 333 to 380. I just droped it back down to stock settings for now, but ill probably end up working on this at a later time. If you want I can post my stock settings from CPUz.

[Verran]

Any ideas as to what it was, I never changed any of the voltages so it was probably that. All i ended up doing was bumping the bus up from 333 to 380. I just droped it back down to stock settings for now, but ill probably end up working on this at a later time. If you want I can post my stock settings from CPUz.

What do you mean what it was? Your clock was simply too high.

 

This is why you test often. If you test at 300FSB and it's fine and then you test again at 305FSB and it's not, well now you know the problem!

[tkrow21]

With a 380FSB on an E8400 you might have to bump up the voltages a bit. It varies from chip to chip; most people will get there without a voltage increase, but I, for example, get there only if I bump up the voltage just a bit.

[Blackninja543]

Well even bumping up the voltage a decent bit i couldnt get it to fully pass OCCT so im testing it at stock settings and see if it still happens.