Ran Prime95 for 5hrs - no problem

suttoni

Distinguished
Jun 22, 2011
14
0
18,510
I don't plan on doing anything real heavy on this computer. I plan on playing Skyrim when it comes out, my question is this: is running prime95 for 5 hrs enough? I know it's a largely debated issue. I've been keeping track of my temps, and from idle they've gone up about 11-15 C on average (from about 39C to 55C). I'm currently using an Intel Core i5 2500K at 3.30Ghz with a Noctua NH-U9B SE2 HSF. Thanks!
 
Solution
Are you running Prime 95 on your stock speeds?? Anyways, I would run it for a good 10 hrs. I've seen errors pop up several times before between 5-10 hrs
Yes, 5 hours is more than enough! And the temps look fine; I would have expected 60 degs C or thereabouts.

My opinion is, unless you are running these tests for a research lab, 30 minutes of stress testing is sufficient. I overclocked my i7-920 to 3.82 GHz and ran prime95 for 10 minutes. My system is running stable for well over a year. Games played include COD 4, Black Ops, FS-X.
 

guskline

Distinguished
Aug 25, 2006
431
2
18,795



Used to be 24 hrs, but I run them overnight. I just ran a 2500K @ 4400 from 9 PM last night till 7 this morning - 10 hrs without a problem. Highest temps 57c AWESOME.
 
You shouldn't even need to run Prime95 if you're going to run stock speeds for the CPU. It's meant to be a torture test for an overclocked processor to make sure it's stable.

If you're overclocking your CPU and using Prime95 for stability testing, don't follow Ubrales' example. Running a stability test for only 10 minutes is stupid and asking for trouble. For most users, a few hours to overnight is good enough. I'm not included in that group -- I do far more stability testing on my own machines.
 

In spite of doing "stupid and asking for trouble..." things, I have an excellent track record in US cutting edge technology and corporate management, and will continue to follow practices that have led me to success in the US! This is not subject to debate!
 
I do stupid things and ask for trouble too -- like running my 2500K at 4.8GHz using 1.504v. But I don't recommend doing those things to newbies or even most veteran users, and I usually put a disclaimer to that effect in the post.

I certainly wouldn't recommend 10 minutes or even 30 minutes of Prime95 stress testing to any user. Thirty minutes is barely enough time to get to stable full load temps, and doesn't give a chance to discover the kinds of errors that don't happen right away.

It's good that you have an excellent track record, but someone else using your methods may not have the same luck. Doing only 10-30 minutes of stress testing is taking too much of a chance with some expensive equipment. I won't recommend that, and I won't stand for someone else to recommend things like that without being challenged.
 

I agree to the content on this post, but not to your previous post which is a direct personal attack on me. No quibbling on words!

Multiple environment overstress testing is great for manufacturers in order to design-in a level of performance that the product will not see in general use. My comments are based on the OP's statement that he is not planning on doing anything heavy on his computer. Under such conditions, no stress testing is necessary.

I don't know where you are coming from, but in this country (US), we do not go around calling others "stupid". Such words are undignified.
 

browsingtheworld

Distinguished
Jun 28, 2011
612
0
19,010
You should test or burn in your overclock and hardware when you first get it so you can tell newegg/whoever it's DOA and just replace it if you do burn it out. Prime95 is a low-level stress test and not really adequate to test the stability of systems.

If you really want to test how stable your system/overclock is run OCCT linpack overnight. Linkpack destroys what most people consider "stable" overclocks that are stable for hours in Prime within 20 minutes.
 
Unfortunately, while Prime95 will sometimes work fine on all 4 cores of a heavily overlocked 2500k system, the same system might then crash playing games, making Prime95 not much more than a core checker/heat sink-thermal load testing program, vice it being a useful overall stability tool...

If running the 2500k at stock 3.3 GHz, there should be little cause for any stabilty or thermal concerns....
 

Uther39

Distinguished
@ Leaps-from-Shadows, Stupid is indeed a strong and undignified word and one you may wish to retract, there are other words you could of used to inhibit the feeling in Ubrales of a personal attack.
Remember chaps, opinions are like assholes, we all have one!
 

We settled our differences in private. All is well now. Thank you for your comment.
 
Prime95 is the accepted standard for testing CPU stability. Linpack is also used, but far less common. I use Prime95 and IntelBurnTest (Linpack with a GUI) for my CPU stability testing.

I do 50 runs on Maximum settings in IBT, which takes about 7.5 hours on my overclock. My 4.8GHz 2500K is perfectly stable and passes with flying colors. I also ran 24 hour of Small FFTs, 24 hours of In-place large FFTs, and 24 hours of Blend in Prime95.
 


lol @ IBT

I have a story about IBT. I have an aggressive overclock on my q6600 (50%). I figured I'd give IBT a whirl. The computer responded faster than 4 SSDs on steriods... blue screen to black screen to no power at all...lol... Wouldn't restart. I reset the cmos. It still wouldn't reset. I removed the graphics card to get to the battery and removed it. It still wouldn't start. At this point I was like "oh hell" so I overdid it and removed the battery again and threw another graphics card in there. Finally it restarts but only because I left the battery out much longer the 2nd time around. I nearly stroked out over all this hehehe.