Hey, I have my phenom 965 @ 3.8ghz on prime. It's currently running at 52 degrees and been running for 20 mins. I know the phenoms like to be cool, as they start to flake @55 degree n 62 is the absolute max temp. Is it safe for me to run prime over night? Or is it too risky
Is it safe to run prime overnight?
- Thread starter w1llyg
- Start date
Soldato
- Joined
- 24 Jul 2006
- Posts
- 8,876
- Location
- Hoddesdon, London, UK
Your logic is flawed. The point is not to see if it can do 8 hours, its to run a series of tests to check stability, a stable machine should be able to run prime forever until something burns out. Various test patterns etc are done over time, i personally use 24hrs for prime and have done since 99 ish as i've seen many 'teen hour' failures but rarely over 24hrs and the machine has always fallen over when it failed prime, doing something or the other.
Passing 8 hours dosen't mean it'll run anything for 8hrs then potentially fall over, it means it has done a lot of test patterns and tested different areas of the ram/cpu/mch etc.. which can't be done in the space of 90mins as someone put it. Your machine can fall over anytime and prime is a good indicator of its stability.
Do what you like, but i know the program and i know overclocking, hate to sound arrogant but a lot of people don't know what they're doing and why they're doing it, i've most assuredly overclocked more machines than most on this site, and while i'm no expert, i have a clue.
watching prime95 is like watching paint dry....if its stable after 20 minutes chances are its cool.....if it aint big deal you get a bsod which you fix with a tweak
Maybe for you, but i like to know i can trust a machine to do anything its capable of at stock, when overclocked.
I'm not gonna write anything else in this thread now, as its a pointless argument, simply put running prime for 20mins-1hr is pointless and most who actually know what they are on about would agree.
Last edited:
It's fine to let it run overnight. If your cooling is good you dont have to worry. LinX or intel burn test are other stress testers which work well. Besides, any modern cpu will shut down when it hits its limit. Ive had cpu's that were 8 hours+ prime stable, yet failed LinX/ibt in minutes.
Same here
Ive ran prime for 15 hours before, hit 69 at about 3pm then started going back down (was next to a locked window back then).
My friend came up, wasnt convinced (I video'd the entire thing) and ran 10 passes of IBT on max (4GB)....failed on the 3rd pass.
very true, same here... with my short eperience with overclocking.
Soldato
- Joined
- 24 Jul 2006
- Posts
- 8,876
- Location
- Hoddesdon, London, UK
Same here, however i find with Sandybridge, something about the latest builds of prime really rock their core and if it passes prime, IBT or Linx is almost a sure bet.
Associate
- Joined
- 20 Dec 2010
- Posts
- 160
how many times does a processor run at 100%, I'm thing Flight Simulator at best, what actually is the point in running Prime95 for 8 hours?
Logic? Why not 5 minutes? Or 1 hour?
I think the 8 hours thing is a marketing hype supposedly performed by companies selling overclocked PC's! I bet they don't even do it...
I think this is where the concept of Prime95 8 hours thing is originating from
Logic? Why not 5 minutes? Or 1 hour?
I think the 8 hours thing is a marketing hype supposedly performed by companies selling overclocked PC's! I bet they don't even do it...
I think this is where the concept of Prime95 8 hours thing is originating from
Soldato
- Joined
- 27 Mar 2004
- Posts
- 14,081
- Location
- Between Realities
I generally do a few hours when finding max overclock and tweaking ram. once its passing these tests, I then do an overnight run. It also helps with the burn in period. Its nothing to do with hype, Thermal paste, Motherboard components will all get hotter when ran for a prolonged period. You guys who dont run an 8 hour test. i bet half you guys would fail a long run. Which means, You're not stable
Soldato
- Joined
- 24 Jul 2006
- Posts
- 8,876
- Location
- Hoddesdon, London, UK
The 8 hours prime concept is relatively new, standard was 24 hours. Prime has been around a lot longer than you've been overclocking it would seem. You also seem to think that a cpu has to be at 100% to show instability or fall over from a poor clock.
Sorry, but this argument is weak, the reason you run for so many hours is to test out various bits of your cpu/chipset/ram and how they react to being tested, its not the same thing over and over and it may not touch upon an unstable area of the cpu etc.. for a few hours, if you don't understand a concept, how can you criticise it?
I'm almost 100% sure the people spouting on about prime/ibt etc.. being pointless are relative newcomers with little experience in overclocking. Theres no such thing as a 10min overclock so many people seem to want (microwave culture?).
My regime:
24hrs prime
100 runs of Linx/ibt - great tool and good addition
various 3d tests
Use the pc hard, game etc.
This has worked for me 99.9% of the time with troublefree rigs after.
No 'tweaking' bsods afterwards for me, no random data loss or lockups.
I'm guessing some of these rigs will fall over at some point, for 'no reason' cause it was 'stable in Crysis'.
Anyway, i should keep my mouth shut and carry on, arguing something pointless like this always drives my blood pressure up.
Last edited:
I agree. The machine is either stable and usable 24/7 or it's a potential BSOD. If you won't run your PC overnight for fear of damaging it then back off the overclock. If you can't afford to damage the machine through overclocking, don't overclock.
Everything you do in overclocking is at your own risk. They don't break often, but when they do, it's potentially YOUR fault.
I'm 100% with Justintime and WJA96 on this.
It's not that many years ago when a person would be laughed off a forum for claiming a stable clock after just a couple of hours testing. The only thing that has changed since then is that it is easier to overclock nowdays. But the basic premise of testing the stability of that overclock remains the same. The entire system needs to be tested, not just the CPU. That includes the memory subsystem. If not, then corruption and data loss will occur at some point. To this day, the best way of doing this is using Prime95 set to "blend".........................it hammers the cpu, memory and memory subsystem.
It takes at least 8hrs for Blend to complete a testing cycle, so the least you could get away with, but 24hrs makes it cut and dried.
To all those peeps that keep saying "Lynx for and hour and a couple of hours of Prime is enough"....................................the only person you are fooling is yourself.
Sometimes it used to take weeks of tweaking and testing to get a stable clock. Nowdays, the clocking part is quick and easy..........................so what is it about peeps that they just can't be bothered to do the testing properly to validate that clock ? Maybe it is that "Microwave" culture that Justintime mentioned...................................and we all know that most microwave food is rubbish.
It's not that many years ago when a person would be laughed off a forum for claiming a stable clock after just a couple of hours testing. The only thing that has changed since then is that it is easier to overclock nowdays. But the basic premise of testing the stability of that overclock remains the same. The entire system needs to be tested, not just the CPU. That includes the memory subsystem. If not, then corruption and data loss will occur at some point. To this day, the best way of doing this is using Prime95 set to "blend".........................it hammers the cpu, memory and memory subsystem.
It takes at least 8hrs for Blend to complete a testing cycle, so the least you could get away with, but 24hrs makes it cut and dried.
To all those peeps that keep saying "Lynx for and hour and a couple of hours of Prime is enough"....................................the only person you are fooling is yourself.
Sometimes it used to take weeks of tweaking and testing to get a stable clock. Nowdays, the clocking part is quick and easy..........................so what is it about peeps that they just can't be bothered to do the testing properly to validate that clock ? Maybe it is that "Microwave" culture that Justintime mentioned...................................and we all know that most microwave food is rubbish.
