Prime 95 stress(torture)test..how long

[darrensimmons]

I have been doind some overclocking of late was wondering how long is the ideal time to run the stress test off prime 95?Don't want to overstress the cpu.

[Crash]

Originally Posted by darrensimmons

I have been doind some overclocking of late was wondering how long is the ideal time to run the stress test off prime 95?Don't want to overstress the cpu.

12 hours minimum,without any error...24 hours recommended

[rflair]

Like Crash said, 24 hours recommended, I've let it run over 96 hours (3 days) when I went away for the weekend a few times.

[Mufflore]

I do it for as long as I am likely to ever play games for.
If you use anything CPU intensive for very long periods, 24hrs should do.

If you do a short test, do it at the warmest time of day.

[Matt26LFC]

Originally Posted by Mufflore

I do it for as long as I am likely to ever play games for.
If you use anything CPU intensive for very long periods, 24hrs should do.

If you do a short test, do it at the warmest time of day.

Yeah I don't really feel the need for all night long stress tests anymore lol

If its stable when I'm gaming, doing everyday things on my rig and doesn't BSOD then its stable enough for me.

If I depend on my rig for my lively hood then thats a differen't matter, however I don't so I don't care

[xaudiox]

Originally Posted by Matt26LFC

Yeah I don't really feel the need for all night long stress tests anymore lol

If its stable when I'm gaming, doing everyday things on my rig and doesn't BSOD then its stable enough for me.

Ive had prime95 run for 24 hours without fail but now I always use Intel Burn Test, 10 loops on extreme preset takes about 20/30 mins and stresses the CPU way more than prime95 ever could.

[BuildeR2]

Originally Posted by xaudiox

Ive had prime95 run for 24 hours without fail but now I always use Intel Burn Test, 10 loops on extreme preset takes about 20/30 mins and stresses the CPU way more than prime95 ever could.

Totally agree here. Exactly what I was going to say. My current Q6600 (old, I know) could make it 48 hours on Prime95 @3.6Ghz with no errors, but would die on IBT. It also will get your temps a good bit higher than Prime95 can, so watch that carefully.

For example:
Max Prime95 temp @3.6Ghz - 59C
Max IBT temp @3.5Ghz - 69C

[Veteran]

On my cpu IBT is dead easy to pass with a low amount of volts for stability success,Linx and Occt are the hardest with prime being somewhere in the middle.

I would say Prime minimum 4 hrs.

[Pill Monster]

5mins? lol

[BLEH!]

I'm with Veteran on this. 4 hours seems to weed out unstabled OCs for me. I use IBT as an initial test of stability, 5 runs on very high, which will crash if something simple is unstable. P95 is more of a nuance thing.

[xaudiox]

My only problem with prime95 is you will never get conformation that your CPU is stable, you can stop after 4hrs but how do you know that if after 5 hrs you will still be stable ? just seems like a time waster to me.
Totally understand people using prime95 as it is more of an endurance test for your CPU but for me IBT on maximum preset (10 pass) is just as good as any stability test.
+ you at least get conformation that your CPU is stable and mine truly has been, always.

[Veteran]

Originally Posted by xaudiox

My only problem with prime95 is you will never get conformation that your CPU is stable, you can stop after 4hrs but how do you know that if after 5 hrs you will still be stable ? just seems like a time waster to me.
Totally understand people using prime95 as it is more of an endurance test for your CPU but for me IBT on maximum preset (10 pass) is just as good as any stability test.
+ you at least get conformation that your CPU is stable and mine truly has been, always.

I use all stability tests not just one,this way you know your machine is truely stable,i cant remember the last time i had a bsod tbh.

[IcE]

Originally Posted by xaudiox

My only problem with prime95 is you will never get conformation that your CPU is stable, you can stop after 4hrs but how do you know that if after 5 hrs you will still be stable ? just seems like a time waster to me.
Totally understand people using prime95 as it is more of an endurance test for your CPU but for me IBT on maximum preset (10 pass) is just as good as any stability test.
+ you at least get conformation that your CPU is stable and mine truly has been, always.

IBT can pass on extreme mode all day on this particular CPU and fail on Prime spectacularly in 5 minutes or less.

However, I don't believe that you actually need more than 8 hours of prime, given that it's extremely unlikely the CPU isn't stable if it can pass 8 hours of the same repeating tests. I'm pretty sure the people that have failed at 24 hours failed for other reasons like a power interruption or some such glitch.

[xaudiox]

Originally Posted by ice445

IBT can pass on extreme mode all day on this particular CPU and fail on Prime spectacularly in 5 minutes or less.

However, I don't believe that you actually need more than 8 hours of prime, given that it's extremely unlikely the CPU isn't stable if it can pass 8 hours of the same repeating tests. I'm pretty sure the people that have failed at 24 hours failed for other reasons like a power interruption or some such glitch.

Every computer is different, in my case ive never had errors with prime95, only IBT can find my CPU's sweet spot, so thats why i stick with it.
Everyone has their own reason why they use a particular stress test, theres nothing wrong with that, use whatever works for you.

[IcE]

Originally Posted by xaudiox

Every computer is different, in my case ive never had errors with prime95, only IBT can find my CPU's sweet spot, so thats why i stick with it.
Everyone has their own reason why they use a particular stress test, theres nothing wrong with that, use whatever works for you.

Yes, I agree with that. Just pointing out that it's odd how one test can straight up not work for some but not others.

[clawhamer]

Yeah, I agree, everyone has their own method for testing stability and what they feel is suitable.

I use IBT – Maximum – 10 pass for the quick and dirty temp\stability test and P95 – Blend – 24h. P95 makes it easy to see where the instability is depending on what test the error occurred in, then you can fine tune where needed.

IMHO, I wouldn’t rely on just one.

[BlackZero]

Different people have different weaknesses in their system, for some it might be the cpu, for others the memory controller/chips and for some plain old heat. This is why testing all these potential weak points is recommended and also why a range of stress testing software as well as simple day to day usage is the most effective way to determine stability. Though it should be noted that stability is relative and every system can and generally does crash at some point, especially when exposed to different software/conditions.

Out of the stress testing utilities mentioned, prime95 blend is the only one that will stress most system components and do so over extended periods while being relatively close to every day usage scenarios. However newer technology brings changes that sometimes make it difficult for many stress testing utilities to remain as effective, this is especially true for sandy bridge based systems where newer power saving technologies and the way they have been Incorporated into the chip design have made it almost impossible to test for weaknesses or instability using traditional stress testing utilities and methods. This is evident from the well known idle bsod issue on sandy bridge based systems caused by power saving techniques, disabling them can cause even greater system instability.

In the instance mentioned above, instability takes place in extremely light load situations and as most stress testing utilities are not designed with this in mind it's extremely difficult to test for instability other than through general usage over extended periods of time.

[hallryu]

Overnight (8 hours) has always been my yardstick with P95.

[---TK---]

minimum 8 hours p95 blend max memory

[Anarion]

I do not usually run Prime really long, about 30 min or so to see that there is not going to be temperature issues. The real stability test is real life usage. If I can game or encode video as long as I want without stability issues, it's stable. Relying in one program only for stability is not wise since it is not uncommon that system can do Prime for hours but then crashes almost instantly, say, in BF3.

[Darren Hodgson]

Originally Posted by ice445

IBT can pass on extreme mode all day on this particular CPU and fail on Prime spectacularly in 5 minutes or less.

That's interesting.

I just created a thread about my 3.8 GHz overclock and my PC seems to have been stable except when I ran Prime95 and it froze my machine after 20-25 minutes (the core temperatures were 69-74 C at the time and seem to have maxed out at those figures). I ran Heaven 3.0 on Extreme, 8xAA/16xAF, 1920x1200 for 40 minutes with no issues as well as 3DMark11 and a few games.

I've never really understood the need to run Prime95 given that it stresses the hardware in ways that normal use with applications and games don't. You could feasibly never see a single crash except when using Prime95. And because it crashes with Prime95 you then raise the voltages and create more heat, perhaps, unnecessarily given that your PC would never be pushed that hard except during stress testing!!!

I've decided that I'm going to just use my PC at the current settings as I would normally. Acronis True Image 2011 is the only application I use that maxes out all 8 threads at 100% and at 3.6 GHz that was 100% stable. If I have any crashes then I'll look into raising voltages and then continue to use it normally until it crashes again. I only play games and stuff on my PC so crashes would only be a minor inconvenience (not that I get many anyway). This seems a more realistic way of testing my own PC than stressing it heavily non-stop for 12 or 24 hours. That could possibly damage it, particularly as my PC is never used heavily for that length of time and is always turned off at night.

[Mufflore]

Originally Posted by Darren Hodgson

That's interesting.

I just created a thread about my 3.8 GHz overclock and my PC seems to have been stable except when I ran Prime95 and it froze my machine after 20-25 minutes (the core temperatures were 69-74 C at the time and seem to have maxed out at those figures). I ran Heaven 3.0 on Extreme, 8xAA/16xAF, 1920x1200 for 40 minutes with no issues as well as 3DMark11 and a few games.

I've never really understood the need to run Prime95 given that it stresses the hardware in ways that normal use with applications and games don't. You could feasibly never see a single crash except when using Prime95. And because it crashes with Prime95 you then raise the voltages and create more heat, perhaps, unnecessarily given that your PC would never be pushed that hard except during stress testing!!!

I've decided that I'm going to just use my PC at the current settings as I would normally. Acronis True Image 2011 is the only application I use that maxes out all 8 threads at 100% and at 3.6 GHz that was 100% stable. If I have any crashes then I'll look into raising voltages and then continue to use it normally until it crashes again. I only play games and stuff on my PC so crashes would only be a minor inconvenience (not that I get many anyway). This seems a more realistic way of testing my own PC than stressing it heavily non-stop for 12 or 24 hours. That could possibly damage it, particularly as my PC is never used heavily for that length of time and is always turned off at night.

Heaven hardly stresses the CPU.
3DMark isnt long enough to give the cpu a workout, its not a constant high load either.

The reason I use it is to guarantee stability for anything I do.
I can get just that bit better overclock if I dont fully stability test, but leaving my machine on 24/7, I get a BSOD or 2 a week out of the blue.
P95 Blend test showed me that I truly need to reduce the overclock to get properly stable.
Prior to that I was using P95 small FFTs and while it gets hotter, it allows an unstable overclock to pass.

[Veteran]

Originally Posted by Darren Hodgson

That's interesting.

I just created a thread about my 3.8 GHz overclock and my PC seems to have been stable except when I ran Prime95 and it froze my machine after 20-25 minutes (the core temperatures were 69-74 C at the time and seem to have maxed out at those figures). I ran Heaven 3.0 on Extreme, 8xAA/16xAF, 1920x1200 for 40 minutes with no issues as well as 3DMark11 and a few games.

I've never really understood the need to run Prime95 given that it stresses the hardware in ways that normal use with applications and games don't. You could feasibly never see a single crash except when using Prime95. And because it crashes with Prime95 you then raise the voltages and create more heat, perhaps, unnecessarily given that your PC would never be pushed that hard except during stress testing!!!

I've decided that I'm going to just use my PC at the current settings as I would normally. Acronis True Image 2011 is the only application I use that maxes out all 8 threads at 100% and at 3.6 GHz that was 100% stable. If I have any crashes then I'll look into raising voltages and then continue to use it normally until it crashes again. I only play games and stuff on my PC so crashes would only be a minor inconvenience (not that I get many anyway). This seems a more realistic way of testing my own PC than stressing it heavily non-stop for 12 or 24 hours. That could possibly damage it, particularly as my PC is never used heavily for that length of time and is always turned off at night.

I forgot to say that running 12gb of ram on an x58 can cause instability with regards to a high oc threshold,(syncronisation between ram,cpu and nb)which you seem to be at considering its a c0 chip.

You could try running your oc with 6gb but that kinda defeats the object,just thought you might like to know.....x58 struggles with high amounts of ram with a high oc on the cpu.

[deltatux]

12 hours for Prime95 and 10 rounds at Maximum for Intel Burn Test.

Prime95 can't test stability unless you do it for hours.

deltatux

[bokah]

5-30mins should be enough

more is just boiling it up unnecessery