Sign in with
Sign up | Sign in
Your question

Is it absolutely necessary to pass prime95?

Last response: in Overclocking
Share
April 15, 2011 4:07:38 AM

I have been noticing how insanely hard it is to run prime95 long-term with no crashes. It is practically impossible for me unless I increase the vcore into an uncomfortable range, and my temperatures rise above what I would want them to be at. However, it is so easy to get stable under things like 3dmark and gaming. My question is, what is the benefit of being stable under prime? I never crash in the OS, only when I run prime95, it's just insane, I have never run anything that has stressed all of my cores to 100% and caused my system to crash.

I have a phenom ii x4 965 BE and it seems impossible to get it any higher than like 3.85 ghz and remain stable under prime. I have tried higher frequencies, and I never crash in the os nor in games, but I get destroyed in prime. Do I really need to be concerned if I'm not stable in prime?
a b K Overclocking
April 15, 2011 5:36:50 AM

No, you don't need to be stable in Prime95.

However, if you aren't stable when overclocking, then your CPU and memory are throwing off small errors at random. Most of these are caught and corrected automatically. Some are not. Enough of them happen and you get data corruption.

Will it happen to you? Probably not ... but you have to decide whether you are really willing to take the chance just to have a bigger overclocking E-peen.
m
0
l
a b À AMD
a c 286 K Overclocking
April 15, 2011 2:25:18 PM

Leaps-from-Shadows said:
No, you don't need to be stable in Prime95.

However, if you aren't stable when overclocking, then your CPU and memory are throwing off small errors at random. Most of these are caught and corrected automatically. Some are not. Enough of them happen and you get data corruption.

Will it happen to you? Probably not ... but you have to decide whether you are really willing to take the chance just to have a bigger overclocking E-peen.


Not agree.
m
0
l
Related resources
a b K Overclocking
April 15, 2011 5:43:39 PM

saint19 said:
Not agree.

I don't agree with it either. I would never run a rig that wasn't Prime95 stable.

But not everyone is like us.

The OP doesn't need to be Prime95 stable. I know quite a few people on forums who make the dumb decision to run everyday overclocks that fail Prime95. They risk a lot by doing so, but it's their system. Besides ... if the OP is asking the question, they have likely already made the decision to ignore Prime95 anyway. No use fighting them on it.
m
0
l
a c 100 K Overclocking
April 15, 2011 7:39:00 PM

Yeah... need to be? Well, technically no. You won't reck your hardware by being unstable. But there's a good chance of data corruption or a BSOD at a very bad time which could easily ruin a day - or more!

If getting stable requires an uncomfortable amount of voltage and heat then you should either learn to deal with it, or don't go that high. That's basically all there is to it for overclockers. Push it and deal with the consequences or play it safe.
m
0
l
April 15, 2011 7:57:59 PM

How much time it takes to fail in prime95? 2 min? 5 min? because if is that the case then your cpu is going to be unstable for everything you do.
m
0
l
a b K Overclocking
April 15, 2011 8:23:07 PM

The problem with memory error, and why I would say yes, you need to be stable. While the error shows up quicker in Prime95; If heat induced, not a big problem if it does not show up in normal use, If if it is not temp related it could be a big issue.

Most of the time it will cause a data error, or an error in an instruction and not a biggy. However ther is a slimchance that it could be a pointer on where the data is to be stored, or where that execution is to take place. On rare occacions that could even be to the Bios - Walla there goes the bios, no bootee. On of the reasons memory should be rock solid when doing a bios update also as in this case it is affecting the bios 100%.

My self - Take your OC down to where it is stable in Prime95.
m
0
l
April 15, 2011 8:36:47 PM

What temps and voltage are you not comfortable with? I say this because I have the same CPU and I can change the multiplier so its clocked at 3.8 and not even raise the voltage and thats with a stock cooler?
m
0
l
April 15, 2011 8:39:31 PM

RetiredChief said:
The problem with memory error, and why I would say yes, you need to be stable. While the error shows up quicker in Prime95; If heat induced, not a big problem if it does not show up in normal use, If if it is not temp related it could be a big issue.

Most of the time it will cause a data error, or an error in an instruction and not a biggy. However ther is a slimchance that it could be a pointer on where the data is to be stored, or where that execution is to take place. On rare occacions that could even be to the Bios - Walla there goes the bios, no bootee. On of the reasons memory should be rock solid when doing a bios update also as in this case it is affecting the bios 100%.

My self - Take your OC down to where it is stable in Prime95.

So would you say the fact that I cant run Primer 95 without going over 65c for more than 3min but never ever see those temps in real life usage is ok? Im at 4.0 ATM with voltage just under 1.45 and in the 4 mins that I can run Prime 95 I get 0 errors but its just to hot. Im games and benches I have yet to see BSOD or anything else crazy and temps wile playing games like crysis and Metro 2033 never go past 57c.
m
0
l
a b K Overclocking
April 15, 2011 8:53:37 PM

Basically, if you don't need to have your PC on for 6 hrs than you don't need to be able to go 6hrs on Prime 95. Prime 95 is pretty much a worst case kind of thing for the CPU. If your CPU is fine for two hours, and that's about the time you spend gaming, then it's stable enough for you. If you need it to be stable 24/7 though that's another matter.
m
0
l
a b K Overclocking
April 15, 2011 8:55:05 PM

I should also add it's important to keep reasonable temps under prime 95. Increasing the vcore will increase your CPU temps and shorten your CPUs life.
m
0
l
April 15, 2011 9:16:46 PM

megamanx00 said:
I should also add it's important to keep reasonable temps under prime 95. Increasing the vcore will increase your CPU temps and shorten your CPUs life.

At the same time there are very very few things that you will do in real usage that are going to peg a 4 core CPU at 100% for any extended period of time and thats what can add to high temps in Prime 95. So as far as heat goes not errors wouldn't monitoring your temps during heavy usage be more accurate? I can say a car gets crap MPG wile its pegged at redline but thats not the proper test for MPG. So temp wise what do you think about that way of thinking?
m
0
l
a c 100 K Overclocking
April 15, 2011 9:26:14 PM

cburke82 said:
At the same time there are very very few things that you will do in real usage that are going to peg a 4 core CPU at 100% for any extended period of time and thats what can add to high temps in Prime 95. So as far as heat goes not errors wouldn't monitoring your temps during heavy usage be more accurate? I can say a car gets crap MPG wile its pegged at redline but thats not the proper test for MPG. So temp wise what do you think about that way of thinking?


Well, that's kind of why I didn't mind stress testing my i5 750 at 4.1ghz and hitting almost 80C when Intel's max is only 72C. Under normal scenarios temps didn't go over 68C I think. But, I did make sure it was stable.

Although I use Intel Burn Test which adds more heat than Prime95, but it's also much faster to determine stability. I think you could use LinX on an AMD CPU... not certain but I don't see why not. It's almost the same thing as IBT (they both use linpack)
m
0
l
April 15, 2011 10:31:48 PM

wolfram23 said:
Well, that's kind of why I didn't mind stress testing my i5 750 at 4.1ghz and hitting almost 80C when Intel's max is only 72C. Under normal scenarios temps didn't go over 68C I think. But, I did make sure it was stable.

Although I use Intel Burn Test which adds more heat than Prime95, but it's also much faster to determine stability. I think you could use LinX on an AMD CPU... not certain but I don't see why not. It's almost the same thing as IBT (they both use linpack)

So I stop prime 95 as soon as it hits 65c is it safe to have it past that for 2 hours or so to determine stability? If so at what temp would one stop the test for fear of damage on a AM3 cpu?
m
0
l
April 16, 2011 12:16:16 AM

An unpredictable computer is a broken computer. Unless you don’t mind random data corruption then it is important that your computer passes Prime95.
m
0
l
April 16, 2011 12:29:53 AM

you don't need to run Prime 95 to see if your computer is working...just do an intel burn test and all is well....never had any problems
m
0
l
April 16, 2011 2:11:36 AM

Well things like random data corruption are definitely something I would not want. That's why I was wondering about this in the first place. I didn't know what exactly the side effects would be for a computer that couldn't remain stable under prime.

So oddly enough, just like cburke82 claimed, I found out I am able to run this cpu at 3.8ghz with no vcore increase at all, and perfectly stable under prime. Once I try to shoot for 3.9, everything goes down hill. I have tried as high as 1.46v, but still cannot get it stable under prime. I really do not care to try any higher at this point, I am more than happy with the fact that I can run it 400 mhz higher with no vcore increase and be stable under prime. I have probably just reached the limit of my chip maybe? Or my motherboard simply freaks out when I try to go any higher.

Well thanks for all the replies guys. Also, what exactly is intel burn test? Is it basically like prime but easier to pass or something?

EDIT: I made a typo in my original post. I meant to say that I cannot get it any higher than around 3.85 ghz
m
0
l
a c 197 K Overclocking
April 16, 2011 2:02:20 PM

nebun said:
you don't need to run Prime 95 to see if your computer is working...just do an intel burn test and all is well....never had any known or detectable problems


First question is, "Do I need to run Prime95?" Obvious answer is "No", particularly if you spend your time gaming.

Second question is, "Should I run Prime95?"
I think so. Even if you are gaming, what happens if you start getting a lot of reset/reboot cycles? Do you blame heat, instability, driver problems, a buggy game, the phase of the moon? You have no starting point.

And if you have a mission essential computer, you should start stress testing at stock frequencies.

Third question is, "How long should I run Prime95?"
I am one of those few people who stress test for 24 hours.

I turn off EIST, work out my OC settings, and do the final testing. Final testing is a 24 hour run doing the small fft's test and a 24 hour run doing the blend test. Then I reenable EIST and do another 24 hours of small fft's.

I am kind of fanatical about the subject of stability. :) 
m
0
l
a c 224 K Overclocking
April 16, 2011 4:35:59 PM

Yoshinat0r said:
I have been noticing how insanely hard it is to run prime95 long-term with no crashes. It is practically impossible for me unless I increase the vcore into an uncomfortable range, and my temperatures rise above what I would want them to be at. However, it is so easy to get stable under things like 3dmark and gaming. My question is, what is the benefit of being stable under prime? I never crash in the OS, only when I run prime95, it's just insane, I have never run anything that has stressed all of my cores to 100% and caused my system to crash.

I have a phenom ii x4 965 BE and it seems impossible to get it any higher than like 3.85 ghz and remain stable under prime. I have tried higher frequencies, and I never crash in the os nor in games, but I get destroyed in prime. Do I really need to be concerned if I'm not stable in prime?


You do not specify what you mean by long term Prime95 run, also you don't specify any settings you're trying to use in your overclock, are you actually creating your own instability problems.

You have a 965 BE which is an unlocked multiplier CPU, you do not have to increase your base clock or FSB, or memory speed and timings to successfully overclock it, you do not have to balance out voltages of Vcore, PLL, VTT, GTL, and tune all those voltages back into a specification window to get a stable overclock.

You can successfully overclock it by simply raising the multiplier and discovering the CPU voltage required to run a higher multiplier and leaving everything else within it's specification window.

I've gotten so many PMs from people claiming they have followed my Black Edition Overclock Guide to the letter, then discovering they haven't, when I ask them to list their settings.

If you are truly overclocking simply by raising the multiplier/vcore and leaving everything else within specifications, long term Prime95 runs like 24hrs are not necessary, even though it will pass it, 2 hrs should be more than enough to show a problem.

I've seen Prime95 long term runs crash at about an hour into a game simply because Prime95 is CPU and memory stressing, when you throw the Graphics card 100% into the picture, you just added an element Prime95 did not test so this belief Prime95 is an infallible testing platform is total horse crap.

More and more today with the continuous releases of the AMD Black Edition and now the Sandy bridge K series unlocked CPUs overclocking is becoming so simple, too many today are still Old Schooling it, but times have changed and we're going to see these old practices fall by the wayside when people start to learn not to over complicate a simple overclock, and create your own instability.

We'll probably soon see better testing and stressing programs coming on the scene that include the full access to stressing the graphics cards too, but until then you can use bench marks to stress test, by using the instructions I placed in the Black Edition Overclock Guide!




m
0
l
April 18, 2011 4:04:09 PM

Quote:
NO! It is probabilistic. If you see it in 3 minutes at 65c then it is very likely you will run into it at a lower load and temp and possibly get corrupt data to disk. If you cannot run it for over 4-6 or so hours and decide to do 24x7 then you are taking a big risk with your data and your system.

And I agree, if it is temperature related then it is not a big deal if you don't usually run at high temperature. But 65c isn't that high, so either your CPU or your memory is unstable - at all temps - it just has a higher probability at higher temps. You won't damage your equipment unless you have over-volted, just your data (equipment can be replaced but not your data!).

There must have been a mis understanding. I never get errors on prime 95 I just dont like running it at high temps and as far as I know the AMD chips are only safe up to 65c correct?? I could be very wrong. I re ran prime 95 this weekend to see were the temps stop and they stabalize at 67c so its not as bad as I thought. I ran it through 11 tests with no errors. Is 67c safe for a Phenom 965?
m
0
l
a b À AMD
a b K Overclocking
April 19, 2011 9:19:07 AM

cburke82 said:
There must have been a mis understanding. I never get errors on prime 95 I just dont like running it at high temps and as far as I know the AMD chips are only safe up to 65c correct?? I could be very wrong. I re ran prime 95 this weekend to see were the temps stop and they stabalize at 67c so its not as bad as I thought. I ran it through 11 tests with no errors. Is 67c safe for a Phenom 965?


AMD Spec --- 62c
m
0
l
a b À AMD
a c 286 K Overclocking
April 19, 2011 1:46:59 PM

cburke82 said:
There must have been a mis understanding. I never get errors on prime 95 I just dont like running it at high temps and as far as I know the AMD chips are only safe up to 65c correct?? I could be very wrong. I re ran prime 95 this weekend to see were the temps stop and they stabalize at 67c so its not as bad as I thought. I ran it through 11 tests with no errors. Is 67c safe for a Phenom 965?


67ºC are high, 55ºC or below are recommendable for avoid problems.
m
0
l
April 19, 2011 3:37:07 PM

saint19 said:
67ºC are high, 55ºC or below are recommendable for avoid problems.

I understand that 67c is high I never go past 57c in real life usage and avg prob 54c
m
0
l
!