Prime 95 crashing while stress testing

Toggle sidebar Toggle sidebar
C

crazypotatos

Honorable
Jul 12, 2013
19
0
10,510
I have an I73770k. and an Asus motherboard.

Ok so yesterday i was stress testing to see if my overclock was stable (1.3v at 4.5ghz). I did the custom prime95 test and it almosted completed the first 13 tests on all 8 "cores" A couple were done, a couple werent. Prime95 crashed right then. Does this mean the OC is unstable? or a problem with prime 95?

EDIT: I did notice the power (on core temp) was reaching 76.7 watts and the max TDP(total distributed power maybe idk?) is 77

Thanks
 
Sort by date Sort by votes
nukemaster

nukemaster

Titan
Nov 23, 2004
19,626
6
74,965
If you are testing custom to use more memory(some users prefer 80-90% use), It may also indicate a memory/compatibility problem. You can try If you are running 4 sticks of memory, try to run 2. If Your memory is overclocked, try it at stock.

If not, try to set the clock to 4.4 and give it another go.

How are the temperatures?
 
Upvote 0 Downvote
C

crazypotatos

Honorable
Jul 12, 2013
19
0
10,510
I wont be OCing the memory. I just have it set at the manufacturer tested stuff. I have 2 sticks. So I put it at 4.4ghz at 1.28v and it is stable for 1 pass through prime95 (about 15 minutes, haven't had time to do more). Temps max out at 68 and are usually at 65. I am running the custom one though. If I run the first option prime95 gives you, the temps will usually jump up to 75.

EDIT: ok so prime95 is in the second test sequence AVX fft length 8k and my temps have gone up to 76 max. Also my power is at 84.7 watts
 
Upvote 0 Downvote
nukemaster

nukemaster

Titan
Nov 23, 2004
19,626
6
74,965
You should run the at least the first 2 tests(Personally 12+ hours of each is good)

While 75c is hot, it is not chip damaging(at least not in short term).

The more stress, the more stable you know it is.

as geofelt says, OCCT is another option but from what I have seen in my own testing, It is even more stressful in some ways(I tested with both of them).

EDIT:
Power consumption will climb as one overclocks.
 
Upvote 0 Downvote
xerxces

xerxces

Distinguished
Dec 28, 2010
328
0
18,810
The real question is why are you OCing? 99.9999% of people will not even notice or need the extra power. The 3770 is just fine on its own, the only thing you are accomplishing is shortening the life the of CPU. Also, there are only four cores on the CPU and 8 "threads". Intel kind of misleads people if you ask me.
 
Upvote 0 Downvote
C

crazypotatos

Honorable
Jul 12, 2013
19
0
10,510


I play FSX which is designed for single core 7ghz cpus which don't exist, hence overclocking
 
Upvote 0 Downvote
geofelt

geofelt

Titan
Oct 9, 2006
54,268
4,130
174,240


FSX has some updates and options.which enable multithreading. Find that and you should do much better.
 
Upvote 0 Downvote
nukemaster

nukemaster

Titan
Nov 23, 2004
19,626
6
74,965

It is not like the OP is pushing for 5.0+ with 1.45+ voltage.

I have had cpus overclocked for many years without any issues as long as you do not use stupid voltage and let it run too hot(Prime95 runs hotter than most games), they do not degrade fast enough for it to be an issue.

I am almost sure the OP's cpu will be LONG obsolete before it dies from overclocking.

I do agree that HT does not help most games.
 
Upvote 0 Downvote
C

crazypotatos

Honorable
Jul 12, 2013
19
0
10,510


I have, mine is completely updated. I haven't enabled ultra high on everything, and I get 30-40fps on the ground (during takeoff and landing). In the air I have gotten up to 120 lol. But I want to run REX as well and good payware aircraft look really nice and require more power.
 
Upvote 0 Downvote
C

crazypotatos

Honorable
Jul 12, 2013
19
0
10,510


FSX doesn't take advantage of hyperthreading very well and favors cpu speed over amount of cores
 
Upvote 0 Downvote
geofelt

geofelt

Titan
Oct 9, 2006
54,268
4,130
174,240
I figure a hyperthread is worth perhaps 25% of a full core.
It is not clear to me if windows is smart enough to dispatch critical tasks to the main cores and not to the hyperthread tasks,
In your case, you might try disabling hyperthreading on your 3770K.
An added benefit is that you can usually oc a bit higher without hyperthreading enabled.

The benefit of the 4930K is that it has 6 full cores. The hyperthreading may or may not be a plus.
But you will add more than 50% to your FSX compute capability. Ivy bridge is some 10% more efficient per clock.
 
Upvote 0 Downvote
DrGonzo

DrGonzo

Honorable
Aug 12, 2013
8
0
10,520
Running Prime 95 or any stress testing software for 12 hours is a fools errand as most stock systems won't pass these tests. Use the computer normally and see what happens. The only time I've ever had any trouble with my over clock is WHILE running prime 95. This caused me to start making infinitesimal adjustments to all of my settings which resulted in the only BSODs I've ever had. I went back to my working OC and forgot about stress testing. I pass Intel Burn Test about 12 out of 15 times on very high. You'll know if your overclock doesn't work because your system won't boot. Don't over think it. I'm sure many members won't like what I have to say but I basically started this account to offer a "no need to stress test" point of view.
 
Upvote 0 Downvote
nukemaster

nukemaster

Titan
Nov 23, 2004
19,626
6
74,965

I did 48 hours on my overclocked system and have NEVER had a stock system fail it.
 
Upvote 0 Downvote
DrGonzo

DrGonzo

Honorable
Aug 12, 2013
8
0
10,520


Well, I hate to tell you this but I've had stock systems fail Prime 95 out of the box on more than one occasion. And when running the test succinctly, I've had varied results. As in, it passes the test one time and fails the test the next. Why would that be? Why would over half of the systems I've encountered inconsistently pass or fail these tests? The point is does the system run without problems. If the answer is yes, no need to stress test. Why would you do that? It's not like back in the day. Now every motherboard, OS, and hardware component has so many locked fail safe measures (which often are the root cause of stress test failure by the way) that it's almost impossible to actual destroy a system, so why? Why this need to stress components in a way that is SO unrealistic. Because most people are sheep and do what the internet tells them that's why. If your system works it works. Stop stressing it to the max.

EDIT :Also, no one has attempted to answer why 24 hours is the "magic" number. Why not 25 or 2000 hours?
 
Upvote 0 Downvote
nukemaster

nukemaster

Titan
Nov 23, 2004
19,626
6
74,965
I use at time because when I started overclocking, renderings took at least that long. Those types of loads had been quite similar to prime95. It was nice to know that when a system was left to do some long term rendering or video encoding(something that takes much less time now) with finish the task.

Just because you do not use your system for anything stressful does not mean that other users do not. I fully understand that many users only use a small portion of the processing power of a computer.

You are on a forum with many power users(who I guess you call sheep) who honestly would have never been able to use your half defective systems.

Think about it this way, Your 50% non prime stable systems also would be unable to run folding@home or any other distributed computing project(Prime95 is actually one as well) as well. They may even have issues running longer rendering sessions or larger video encoding sessions as well.

Simply put, if a system fails Prime95 it is defective(or using incompatible components. Check the QVL) or clocked too far(RMA if it is at stock and sorry for the bad luck).

Saying if the system runs(or works as you call) it is stable is pure BS.
I can push a system just over the edge and it may not crash for months, but at some point it will(under normal use).
 
Upvote 0 Downvote
DrGonzo

DrGonzo

Honorable
Aug 12, 2013
8
0
10,520


I apologize for the sheep comment that was rude. My real question is: Why is 24 hours this magic number? If four or eight hours isn't enough then why is twenty-four enough. It seems completely arbitrary to me. As a system that is stable according to Prime 95 for twenty-four hours, may not be stable according to Prime 95 for twenty-five or thirty-six hours, so by that logic you would need to endlessly run Prime 95 to make sure your system was stable. Which is fine as long as you don't want to use the machine for anything other than folding or stability testing. This is a legitimate question and all of the "power users" seem to simply ignore it.

Also, why would a system pass Prime 95 for one twenty-four hour duration and fail the next? It's much easier to see these result when using Intel Burn Test since you don't have to wait twenty-four hours. This is a very common occurrence and is in fact the reason I stopped relying on both of these programs. If the end user is going to fold, I test the system through folding. If the end user is going to game, I test the system by gaming. I just really want to know how these test can be seen as the definitive stress test when they pass some fail some.


 
Upvote 0 Downvote
nukemaster

nukemaster

Titan
Nov 23, 2004
19,626
6
74,965
I use 24 because I have seen some errors crop up after 12(please note this was a memory issue not a cpu issue. It would have a very limited crash potential, but it would happen from time to time.).

Very few systems are run at this load over much more time.

The amount of prime95 or other stress tests depends on the user. After all, if you are not doing anything stressful, prime for just a few hours to ensure your heatsink is on right/working the way you want it to and no other problems are found is more than enough.

Nothing is more of a pain than having a system/program crash midway though work or play. I do this to ensure that it has the smallest chance of happening.

Another case of using Prime95 was when I wanted to lower the voltage on a cpu, This allowed me to find the lowest voltage the system could run at. It was also a good test for the cooling system(SFF case).

I generally do not run Prime95 on a stock system(Just like I do not stress video cards with anything but games), just for overclocking or undervolting. I will run it on systems that crash to ensure that heat or errors are not the cause.

Unfortunately, I did have a video card that was bad at stock(more to the point Nvidia's boost was pushing it into instability). It seemed ok for 2 months because I was not playing anything too demanding, but as soon as I either pushed the settings or played certain games it would driver crash constantly or sometimes it would artifact quite bad.

So sometimes a bit of stress testing may avoid having to RMA with the cards maker(at your own cost many times). For the first 30 days, My RMAs come from the supplier with pre-paid shipping both ways.

Anyway, I think we need to stop thread jacking.

PM if you wish to discus this further or even start a thread about if Prime95 is worth it(I would guess you may find just as many for as against it).
 
Upvote 0 Downvote
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom