Troubles with Prime95

L

loss4words

Weaksauce
Joined
Aug 31, 2007
Messages
107
Hey guys,

I've OC'ed my Q6600 a little bit by pushing the FSB to 340Ghz. CPU temps are around 36C-40C when idle. Vista seems to be running fast and stable but I'm getting getting some weird stuff when I'm trying to stress test the CPU with Prime95. Here's what I get:


FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
Torture Test ran 2 minutes - 1 errors, 0 warnings.
Execution halted.

Should I be worried? I didn't even know I had any hardware issues until I ran this program? Which hardware could be faulty?
 
All it is saying is that "a" voltage settings is either too much or too little. So if you care about being prime 95 stable, play with the voltages more... Not just Vcore, but every option you can..
But if you can use your computer as your normally would, without problems, then ignore it..
No matter what settings I try on the Blend test it always fails under 1 minute, yet I can run SMP Folding and play COD4 and Crysis without problems..
So in short, Stress testing is not the end all be all of stability..
 
It means your computer is not stable. You are getting match errors. Wether or not you should worry depends what you use it for. That said if I get errors in prime within 24 hours I don't consider the comptuer stable at all. I only use stable overclocks and will downclock or increase voltages acordingly to make my PC stable. I don't want corrupted data.
 
Justintoxicated said:
It means your computer is not stable.
Click to expand...

I have found this not to be true at all.
And if you want to use that as "your" standard, then by all means, I will not stop you there, but don't tell people that if you fail stress tests either using Prime95, orthos or TAT that your computer is not stable.

Like I said, I can fail stress tests in prime 95, and now that I think about it, a few in orthos...
YET, I can run SMP folding for days on end, play games like Crysis and COD4 for hours on end, and not have any problems.

For me, through personnel experience, Stress testing is BULL ****.
But, again, if you want to hold yourself to that standard, thats fine, but don't swear by it to others..
 
Blueagle87 said:
I have found this not to be true at all.
And if you want to use that as "your" standard, then by all means, I will not stop you there, but don't tell people that if you fail stress tests either using Prime95, orthos or TAT that your computer is not stable.

Like I said, I can fail stress tests in prime 95, and now that I think about it, a few in orthos...
YET, I can run SMP folding for days on end, play games like Crysis and COD4 for hours on end, and not have any problems.

For me, through personnel experience, Stress testing is BULL ****.
But, again, if you want to hold yourself to that standard, thats fine, but don't swear by it to others..
Click to expand...


Just because things run on your PC does not mean there are no errors. Prime is CLEARLY not able to do complex factoring in this case and is having rounidng errors. It is exactly that " errors". Errors will lead to instability. On a "Stable System" prime will run without errors for days on end.

Hey my car might run fine but also be pooring black smoke out of the exhaust. If it's spewing out black smoke then there is a problem. I could keep driving it buring up oil but eventualy it can lead to engine failure. Same thing with needing an alignment. Car may drive straight but your wearing out the tires due to tow camber etc, there is still a problem. or perhaps you only drifve on sand which is way to forgiving if parts are not perfectly lined up.. So if your driving on sand all day it may not show up, but once you get back on the road you will notice the excessive tire wear.

Keep in mind that when playing games alot of the math calculations are now done by the GPU. If thats all you use your computer for then it might not be a problem for it to not be 100% stable. (Driving with slightly mis-alligned parts on sand). I do have to wonder if your actualy getting errors in folding and sending off bad info...We will never know. Fact is if your getting errors and problems then your not 100% stable, as that is in the definition of stable...
 
Explain to me why SMP Folding runs fine, without errors and prime 95 does not.
Surely you must admit SMP Folding does the same types or similar types of math calculations..

If I can use my computer as I normally would, which includes SMP Folding, and there are no errors, blue screens, restarts or otherwise "problems" running any program, then why should I care if Prime 95 or another stress test program returns errors. It clearly is not making a difference.. If by your defintion, then yes, I am not stable, but normal use stable is all anyone wants, therefore I am stable.
 
Blueagle87 said:
Explain to me why SMP Folding runs fine, without errors and prime 95 does not.
Surely you must admit SMP Folding does the same types or similar types of math calculations..

If I can use my computer as I normally would, which includes SMP Folding, and there are no errors, blue screens, restarts or otherwise "problems" running any program, then why should I care if Prime 95 or another stress test program returns errors. It clearly is not making a difference.. If by your defintion, then yes, I am not stable, but normal use stable is all anyone wants, therefore I am stable.
Click to expand...

ignorence is bliss ;) j/k
if its ok for you then it's ok, but I'll bet if you took away your OC you could run prime 95 just fine. That is the purpose in the program after all, stress the CPU and see how it responds. Just because SMP Folding runs fine could it be possible that is is not designed to do as much error checking? I'm stable up to 2800MHZ but if I go anymore I will get errors in prime after a few hours. Hence I don't consider my PC stable at over 2800MHZ. It will run just not 100% stable.

Considering the OP can only run prime95 for 2 minutes before failure, I would definately recommend reducing the OC. It's quite obvious that something is wrong and your processor is not able to properly compute whatever it is that prime95 is computing durring that test. Should he choose to ignore the problem, then thats fine by me...not my PC.
 
Blueagle87 said:
Explain to me why SMP Folding runs fine, without errors and prime 95 does not.
Surely you must admit SMP Folding does the same types or similar types of math calculations..
Click to expand...

What kind of error checking is built into SMP Folding? True, Prime95 is an artificial test that does not result in anything (unless you're actually trying to find new Merseinne Prime Numbers)...but SMP Folding is not designed to be run on a system that may have rounding errors due to overclocking.

Especially since an EUE might be an error in your Folding...or it might be a valid result...
 
Justintoxicated said:
1.ignorence is bliss ;) j/k
if its ok for you then it's ok, but I'll bet if you took away your OC you could run prime 95 just fine. That is the purpose in the program after all, stress the CPU and see how it responds. Just because SMP Folding runs fine could it be possible that is is not designed to do as much error checking? I'm stable up to 2800MHZ but if I go anymore I will get errors in prime after a few hours. Hence I don't consider my PC stable at over 2800MHZ. It will run just not 100% stable.

Considering the OP can only run prime95 for 2 minutes before failure, I would definately recommend reducing the OC. It's quite obvious that something is wrong and your processor is not able to properly compute whatever it is that prime95 is computing durring that test. Should he choose to ignore the problem, then thats fine by me...not my PC.
Click to expand...
I still don't understand your arguement. I understand that you "may" hold yourself to the standards of these stress testing programs. But let me ask you this, Why do you use stress testing programs in the first place? A guess would be, we use stress testing programs to asertain whether such and such settings will be stable for normal use? Right? But what happens if the stress test programs say your overclock is unstable and normal use says its not? Tell me, why should I care what these stress test programs say if they are telling me that it is unstable, and yet, I can do EVERYTHING on the computer that I want to do without problems..

Under the blend test, Mine fails in under a minute, reguardless of the clock or Vcore.

rainman said:
What kind of error checking is built into SMP Folding? True, Prime95 is an artificial test that does not result in anything (unless you're actually trying to find new Merseinne Prime Numbers)...but SMP Folding is not designed to be run on a system that may have rounding errors due to overclocking.

Especially since an EUE might be an error in your Folding...or it might be a valid result...
Click to expand...

I have no idea about the error checking, but SMP client will give one indications if your overclock is unstable. Either through restarts, blue screens, or the client may state gromacs can no longer continue and kill the work unit. It may also give FILe IO errors.

EUE? End Unit error?

The other interesting item is that it only fails in the Blend test, nothing else. It will run fine in the other two tests.
 
Blueagle87 said:
I still don't understand your arguement. I understand that you "may" hold yourself to the standards of these stress testing programs. But let me ask you this, Why do you use stress testing programs in the first place? A guess would be, we use stress testing programs to asertain whether such and such settings will be stable for normal use? Right? But what happens if the stress test programs say your overclock is unstable and normal use says its not? Tell me, why should I care what these stress test programs say if they are telling me that it is unstable, and yet, I can do EVERYTHING on the computer that I want to do without problems..

Under the blend test, Mine fails in under a minute, reguardless of the clock or Vcore.



I have no idea about the error checking, but SMP client will give one indications if your overclock is unstable. Either through restarts, blue screens, or the client may state gromacs can no longer continue and kill the work unit. It may also give FILe IO errors.

EUE? End Unit error?
Click to expand...

I think my explanation above should suffice. But if you want another example :) ...Your driving your car and the check engine light comes on. Well it's driving fine still so why fix it? You know your computer is computing correctly because why? Because you cannot see the problems or errors it is spitting off? When I'm using a calculator I like to think I am using one I can trust, I don't trust a PC that cannot pass prime. My friend used to OC too and would do the same shit. Run his PC anyways...

Bluescreens are OS system errors and reboots can be for various reasons. Even a stable PC could bluescreen (software problem) or even lockup (software problem) No PC is 100% stable in reality simply due to software like windows. Of course these are the first signs of instability but not the end all. What if you were using your computer to actualy find new prime numbers and it was giving you bogus answers but no blue screens or lockups? I would definately consider this a problem and I would not define the computer as stable...
 
Justintoxicated said:
I think my explanation above should suffice. But if you want another example :) ...Your driving your car and the check engine light comes on. Well it's driving fine still so why fix it? You know your computer is computing correctly because why? Because you cannot see the problems or errors it is spitting off? When I'm using a calculator I like to think I am using one I can trust, I don't trust a PC that cannot pass prime. My friend used to OC too and would do the same shit. Run his PC anyways...

Bluescreens are OS system errors and reboots can be for various reasons. Even a stable PC could bluescreen (software problem) or even lockup (software problem) No PC is 100% stable in reality simply due to software like windows. Of course these are the first signs of instability but not the end all. What if you were using your computer to actualy find new prime numbers and it was giving you bogus answers but no blue screens or lockups? I would definately consider this a problem and I would not define the computer as stable...
Click to expand...

Thats all well and good but how about explaing or giving a reason for have the computer left on for four days striaght running SMP Folding. Going through seven SMP work units without error.
Surely 72 hours is more than ample test for stability?

I understand your point but it still doesn't make since why these errors are not comming out in normal use even when I leave the computer on all the time, and when its not gamming, its SMP Folding..

rainman said:
That sounds like a memory problem.
Click to expand...

Maybe, but it agains begs the question, what does it matter If I can use the computer as I normally would without errors or problems.
 
Blueagle87 said:
Maybe, ut it agains begs the question, what does it matter If I can use the computer as I normally would without errors or problems.
Click to expand...

Because under normal use, your system is now working harder to overcome the errors when/if it detects them.
 
Blueagle87 said:
Thats all well and good but how about explaing or giving a reason for have the computer left on for four days striaght running SMP Folding. Going through seven SMP work units without error.
Surely 72 hours is more than ample test for stability?

I understand your point but it still doesn't make since why these errors are not comming out in normal use even when I leave the computer on all the time, and when its not gamming, its SMP Folding..



Maybe, but it agains begs the question, what does it matter If I can use the computer as I normally would without errors or problems.
Click to expand...


I can do advanced calculus all day and be 100% right, so long as I don't check my answers :)
 
Thats awesome, that I can use my computer without problems even though these so called stress testing programs say it nots fit too..
Sad, but awesome..

Stability during normal use is all any user wants, therefore Im all set.


Isn't that sad?
I just don't see why people stress their systems to a point that will never be replicated through normal use.
 
Thanks for all the replies, everyone. My main concern was that some piece of hardware was failing or was having issues. Apparently, it was all due to my poor overclocking. I reset everything to defualt and I'm getting no errors thus far. And there I thought if I just bumped FSB a little that would be the end of it...guess I need to go and read some more on OCing before I really screw something up :)

Another question I have is that BIOS and Speedfan are giving me different CPU temps. BIOS says it's 46C, Speedfan says it's 38. So which one is correct?
 
Again, read the sticky at the top of the forum.

But the quick answer is: SpeedFan 4.33 and lower do not record temp correctly. They need a +15 C offset. SpeedFan 4.34 is supposed to correct this.

CoreTemp 0.95.4 works (but not under Vista x64).

Everest Ultimate 4.20.1170 works (but is not free).
 
Everest Ultimate is a very nice utility, I can vouch for it. I have been using it on and off for the better part of four years now.



As you can see from the screen shot, it offers alot of different configurations and information displays about your computer..
 
Thanks for responding, guys.

I have Everest Ultimate 3.50.761 and this is what it shows me as far as temps go. They look to be about the same as Speedfan's temps...

my.php
 
Go into the CPUID tab and see what it reports for Tjunction. If it is 85, you need to add 15 to all CPU temps. If it is 100, then you might be getting accurate temps.
 
I would also check with coretemp 0.95.4, just to be sure.
And honestly I would just use coretemp over speedfan or everest for permanent use.
 
rainman said:
Go into the CPUID tab and see what it reports for Tjunction. If it is 85, you need to add 15 to all CPU temps. If it is 100, then you might be getting accurate temps.
Click to expand...

If Tjunction is the same as TJmax Temperature than yes, it's showing me 100C.

If my temps are correct, is 37C idle at stock speed normal for Q6600 B3 stepping? Should I be overclocking this thing at all? I think I've read somewhere that if you have around 40C at stock speed overclocking isn't recommended.
 
loss4words said:
If Tjunction is the same as TJmax Temperature than yes, it's showing me 100C.

If my temps are correct, is 37C idle at stock speed normal for Q6600 B3 stepping? Should I be overclocking this thing at all? I think I've read somewhere that if you have around 40C at stock speed overclocking isn't recommended.
Click to expand...

Where did you read that?

I idle at 38-40 C. Max out at 59-61 C under full load. Seems to work just fine.
 
You must log in or register to reply here.
Back
Top