Prime95 Error on my new IC7-G Rig

[Wurrmm]

Well, I let prime95 go last night.
IC7-G
Mem XMS PC3500C2 @ 400MHz 2-3-3-7 2.7v
CPU P4C 2.8GHz @ Default clock and 200FSB
Performance Mode @ Fast

Started it at 1:27 EST and here is what happened:

[Wed Jun 04 01:27:44 2003]
Self-test 1024K passed!
.
.
.
[Wed Jun 04 08:59:02 2003]
Self-test 1536K passed!
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.

Is this good or bad??? Is there anyway to determine what caused the error (this is my first time using Prime95)?? Anands Review said they went 24hrs straight aft what seems like more aggressive settings. Is there memory, CPU, mobo, rated better than most retail???

 

[Thor86]

Originally posted by: Wurrmm
Well, I let prime95 go last night.
IC7-G
Mem XMS PC3500C2 @ 400MHz 2-3-3-7 2.7v
CPU P4C 2.8GHz @ Default clock and 200FSB
Performance Mode @ Fast

Started it at 1:27 EST and here is what happened:

[Wed Jun 04 01:27:44 2003]
Self-test 1024K passed!
.
.
.
[Wed Jun 04 08:59:02 2003]
Self-test 1536K passed!
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.

Is this good or bad??? Is there anyway to determine what caused the error (this is my first time using Prime95)?? Anands Review said they went 24hrs straight aft what seems like more aggressive settings. Is there memory, CPU, mobo, rated better than most retail???

Most likely your ram peformance setting is not taking well. Try 2.5-3-3-7 timings and disable performance setting and try torture testing again.

 

[Wurrmm]

Actually, I was just told 7.5 hrs is pretty good. I am gona leave it as is. But thanks.

 

[chocoruacal]

Originally posted by: Wurrmm
Actually, I was just told 7.5 hrs is pretty good. I am gona leave it as is. But thanks.

That doesn't mean there isn't a problem, and that it won't appear again down the road. If its a new build...wouldn't you want to root out the problem while you still have the opportunity to return parts Whatever floats your boat, though

 

[jdogg707]

Originally posted by: Wurrmm
Actually, I was just told 7.5 hrs is pretty good. I am gona leave it as is. But thanks.

Kill the performance setting, it doesn't really do anything but mess with system stability! And if you are getting a Prime error, then there is a problem...no matter how long you let it run...but if complete system stability doesn't matter...leave it as is.

BTW, shouldn't your RAM run at 2-2-2-6 or something at DDR400? I know Corsair PC3200C2 rates at 2-3-3-6...

 

[Wurrmm]

It is my belief that not everyone can get something perfect, and it is likely that same thing would happen even if I got all replacement parts. Not everyone uses Prime95. I am also sure that if I turned my performance mode to disabled then it would run flawlessly. I just finshed reading the Scientific American Article call Computer Heal Thyself and it has led me to this conclusion.

 

[Wurrmm]

I am thinking about disabling Performance mode and going to 2-2-2-6 2.7v which my memory supports. But I really like what I have now.

 

[Wurrmm]

Just found out I had a bad stick. I indivisually tested each with memtest 86 and even though in dual channel, no errors, one got errors and the other did not. I also set my timings to 2-2-2-6 2.7v and one stick was bad and the other was not, where as I was getting constant errors in dual at those times. If the new stick is good, then I should beable to get 2-2-2-6 perfect with both from memtest86.

 

[Aboroth]

If Prime95 finds something wrong then there *is* something wrong with the stability of your hardware, and other applications are affected, not just Prime95. Maybe you don't notice problems but don't be surprised if you have weird stuff happen some day.

 

[Wurrmm]

Gona give prime95 another go tonight.....before I go to bed. I only have one stick in now, but I want to see what happens.

 

[jdogg707]

Originally posted by: Wurrmm
It is my belief that not everyone can get something perfect, and it is likely that same thing would happen even if I got all replacement parts. Not everyone uses Prime95. I am also sure that if I turned my performance mode to disabled then it would run flawlessly. I just finshed reading the Scientific American Article call Computer Heal Thyself and it has led me to this conclusion.

But the thing is, shouldn't computer hardware, under ideal conditions with everything functional be able to produce near perfect results? I mean Prime95 deals with mathematical equations, how can an equation be run in between right and wrong, it's either a 1 or 0 correct? Also, shouldn't with your computer system be looking for complete stability? I mean, when I load my personal data onto a system I want it to be as secure as possible and not have to worry about waking up to find my hard drive littered with bad sectors or corruption in my memory!

 

[WarCon]

I think there is some confusion as to the usefulness of Prime95 as a stability bench. Alot of the more hardcore people don't think its a good sign of stability, but not because it falsely reports problems. It sometimes will run endlessly even though a system will fail to be able to encode video (just an example btw) at the speed its at, and thats why some people don't use it as their standard for stability.

But if it has an error condition, then something is unhappy (even if only very marginally) in your system.

 

[Duvie]

Originally posted by: WarCon
I think there is some confusion as to the usefulness of Prime95 as a stability bench. Alot of the more hardcore people don't think its a good sign of stability, but not because it falsely reports problems. It sometimes will run endlessly even though a system will fail to be able to encode video (just an example btw) at the speed its at, and thats why some people don't use it as their standard for stability.

But if it has an error condition, then something is unhappy (even if only very marginally) in your system.

warcon nailed it!!!! And this is more so associated with the newer p4 chips like yours wurrmm......

I would pick a time you are confortable with...I like 1 hours during preliminary when I am first trying to find the cpus overall limit then 12 hours max when I think I have it nailed down. Then live with it. If it errors within the 12 hours realize it does not met your parameters and fix it.

I think you may have solved it, but I think it is even more crucial that a system not even running out of spec errors in prime95 means you want to solve this now...

 

[AdamK47]

I agree with some of you here. I use Prime95 as a preliminary stability test. If it passes Prime95 I move on to something else like Memtest, or Memtest + Prime95. Then the next day I'll loop a stressful DX8.1 or DX9 demo.

 

[Duvie]

I like to run a nice multiple pass 2 hour plus Divx movie with high bitrate and all the bells and whistles....NOt some Divx software itself is buggy but make sure it runs fine at default levels before you use it as a stress tester...

 

[Wurrmm]

I am hoping I can go all night and into the next day with Prime95 now that I have taken out the bad mem chip. The RAM guy said I should be able to run my XMS PC3500C2 at 2-2-2-6 2.7v and 400MHz. Since the chip I took out was getting all sorts of erros at that speed I am gona replace it.....test the new chip then try my good chip with the new chip and go from there.

 

[Tom]

I've always been curious about whether external events, power surges, cosmic rays, noodles, could cause Prime95 to glitch ?

 

[Wurrmm]

Noodles??? Well, if you spill ramen on your rig while running Prime95, I am pretty sure it will crash.