• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Prime95 fatal error, need help to find the problem

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

dagp06

Registered
Joined
Jan 5, 2015
Hi everyone,

I used Prime95 to stress my computer and after less than a minute, my screen froze and I wasn't able to do anything for several minutes. I rebooted and looked at the results.txt.

"FATAL ERROR: Rounding was 0.484375, expected less than 0.4
Hardware failure detected, consult stress.txt file."

I have already run a Memtest for 28 passes and there wasn't a single error so there must me a problem somewhere else than memory.

I built my computer last summer, here's a list of my hardware:

Mobo: ASUS M5A97 LE R2.0
CPU: AMD FX(tm)-8350 Eight-Core Processor
GPU: NVIDIA GeForce GTX 760
PSU: Seasonic M12 II Bronze 650W
Ram: 1x 8Go G. Skill Ripjaws
SATA Disk: Hitachi HDS72202 (2To)

I'm running windows 7 pro 64 bits.

Tell me if you need anymore information!

Thanks
 
Which stress test did you use in P95?
What were the temps when doing the stress test and which program? (screenshots would be awesome here).
Are you overclocked?

That said, that motherboard, regardless of what the CPU compatibility list says, has NO cooling on the VRM area around the socket. I am betting that is getting to hot and shutting things down. I think a better motherboard is in your future...
 
I did the Small FFTs test. I cannot take a screenshot as my screen freezes. Should I wait until it unfreezes or will it stay like that forever until I reboot?
It freezes in about 5 seconds. I don't think the temperature has time to get that high in such short time.

I am not overclocked.
 
Although in my mind and mind of many, this:
Mobo: ASUS M5A97 LE R2.0
CPU: AMD FX(tm)-8350 Eight-Core Processor << are not a good combination over the long haul, it maybe that you have dust in the fins of coolers which is now inhibiting cooling to the point of pushing the system over the edge.

Frozen screen will in all likelihood remain until a reboot.
RGone...ster.
 
It would stay like that forever. Start Hwmonitor and have it running, start P95. When it freezes, use your phone to take a picture of the screen and upload that?

I'm more concerned over the motherboard and its woefully under equipped power delivery area for the CPU you have running then the temps of the CPU to be honest.

What CPU cooler is on the 8350?
Is this the first time you stress tested? If not, it passed before?
What case do you have? What is the airflow inside of it like (fan locations and direction)?
 
Last edited:
Hi.

I did what you asked me to. Here are 3 pictures:

HWMonitor when its frozen:
http://postimg.org/image/wujxqhl3f/

Prime95 when it's frozen (we see that test #6 gets an error):
http://postimg.org/image/4vo8a2lu1/

And my computer (sorry for bad light/quality). My case is the Antec Eleven Hundred V2 Gaming Case
I have 3 fans on my case. One on the back and 2 on top, but I had only room for 2 fans on my mobo so I only plugged one of the top fans. Then I have 2 fans blowing in the direction of CPU (one on top and one in the back).
The airflow doesn't seem to be a problem here, but I'll let u judge by yourself.
http://postimg.org/image/m78kzqjo5/

I don't know what CPU cooler is on the 8350. I didn't add one.
I tested 3 times today, computer froze about 5 seconds after test started the 3 times. It was the first times I stress tested.

Thanks for taking your time to help me.
 
Here is what I see...

1. That cooler may not even be able to support that CPU under load. but who knows because when you hammer the CPU, it freezes...
2. Your airflow in that case is not good enough for the system. No intake fans? SOMETHING is going to need to cool that power delivery area. If you don't plan on upgrading the motherboard (I would...) you need to get those things heatsinked and have fans on them to have a chance with that CPU and it not throttling. You need to have intake fans too. Either on the side, or on the front. If you don't replace that mobo, I would put two on the side blowing in on the mobo. You also have another fan header on the middle of the board (CHA FAN). Perhaps a fan controller is in order for you if you do not buy a motherboard that can support more fans (or a splitter for low power fans).
3. When you took the Hwmonitor shot, was that AFTER it froze after the stress test? We want to see CPU LOAD temperatures, not idle and it looks like nothing moved at all there.

Does it freeze gaming? Internet? etc? Have Hwmonitor run and play a game for 15 mins and see what the temps are...
 
It never freezes at any other time than with Prime95.

I've run Watch Dogs, and the reason why I first stress tested my cpu happened: BSOD. BluescreenView tells me it's caused by hal.dll and ntoskrnl.dll.
Last time I played (more than a month ago), I've played at least 20 minutes before the BSOD occured, but today, the game opened, I clicked on "Start Game" and while loading --> BSOD.
 
I have to be honest. I am not sure where I would start if I was you. I suppose the cheapest way to do it is to get more fans in there blowing on the power area around the socket and see if that helps. If that doesn't do it, add heatsinks on the VRM area (may want to do that AND fans actually). Even open up the side of the case and put a house fan blowing on it THEN start a stress test. If that doesn't work, I would get another motherboard.

Try putting a load on the GPU only with something like Unigine Heaven. That stresses the CPU minimally but beats on the GPU. I am asking this to see if that quality PSU is still doing its job. If Heaven doesn't freeze/BSOD, then its the motherboard/that power area giving you fits. Seeing as how its pretty instant though, I'm thinking the motherboard power area is giving out after a year of running that 8350 with no heatsinks and poor case airflow.
 
Last edited:
I used FurMark earlier this morning, but how long should I let it run? I've run it for 5-6 minutes, and the temp of my GPU became steady at 74°C.

And I'm sorry if this is a stupid question (lol), but if I get a new motherboard, is it possible to sell the one I own now? Cause if I understood, it's not defective, it's just not the one I need. I don't know if it's possible to sell a kinda new but used mobo.

And I saw on the web that ASUS M5A97 R2.0 should be fine (I have a ASUS M5A97 R2.0 "LE"). What do you think?
 
Give it 10-15 mins if you are freezing/BSOD at 5 mins or so in the CPU tests.

Well, no way to be sure its not defective and on the way out. It may work with a hex/quad FX CPU, who knows. There isn't a 'kind of new, its just used, LOL! :)

I saw that too at OCN in their database, but I think in your situation, particularly due to the lack of airflow in your case, you need a more robust solution that has heatsinks on the power delivery area.
 
Ok I will run furMark for 10-15 mins.

Do you have any particular heatsink you would recommend to me (with a link if possible (for canadian buyers lol))? And what can I do to have my 3rd fan working?
 
Heatsink for the CPU? Coolermaster Hyper 212 Evo is a good bang for the buck.

As far as your 3rd fan (and frankly you need more as I said earlier), as also mentioned earlier, you should have a spare CHA FAN header on your board just above the PCIe slots. Does that Antec case have a fan controller built in? If not, you will likely need either splitters so you can run two fans on one header, or a fan controller.

Here is a great read from RGone who helped above: http://www.overclockers.com/forums/...s-gt-gt-Using-the-AMD-FX-Bulldozer-Piledriver
 
I don't have a fan controller so I will go for the splitters, I'll look that up, I'm a newbie when it comes to hardware. I will check if I have a spare fan header on my board.

I runned FurMark for about 12-13 minutes without any problem.

I'm off for today. Thanks a lot for your help!
 
With splitters though, you have to be careful not to overload the header. No more than 12W/1A on it (you can see this information on the back of the center of the fan). But with cheap case fans, you should be fine.

I saw you have another header on your board. I mentioned where it was located. You should have four total. 3 up top by the CPU, and one above the PCIe slots.
 
Last edited:
Judging by the freezing, I will bet that it's the motherboard. The 4+2 Vrm section can't handle the power delivery when under 100% load. You can try putting a spot fan on the Vrm heatsinks and see if it helps.
 
I opened my pc up and I found the 4th fan header you were talking about, so my 3rd fan is now working.
I runned Prime95's same test with one side opened and a house fan in front of it as you suggested, here are the results:
http://postimg.org/image/bn75fo9r5/

It froze, but much later than before. It lasted ~50 seconds instead of 5 seconds. But still, it froze.
I don't know if adding fans and heatsinks will completely solve the problem. What's your guess?
 
Back