[SOLVED] Fatal error during Prime95

May 18, 2021
35
2
45
So, in the last weeks I've been having PC restarts (power loss and turn on again, kernel power 41) during gaming/under load. Today I decided to run Prime95 to see if it was going to make my PC crash and mainly which were the PSU voltages under load (because I had a suspicion about it, and indeed, 5V shows 5.269-5.280 in hwinfo64).

Thing is, i was going to turn off the test some minutes after starting it and saw that the stress test had stopped, around 11 minutes after the beginning of it. It was a fatal error, specifying this on the results file:

FATAL ERROR: Final result was 00000000, expected: 87610A79.
Hardware failure detected running 288K FFT size, consult stress.txt file.

(Repeating 6 times, because of the 6 cores/threads)

I don't know if this has something with the problem I talked about earlier, and if maybe CPU can be the cause (I'm afraid of this now). My motherboard is Gigabyte H310M M.2 2.0, CPU is i5-9400 and RAM is a Mushkin one (don't remember the model), one 8 GB stick at 2666 MHz CL19. I know this kind of problems is oftenly seen in overclockers but I'm not running an overclocked processor, as it is locked to it; anyways i have turbo boost enabled which boosts my CPU from 2.9 GHz to 3.9 GHz.

I appreciate any suggestion or idea.
 
Solution
Try running Prime 95 in blended mode , or small FFTs (with AVX/AVX2 disabled); blended mode which causes much more shifting of data to/from RAM, but, default mode with AVX or AVX2 enabled was formerly considered a partial overload...

If running Prime95 in default mode (with AVX enabled), it is used to be considered a partial overload, and certainly might be a stressor on a weak PSU or some mainboards' VRM circuitry... Perhaps that CPU will not pass whatever mode your running on default core voltage. If you have another RAM stick, try it, can't hurt...

It is indeed fairly unusual to see a failure in non-overclocked CPUs ...

popatim

Titan
Moderator
you can run memtest86+ for 6+ hours and see if there are memory errors but this could just be an incompatibility with the motherboard.
If you et a memory error you can quite he run. Shut down the PC, remove 1 stick (make sure its in the correct slot for a 1 stick setup) and then run it again and then test the other stick.
 
May 18, 2021
35
2
45
you can run memtest86+ for 6+ hours and see if there are memory errors but this could just be an incompatibility with the motherboard.
If you et a memory error you can quite he run. Shut down the PC, remove 1 stick (make sure its in the correct slot for a 1 stick setup) and then run it again and then test the other stick.
So, do you mean this should be a memory error? I'll see if it's possible running Prime95 just stressing CPU later. I'll do memtest86 in my next AFK time.
 
Last edited:
May 18, 2021
35
2
45
you can run memtest86+ for 6+ hours and see if there are memory errors but this could just be an incompatibility with the motherboard.
If you et a memory error you can quite he run. Shut down the PC, remove 1 stick (make sure its in the correct slot for a 1 stick setup) and then run it again and then test the other stick.
Hey, I ran the default 4 passes of memtest86, took a bit more than 2 and half hour and didn't detect any error. Could this be something with the CPU or Prime95?
 
Try running Prime 95 in blended mode , or small FFTs (with AVX/AVX2 disabled); blended mode which causes much more shifting of data to/from RAM, but, default mode with AVX or AVX2 enabled was formerly considered a partial overload...

If running Prime95 in default mode (with AVX enabled), it is used to be considered a partial overload, and certainly might be a stressor on a weak PSU or some mainboards' VRM circuitry... Perhaps that CPU will not pass whatever mode your running on default core voltage. If you have another RAM stick, try it, can't hurt...

It is indeed fairly unusual to see a failure in non-overclocked CPUs ...
 
  • Like
Reactions: JuanOlgomiec
Solution

TRENDING THREADS