Prime95 confusion HELP

[lukej92]

I just ran prime95 because i have been experiencing bsod and have tried just about everything, thought id give this a shot.
Anyway,
First time i run it 4 workers failed pretty much straight away (maybe 5 minutes in?)

Second time i run it 2 workers failed after 30 minutes

bringing back this error..
[May 17 21:07] Worker starting
[May 17 21:07] Setting affinity to run worker on logical CPU #5
[May 17 21:07] Beginning a continuous self-test to check your computer.
[May 17 21:07] Please read stress.txt. Choose Test/Stop to end this test.
[May 17 21:07] Test 1, 26000 Lucas-Lehmer iterations of M12451841 using AMD K10 type-2 FFT length 640K, Pass1=640, Pass2=1K.
[May 17 21:14] Self-test 640K passed!
[May 17 21:14] Test 1, 3200000 Lucas-Lehmer iterations of M172031 using AMD K10 type-1 FFT length 8K, Pass1=32, Pass2=256.
[May 17 21:22] Self-test 8K passed!
[May 17 21:22] Test 1, 21000 Lucas-Lehmer iterations of M14155777 using AMD K10 type-2 FFT length 720K, Pass1=320, Pass2=2304.
[May 17 21:28] Self-test 720K passed!
[May 17 21:28] Test 1, 1800000 Lucas-Lehmer iterations of M250519 using AMD K10 type-0 FFT length 12K, Pass1=48, Pass2=256.
[May 17 21:35] Self-test 12K passed!
[May 17 21:35] Test 1, 18000 Lucas-Lehmer iterations of M15597569 using AMD K10 type-2 FFT length 800K, Pass1=640, Pass2=1280.
[May 17 21:37] FATAL ERROR: Rounding was 0.5, expected less than 0.4
[May 17 21:37] Hardware failure detected, consult stress.txt file.
[May 17 21:37] Torture Test completed 4 tests in 30 minutes - 1 errors, 0 warnings.
[May 17 21:37] Worker stopped.


what should i do? what could be wrong

 

[RaDiKaL_]

Monitor temps and voltages with HWMonitor Pro or AIDA64 while running Prime95, could be an overheating issue or a failing PSU, also which kind of test are you running? if you use the blend or Large FFT it could be a RAM issue, try with the small FFT option instead.

 

[ezskills]

Are you overclocking

 

[lukej92]

Monitor temps and voltages with HWMonitor Pro or AIDA64 while running Prime95, could be an overheating issue or a failing PSU, also which kind of test are you running? if you use the blend or Large FFT it could be a RAM issue, try with the small FFT option instead.

Its my first time running prime95 I selected the blend test.
I had a feeling my psu was failing, I was watching the power on hwmonitor and it was jumping from 50w down to 30w then back up to 50w and all over the place to be honest, I have a new PSU coming tomorrow anyway because the PSU i currently have is awful, Its a 550w coolermaster gx series.
I run memtest for 8 hours with no problem before i run prime95

EDIT: I did monitor the temps and at full load it was 50c

UPDATE: I run prime95 on small fft for 8 hours with no problems at all!
As soon as i try to run blend test it fails either 2, 3 or 4 workers

 

[lukej92]

Are you overclocking

I did have it overclocked but i reset everything last week, I didn't really take it too high either.