PC Processors

[19hba66]

When running Prime95 stress-test on a non-OC Ryzen 3600 system, 6 threads return fatal error and stops with the following message on Worker #1, 4, 7, 8, 11 and 12:
#1
[Dec 9 10:55] Worker starting
[Dec 9 10:55] Beginning a continuous self-test on your computer.
[Dec 9 10:55] Please read stress.txt. Choose Test/Stop to end this test.
[Dec 9 10:55] Test 1, 31000 Lucas-Lehmer iterations of M9537183 using FMA3 FFT length 480K, Pass1=384, Pass2=1280, clm=1.
[Dec 9 11:00] Test 2, 31000 Lucas-Lehmer in-place iterations of M9437185 using FMA3 FFT length 480K, Pass1=384, Pass2=1280, clm=1.
[Dec 9 11:02] FATAL ERROR: Rounding was 2922748011, expected less than 0.4
[Dec 9 11:02] Hardware failure detected, consult stress.txt file.
[Dec 9 11:02] Torture Test completed 1 tests in 6 minutes - 1 errors, 0 warnings.
[Dec 9 11:02] Worker stopped.
#4
[Dec 9 10:55] Worker starting
[Dec 9 10:55] Beginning a continuous self-test on your computer.
[Dec 9 10:55] Please read stress.txt. Choose Test/Stop to end this test.
[Dec 9 10:55] Test 1, 31000 Lucas-Lehmer iterations of M9537183 using FMA3 FFT length 480K, Pass1=384, Pass2=1280, clm=1.
[Dec 9 11:00] Test 2, 31000 Lucas-Lehmer in-place iterations of M9437185 using FMA3 FFT length 480K, Pass1=384, Pass2=1280, clm=1.
[Dec 9 11:02] FATAL ERROR: Rounding was 3.598465873e+011, expected less than 0.4
[Dec 9 11:02] Hardware failure detected, consult stress.txt file.
[Dec 9 11:02] Torture Test completed 1 tests in 6 minutes - 1 errors, 0 warnings.
[Dec 9 11:02] Worker stopped.
#7
[Dec 9 10:55] Worker starting
[Dec 9 10:55] Beginning a continuous self-test on your computer.
[Dec 9 10:55] Please read stress.txt. Choose Test/Stop to end this test.
[Dec 9 10:55] Test 1, 31000 Lucas-Lehmer iterations of M9537183 using FMA3 FFT length 480K, Pass1=384, Pass2=1280, clm=1.
[Dec 9 10:59] Test 2, 31000 Lucas-Lehmer in-place iterations of M9437185 using FMA3 FFT length 480K, Pass1=384, Pass2=1280, clm=1.
[Dec 9 11:01] Self-test 480K passed!
[Dec 9 11:01] Test 1, 7000000 Lucas-Lehmer in-place iterations of M83839 using FMA3 FFT length 4K.
[Dec 9 11:02] FATAL ERROR: Rounding was 0.5, expected less than 0.4
[Dec 9 11:02] Hardware failure detected, consult stress.txt file.
[Dec 9 11:02] Torture Test completed 2 tests in 6 minutes - 1 errors, 0 warnings.
[Dec 9 11:02] Worker stopped.
#8
[Dec 9 10:55] Worker starting
[Dec 9 10:55] Beginning a continuous self-test on your computer.
[Dec 9 10:55] Please read stress.txt. Choose Test/Stop to end this test.
[Dec 9 10:55] Test 1, 31000 Lucas-Lehmer iterations of M9537183 using FMA3 FFT length 480K, Pass1=384, Pass2=1280, clm=1.
[Dec 9 11:00] Test 2, 31000 Lucas-Lehmer in-place iterations of M9437185 using FMA3 FFT length 480K, Pass1=384, Pass2=1280, clm=1.
[Dec 9 11:02] Self-test 480K passed!
[Dec 9 11:02] Test 1, 7000000 Lucas-Lehmer in-place iterations of M83839 using FMA3 FFT length 4K.
[Dec 9 11:02] FATAL ERROR: Rounding was 0.5, expected less than 0.4
[Dec 9 11:02] Hardware failure detected, consult stress.txt file.
[Dec 9 11:02] Torture Test completed 2 tests in 6 minutes - 1 errors, 0 warnings.
[Dec 9 11:02] Worker stopped.
#11
[Dec 9 10:55] Worker starting
[Dec 9 10:55] Beginning a continuous self-test on your computer.
[Dec 9 10:55] Please read stress.txt. Choose Test/Stop to end this test.
[Dec 9 10:55] Test 1, 31000 Lucas-Lehmer iterations of M9537183 using FMA3 FFT length 480K, Pass1=384, Pass2=1280, clm=1.
[Dec 9 10:59] Test 2, 31000 Lucas-Lehmer in-place iterations of M9437185 using FMA3 FFT length 480K, Pass1=384, Pass2=1280, clm=1.
[Dec 9 10:59] FATAL ERROR: Rounding was 0.5, expected less than 0.4
[Dec 9 10:59] Hardware failure detected, consult stress.txt file.
[Dec 9 10:59] Torture Test completed 1 tests in 4 minutes - 1 errors, 0 warnings.
[Dec 9 10:59] Worker stopped.
#12
[Dec 9 10:55] Worker starting
[Dec 9 10:55] Beginning a continuous self-test on your computer.
[Dec 9 10:55] Please read stress.txt. Choose Test/Stop to end this test.
[Dec 9 10:55] Test 1, 31000 Lucas-Lehmer iterations of M9537183 using FMA3 FFT length 480K, Pass1=384, Pass2=1280, clm=1.
[Dec 9 10:59] Test 2, 31000 Lucas-Lehmer in-place iterations of M9437185 using FMA3 FFT length 480K, Pass1=384, Pass2=1280, clm=1.
[Dec 9 10:59] FATAL ERROR: Rounding was 0.5, expected less than 0.4
[Dec 9 10:59] Hardware failure detected, consult stress.txt file.
[Dec 9 10:59] Torture Test completed 1 tests in 4 minutes - 1 errors, 0 warnings.
[Dec 9 10:59] Worker stopped.

All cores temperature max out at 95°C.
In a second run Worker #1, 4, 7, 11 and 12 errors, but #8 keeps going.
Is this indication of a faulty processor?

[misterj]

19hba66, impossible to even guess with no information, but  I would be more suspicious of the memory.  Here is the AMD Online Support link to report the errors to AMD.  If you want to see if any users can help you, please post all your parts and a screenshot of Ryzen Master (R) - simply drag-n-drop your image into your reply.  I have seen these errors before but do not remember the problem but not processor.  Please try searching.  Enjoy, John.

[19hba66]

misterj, thx for your answer. Here is the specs and screenshot

Specs:
Fatal1ty B450 Gaming-ITX/ac (3.70)
AMD Ryzen 5 3600 cooled by an ID-COOLING IS-60
2x8 GB G.Skill F4-3200C14D-16GFX Flare X
MSI Radeon RX 560 AERO ITX 4G OCV2
1TB Intel SSD 660p (NVMe)+1 TB 860 QVO SSD (SATA)
Corsair SF450 PSU (450W)
All mounted in Fractal Node 202 and one Fractal Venturi HP-12 PWM fan

[misterj]

19hba66, 450 Watts is way too little and a good candidate for the source of the problem.  Are you running an XMP?  If so, please stop.  In fact, do a Clear CMOS.  Please tell us the version of W10 you are running.  Are you running the latest BIOS version available?  Run Prime95 and post another RM shot after these changes.  Thanks and enjoy, John.

[19hba66]

misterj. I've tried a 550W PSU with the same result. What do you base your advice on regarding the 450W being to small? Every PSU calculator I've used suggests a 300W PSU for this configuration. I then thought 450W would be adequate.

I am running XMP with manual setup by advice from DRAM Calculator, and it runs better than it ever had, just not the Sleep/Hibernater, but that was also a problem before using XMP/DRAM Calc.

Windows is vers. 1909 (18363.476).

Running the latest UEFI 3.70.

I'm a bit surprised by the results from Prime95 because, apart from the boot-loops when using Sleep/Hibernate, the system runs fine, no BSOD's.

[misterj]

19hba66, I am heavy on power and CPU cooling.  I have a 1250 Watt on my 2990WX and had a 750Watt on my 1800X.  I suspect two sources of your errors:  power or memory.  Please do a Clear CMOS, killing the XMP.  Prime95 should run clean essentially forever.  After the test you can run whatever.  I also suggest you open an AMD Online Support ticket.  Thanks and enjoy, John.

[19hba66]

Misterj, thanks, I will try that.

[misterj]

Thanks, 19hba66.  If the problem continues, then please test with one and then the other memory stick.  Thanks and enjoy, John.

[elstaci]

First make sure you have your BIOS/UEFI set to default or factory settings to avoid any errors due to Bios settings being changed.

By any chance are you getting any BSOD's?

This tech site give some good tips on how to check if the motherboard or CPU is defective: https://www.computerhope.com/issues/ch001089.htm

There is a link to download their CPU Stress testing software called CPU HOT TEST Lite version which is free.

From what I have read, it could be a faulty CPU, needing to increase CPU voltages by the smallest increments possible, bad Memory, or possibly bad motherboard.

You can always run MEMTEST86 to eliminate bad RAM MEMORY. If it checks good, remove all the RAM except one RAM module and see it the errors reoccur.

Try stressing your CPU with a different Stress test. I personally like OCCT.  Stress test using the large packet CPU Test and see if you see anything abnormal during the test like temperatures, voltages, etc. No need to install the program. run by itself and is free.

Several sites concerning Prime95 suggests increasing by the smallest increments possible the CPU VCore voltage and see if the errors reoccurs.

Finally, if you want you can always open a AMD Warranty Request and upload your Prime95 test results and see if they believe your processor needs to be RMAed from here: https://www.amd.com/en/support/kb/warranty-information/rma-form 

They may suggest you try some tests or settings first to determine if the cpu is defective. A few Users here have indicated that they had defective 3rd generation processors.

[19hba66]

elstaci, thanks for your comprehensive answer.

No, no BSOD's at all. The system runs OK, but cannot Sleep or Hibernate without boot-loop.

Memtest86 has been run and found no errors.

Hot CPU Tester and OCCT didn't find any errors. I will set them to run for a longer period, than initially tetsted.

I will contact AMD for further actions.