Question New PC - Prime 95 stress tests errors

[Aegis91]

Hi everyone,

Another member here (@remixislandmusic ) has been kindly helping me out with buying a new PC and troubleshooting. Please see my signature for my latest build. I have had a lot of issues so far, so I will give the run down:

1) When I installed windows 10, everything was fine. Then I went to the AMD website and updated my drivers. This caused everything to crash. Tried to repair the OS, and that failed with me getting so many errors at the Windows Recovery screen. So I tried to clean install Windows 10. That did not work, again with many errors. Fortunately, ASROCK has this "SDD secure clean". Using that and installing the OS as a clean install worked fortunately.

2) However, whenever I log-in into Windows 10, I get an error asking me to continue or stop running the following script: https://logincdn.msauth.net/16.000.28230.00/ConvergedLoginPaginatedStrings.en.js . Google says related to Outlook or OneDrive, but I am not sure what this is beyond that. Any ideas?

3) My MB defaulted to DRAM frequency of 2400. AMD website says my CPU supports 2933 MHz and my RAM is 3000MHz. I set it to 3000MHz. I ran MemTest86 - Official Site of the x86 Memory Testing Tool and it was fine. Nothing popped up.

I ran Prime95 torture test and I got several "FATAL ERROR: Rounding was 0.5, expected less than 0.4. Hardware failure detected, consult stress.txt file." errors. Here is what I did:

I went back into the BIOS and set the RAM back to the default of 2400MHz. Did the Prime95 test and it was fine. Went back into the BIOS and put the RAM to 2933MHz. Did the Prime95 torture test, got no error. Then went back into the BIOS and put the RAM to 3000MHz. Error. Went back to 2933MHz, got an error. Went back to 2400Mhz, no error. Went back to 2933MhZ, no error. Simply re-ran the stress test again at 2933Mhz, no error.

For some reason, it was defaulted to set worker to core # 8 for all the tests. I only ran the tests for 5-6 minutes (5 completed tests). Usually the errors happened between tests 1 - 2. Temperature goes up to 77c to 80c. Idle temperatures 38c-40c.

I don't want to overclock at this point.

I am not sure what all this means. But should I be concerned? What parts should I get replaced? If I should be concerned, how can I better diagnose the problem?

Thank you so much,
Aegis.

 

[NightHawkRMX]

I think your hardware is fine.

First, if you are happy with 2933mhz ram, leave it there since its stable. If you want 3000mhz, try raising the dram voltage to 1.375v

 

[DMAN999]

I would just leave the RAM at 2933 MHz.
I don't think setting it to 3000MHz vs 2933 MHZ will give you any real world performance increase.
But as remixislandmusic said bumping the DRAM voltage up to 1.37 or 1.375 may allow you to run it at 3000 MHZ.

 

[Aegis91]

This is weird. I completed another stress test just see. I stopped. It said "0 errors, 0 warnings". But a results text file was generated during this and said:

[Wed Jun 26 20:33:10 2019]
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.

Any advice?

 

[NightHawkRMX]

What speed was your ram at for that run?

 

[Aegis91]

2933Mhz

 

[NightHawkRMX]

Try bumping the dram voltage up a bit to get it stable.
1.375 should work.

 

[Aegis91]

What's the consequences of that? I am nervous about adjusting the voltage haha.

Is it the voltage for the RAM? Or is it a different setting?

Lastly, are we sure this isn't a hardware issue?

 

[TJ Hooker]

Full system specs?

 

[Aegis91]

Full system specs?

What format / prnt screen do you want?

 

[NightHawkRMX]

Dram voltage is what u adjust.

Increasing it slightly from the default 1.35 shouldnt have much adverse effects since your ram has heatspreaders.

Hardware is likely fine, but changing the voltage may fix the issue.

Pc part picker forum form is fine.

 

[NightHawkRMX]

Dram voltage is what u adjust.

Increasing it slightly from the default 1.35 shouldnt have much adverse effects since your ram has heatspreaders.

Hardware is likely fine, but changing the voltage may fix the issue.

Pc part picker forum form is fine.

 

[Aegis91]

Full system specs?

Hi there,

The build link is in my signature (new build).

Please see here:

https://imgur.com/a/kr1UT6B

View: https://imgur.com/a/kr1UT6B

 

[TJ Hooker]

When you're setting your RAM speed, are you doing so by enabling the XMP profile? Or are you just manually changing the memory multiplier and nothing else?

 

[Aegis91]

When you're setting your RAM speed, are you doing so by enabling the XMP profile? Or are you just manually changing the memory multiplier and nothing else?

I was setting the multiplier. Like the little box that says the speed. I move it up to 2933.

Another issue is that when I switch users or log off, I sometimes get a black screen. Restart solves this. But it is weird.

 

[TJ Hooker]

You should be setting by enabling the XMP profile in the BIOS. May also be referred to as the DOCP profile.

 

[Aegis91]

Yeah, you mean this:

https://imgur.com/a/VtJm8xx

View: https://imgur.com/a/VtJm8xx

(Hit the arrow to see when I select the profile)

Now it seems to set it to 3000 when I select that option. I tried running Prime 95 again without making the voltage change, and keeping the frequency to 2933. I did not get any errors this time which is weird...

 

[NightHawkRMX]

Move that option to get 3000mhz. Run prime 95.

Again, if unstable, a minor voltage jump will help.

 

[Aegis91]

Yes. As the other poster said - select the XMP profile versus setting the voltage manually.