mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > Software

Reply
 
Thread Tools
Old 2020-09-06, 14:53   #1
Beinish
 
Sep 2020

19 Posts
Default Test found 2 errors a few seconds after starting

Hello,

I started a Prime95 test with Large FFTs and 2 threads(?) found an error and stopped after a few seconds:
Quote:
[Sun Sep 6 17:49:05 2020]
FATAL ERROR: Rounding was 0.4582755465, expected less than 0.4
Hardware failure detected, consult stress.txt file.
FATAL ERROR: Rounding was 0.4998263118, expected less than 0.4
Hardware failure detected, consult stress.txt file.

The rest of the tests kept working. Should I run it again and leave them with the error?

Click image for larger version

Name:	prime95_eBl8oj8Ukq.png
Views:	30
Size:	126.1 KB
ID:	23275

Rig:

CPU: Ryzen 5 3600
Motherboard: ASUS TUF B450M Plus Gaming
Memory: Corsair Vengeance LPX 16GB (2x8GB) DDR4 DRAM 3200MHz C16
Graphics Card: EVGA RTX 2070 Super (Currently with a GTX950 for testing)
PSU: Antec EA750G Pro 750W Gold
Hard Drives: Crucial P1 1TB 3D NAND NVMe PCIe M.2 SSD

Last fiddled with by Beinish on 2020-09-06 at 15:03
Beinish is offline   Reply With Quote
Old 2020-09-06, 16:45   #2
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

41·113 Posts
Default

Why so many workers at 12? Isn't that a 6-core cpu? PRP or LL testing perform better with no HT use, generally. Prime95 has benchmarking features to determine how many cores/worker is greatest throughput. Consider running 1,2,3,or 6 workers, not 4 or 12. You might find it useful to read https://www.mersenneforum.org/showthread.php?t=23900 or https://www.mersenneforum.org/showthread.php?t=24588

Last fiddled with by kriesel on 2020-09-06 at 16:50
kriesel is offline   Reply With Quote
Old 2020-09-06, 17:08   #3
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

112×59 Posts
Default

Quote:
Originally Posted by kriesel View Post
Why so many workers at 12?
This is a stress test. More workers = more stress.

Most stress test errors are RAM related. Try overvolting your RAM or underclocking your RAM.
Prime95 is offline   Reply With Quote
Old 2020-09-06, 17:11   #4
Beinish
 
Sep 2020

238 Posts
Default

Quote:
Originally Posted by kriesel View Post
Why so many workers at 12? Isn't that a 6-core cpu? PRP or LL testing perform better with no HT use, generally. Prime95 has benchmarking features to determine how many cores/worker is greatest throughput. Consider running 1,2,3,or 6 workers, not 4 or 12. You might find it useful to read https://www.mersenneforum.org/showthread.php?t=23900 or https://www.mersenneforum.org/showthread.php?t=24588
Sorry for the ignorance, but I barely understand anything for what you sent.
I want to test out my CPU as I have reason to believe it's causing BSODs and crashes in some apps and games. I need "evidence" that there's an issue with the CPU/Mobo/RAM to go get my warranty working...

Would you mind telling me which tests to run?
It's indeed a 6 core 12 threads CPU.

A few weeks ago I ran Small FFTs but saw no errors after about 3-4 hours. This time I'm running large FFTs because I read online it involves the RAM as well, which I want to test even though Memtest86 found no issues after multiple tests.

This is the test I ran:
Click image for larger version

Name:	prime95_9ICoetjHr1.png
Views:	16
Size:	12.6 KB
ID:	23276

Quote:
Originally Posted by Prime95 View Post
This is a stress test. More workers = more stress.

Most stress test errors are RAM related. Try overvolting your RAM or underclocking your RAM.
I will increase my voltage to 1.35v, even though everything is stock? I'm not overclocking anything (no CPU, RAM, GPU overclocking whatsoever).
This might indicate a RAM issue? Should I run it with one stick at a time maybe? I ran multiple Memtest86, Testmem5 and HCI Design Memtest and found no errors in my RAM in any test.
By the way the RAM is at stock 2133mhz.

Last fiddled with by Beinish on 2020-09-06 at 17:20
Beinish is offline   Reply With Quote
Old 2020-09-06, 17:21   #5
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

157438 Posts
Default

Quote:
Originally Posted by Beinish View Post
I will increase my voltage to 1.35v, even though everything is stock? I'm not overclocking anything (no CPU, RAM, GPU overclocking whatsoever).
This might indicate a RAM issue? Should I run it with one stick at a time maybe? I ran multiple Memtest86, Testmem5 and HCI Design Memtest and found no errors in my RAM in any test.
Don't put too much faith in "stock". It is not at all uncommon for memory to have trouble running at stock speed.

So, yes try overvolting to 1.35 (or increase in 0.05 increments). If uncomfortable wit overvolting, reduce the RAM speed and re-test.

Do not remove a mem stick. That could fix the problem, but is unhelpful in diagnosing the culprit.

Memtest does not find all memory errors. No test is perfect.
Prime95 is offline   Reply With Quote
Old 2020-09-06, 17:27   #6
Beinish
 
Sep 2020

19 Posts
Default

Quote:
Originally Posted by Prime95 View Post
Don't put too much faith in "stock". It is not at all uncommon for memory to have trouble running at stock speed.

So, yes try overvolting to 1.35 (or increase in 0.05 increments). If uncomfortable wit overvolting, reduce the RAM speed and re-test.

Do not remove a mem stick. That could fix the problem, but is unhelpful in diagnosing the culprit.

Memtest does not find all memory errors. No test is perfect.
Gotcha.
I set the voltage at 1.250 as to stocks 1.200. When I started the test it still errored our on worked #1 and #2 (same workers as before) but this time it took a bit more time to show the error (like 5 seconds as opposed to 1 second if that matters).

Quote:
[Sun Sep 6 20:24:07 2020]
FATAL ERROR: Rounding was 2.216375913e+14, expected less than 0.4
Hardware failure detected, consult stress.txt file.
FATAL ERROR: Rounding was 4.405092638e+18, expected less than 0.4
Hardware failure detected, consult stress.txt file.
Should I continue adding voltage by 0.05 and testing every time or should I let it keep running and then do it? Also, where do I stop and at what point can I start believing the CPU/RAM is faulty?

Thank you for the help sir.
Beinish is offline   Reply With Quote
Old 2020-09-06, 17:31   #7
moebius
 
moebius's Avatar
 
Jul 2009
Germany

18C16 Posts
Default

Quote Kriesel: LL testing perform better with no HT use, generally.
For Ryzen CPU's it depends on the FFT Size, better performing with no HT at the wavefront f.e 5,5 FFT, better performing with HT at f.e. 2,75M FFT.

Quote Beinish: found an error and stopped after a few seconds:
Is the RAM recommended by the motherboard manufacturer (in the list on the manufacturer's website)?
Are you using the corresponding XAMP profile for this RAM in the BIOS?

Last fiddled with by moebius on 2020-09-06 at 17:31
moebius is online now   Reply With Quote
Old 2020-09-06, 17:39   #8
Beinish
 
Sep 2020

19 Posts
Default

Quote:
Originally Posted by moebius View Post
Quote Kriesel: LL testing perform better with no HT use, generally.
For Ryzen CPU's it depends on the FFT Size, better performing with no HT at the wavefront f.e 5,5 FFT, better performing with HT at f.e. 2,75M FFT.

Quote Beinish: found an error and stopped after a few seconds:
Is the RAM recommended by the motherboard manufacturer (in the list on the manufacturer's website)?
Are you using the corresponding XAMP profile for this RAM in the BIOS?
It is on the recommended list, yes.
I'm not overclocking my RAM but I did try to enable DOCP before. I think the computer didn't boot up, I had to manually change some settings and voltage.
Beinish is offline   Reply With Quote
Old 2020-09-06, 17:49   #9
paulunderwood
 
paulunderwood's Avatar
 
Sep 2002
Database er0rr

2×1,723 Posts
Default

Have you tried running everything at stock except for the RAM at 3000Mhz?
paulunderwood is offline   Reply With Quote
Old 2020-09-06, 17:50   #10
moebius
 
moebius's Avatar
 
Jul 2009
Germany

22·32·11 Posts
Default

Try it several times if the PC with the XAMP / DOCP profile does not start up. For me it is like that, once the PC has started, it is also Prime-stable and even without crashes.
moebius is online now   Reply With Quote
Old 2020-09-06, 18:19   #11
Beinish
 
Sep 2020

19 Posts
Default

Quote:
Originally Posted by paulunderwood View Post
Have you tried running everything at stock except for the RAM at 3000Mhz?
No, but after you wrote that I did and still the same result. first two testers fail instantly and the 3rd one (#12 I believe) fails after 15-20 minutes or so.

Quote:
Originally Posted by moebius View Post
Try it several times if the PC with the XAMP / DOCP profile does not start up. For me it is like that, once the PC has started, it is also Prime-stable and even without crashes.
Can't. When I enable DOCP the computer boots, shuts down, boots again, shuts down again and when it finally boots up, it's on stock settings (2133mhz). If I go back to the BIOS, I see it's on the same settings I left them on (DOCP, so 3200mhz, 1.35v dram, etc), but it's as if the settings don't apply after the finally boot. Sometimes it doesn't even boot at all and I need to reset the CMOS to boot up again.

Edit:
Just successfully enabled DOCP, no issues but the stress test failed anyway on the same workers.

Last fiddled with by Beinish on 2020-09-06 at 18:39
Beinish is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Hardware errors have occurred during the test! rgirard1 Software 16 2020-09-10 22:16
I found the primality test, there seems to be no composite numbers that pass the test sweety439 sweety439 7 2020-02-11 19:49
2+ fast errors during torture test Kai.s0r Hardware 5 2010-08-06 00:49
LLR test starting values kuratkull Information & Answers 6 2008-03-05 13:02
Errors during Torture Test sjhanson Hardware 20 2003-02-02 23:28

All times are UTC. The time now is 02:54.

Thu Oct 29 02:54:49 UTC 2020 up 49 days, 5 mins, 1 user, load averages: 1.21, 1.45, 1.54

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.