mersenneforum.org  

Go Back   mersenneforum.org > New To GIMPS? Start Here! > Information & Answers

Reply
 
Thread Tools
Old 2021-01-03, 18:39   #1
Bitter
 
Feb 2019

2·5 Posts
Default Prime95 v 30.3b6 64bit enable more verbose logging?

I'm trying to pin down some issues in a computer and prime95 has been great at causing the failure but I haven't had a detailed enough log to know which worker is failing each time. What's happening is either the whole PC reboots with a windows event error about bug check or prime95 crashes and closes. Either time the results.txt only tells me at what point in testing the problem happened and what the rounding error was, but not which worker had the failure. I would like to know which worker is failing each time to pin down if I have a bad core in the CPU or if it's a more general failure. Only once in many hours of testing have I been present to see a worker fail. It fails at different points and different times so it's my issue, not a Prime95 issue.

How do I get a more verbose log from Prime95 running Windows 10?
Bitter is offline   Reply With Quote
Old 2021-01-03, 20:09   #2
ixfd64
Bemusing Prompter
 
ixfd64's Avatar
 
"Danny"
Dec 2002
California

34·29 Posts
Default

You can set Debug=1 in your prime.txt file, although this is mainly for troubleshooting connection issues. Checking the Event Viewer logs will probably be more helpful.
ixfd64 is offline   Reply With Quote
Old 2021-01-03, 20:12   #3
Bitter
 
Feb 2019

2×5 Posts
Default

Quote:
Originally Posted by ixfd64 View Post
You can set Debug=1 in your prime.txt file, although this is mainly for troubleshooting connection issues. Checking the Event Viewer logs will probably be more helpful.
Event viewer has only informed me that it wasn't a planned reboot and that I failed a bug check, pretty generic faults that don't point me to which worker thread failed. I was hoping if it was the same worker each time it could indicate a specific CPU core that was at fault and warranty the CPU.



I'll try setting the debug switch anyway. Are there any other ways to log or to incrementally save the text output in the Prime95 worker window?
Bitter is offline   Reply With Quote
Old 2021-01-03, 20:40   #4
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

730110 Posts
Default

Quote:
Originally Posted by Bitter View Post
Are there any other ways to log or to incrementally save the text output in the Prime95 worker window?
I'm afraid you are out of luck.

FWIW, a bad CPU core is really rare. The most common culprit is RAM. Try a small FFT torture test (more CPU-only testing, little RAM testing).
Prime95 is online now   Reply With Quote
Old 2021-01-04, 03:10   #5
Bitter
 
Feb 2019

2×5 Posts
Default

Quote:
Originally Posted by Prime95 View Post
I'm afraid you are out of luck.

FWIW, a bad CPU core is really rare. The most common culprit is RAM. Try a small FFT torture test (more CPU-only testing, little RAM testing).
I know it's rare but with how Ryzen boosts it's possible that one isn't stable at boost speeds I suppose. It runs small FFT without fail, pretty sure the whole thing fits in the 16MB L3 cache so no RAM is tested. I have a smaller set of RAM that's the same speed to swap in for overnight testing I'm going to try out but I have a funky feeling about the board/cpu since I noticed more than normal vdroop on the SOC voltage readings and it doesn't like the post after changing that voltage up to compensate for the droop, sometimes beeping out a memory error code sometimes just not posting at all.
Bitter is offline   Reply With Quote
Old 2021-01-04, 03:34   #6
VBCurtis
 
VBCurtis's Avatar
 
"Curtis"
Feb 2005
Riverside, CA

29·32 Posts
Default

If it's passing a test that fits in cache, but fails work that doesn't fit in cache, you have a really strong indication that RAM is not good while CPU is good.
It's possible the motherboard is bad, too- there is such a thing as a RAM channel being bad on a board, for instance.
VBCurtis is offline   Reply With Quote
Old 2021-01-04, 06:18   #7
retina
Undefined
 
retina's Avatar
 
"The unspeakable one"
Jun 2006
My evil lair

2×34×37 Posts
Default

Quote:
Originally Posted by Bitter View Post
Event viewer has only informed me that it wasn't a planned reboot and that I failed a bug check, pretty generic faults ...
At the risk of simply reiterating what the previous posters said: Check your RAM. Or perhaps if you are lucky, your RAM timing settings. Slow it down, get it stable, then bring it up slowly till it fails, back it off a small amount, and done.
retina is online now   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
GMP-ECM with --enable-openmp flag set in configure = bad results? GP2 GMP-ECM 3 2016-10-16 10:21
Network logging? Xyzzy Linux 10 2015-09-25 17:55
Strange question on logging out Brian-E Forum Feedback 3 2007-12-14 13:18
Running mprime in verbose mode under linux? linux newbie Software 4 2006-03-21 16:13
HOWTO: enable the HLT command on nforce2 chipsets with freeware utils! graysky Hardware 4 2004-10-17 10:23

All times are UTC. The time now is 06:47.

Tue Jan 19 06:47:11 UTC 2021 up 47 days, 2:58, 0 users, load averages: 1.58, 1.57, 1.66

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, 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.