LLR Crashing

[mdettweiler]

I'm currently at about 90% on my range--but I've noticed that ever since about 85%, LLR has been crashing every once in a while on me. I'll get this notice from Windows that LLR.exe encountered a problem and was shut down--and sometimes, that can be disastrous when it comes to lost CPU time. For example: I was out of the house most of today, but apparently, not long after I left, LLR crashed, and when I came back, 5 hours and 45 minutes of CPU time had been wasted, according to Task Manager's "system idle process" CPU time count.

I know my computer isn't cranking out bad residuals--some of the BOINC projects I participate in (on the one core, while LLR uses the other) are based on the standard LLR.exe with a wrapper, and I haven't had any problems with them, and the results are always matching up with the other systems that are also assigned to crunch those workunits--so at least I can trust my residuals. However, I am quite befuddled how LLR could run without a hitch all the way up to 85% through my range, then all of a sudden have all these crashes. I've sometimes tried deleting the "z*****" file from my LLR directory, which causes LLR to start over testing the number it was working on (in case the crashes were the fault of the checkpoint file getting corrupted), but it still crashes.

I've noticed that this has mainly happened ever since one particular BSOD my system had a few days ago; could that be part of the problem?

Does anyone know what might be going on, and possibly a fix, besides starting up LLR again manually each time it crashes? (I have LLR in my startup folder set to run minimized, FYI.)

[VBCurtis]

This, or something similar, has happened to a few of us this year. I found that one copy of LLR would crash repeatedly on a specific file; I cut and pasted the work from inside the file into another sieve file, moved it into another folder containing LLR, and ran it without issue.

I do not claim this process will fix your problem, but somehow it fixed mine. I suggest you download a fresh LLR (or just re-unpack it from the zip) into a new folder not contained in the old LLR folder, move the work file over, test, and report if it fixed your problem.

-Curtis

[mdettweiler]

Originally Posted by VBCurtis

This, or something similar, has happened to a few of us this year. I found that one copy of LLR would crash repeatedly on a specific file; I cut and pasted the work from inside the file into another sieve file, moved it into another folder containing LLR, and ran it without issue.

I do not claim this process will fix your problem, but somehow it fixed mine. I suggest you download a fresh LLR (or just re-unpack it from the zip) into a new folder not contained in the old LLR folder, move the work file over, test, and report if it fixed your problem.

-Curtis

Okay, that's what I did. I downloaded LLR again, unzipped it into a new folder, and copied all the text files (650to655.txt, lresults.txt, and primes.txt) and llr.ini over by opening them, copying their contents to the clipboard, and pasting them in a new file of identical name in the new folder. (I left the z**** file, though.) I started up the new copy of LLR and it seems to be working fine, and has picked up where the old one left off. I'll report here if I find any more problems.

Thanks!

[mdettweiler]

Originally Posted by VBCurtis

This, or something similar, has happened to a few of us this year. I found that one copy of LLR would crash repeatedly on a specific file; I cut and pasted the work from inside the file into another sieve file, moved it into another folder containing LLR, and ran it without issue.

I do not claim this process will fix your problem, but somehow it fixed mine. I suggest you download a fresh LLR (or just re-unpack it from the zip) into a new folder not contained in the old LLR folder, move the work file over, test, and report if it fixed your problem.

-Curtis

It seemed to fix the problem for a little while (about a day or two) but now it seems to be coming back. I was doing some P-1 work for Riesel Sieve using Prime95 (which seems to be coded similarly, at least on the outside, to LLR), and I was getting the same problem. Even when I tried some LLRNet with the Sierpinski/Riesel Base 5 project, it was crashing every once in a while--at least once, if not twice, in a day. I tried the trick that you had mentioned, and it seemed to help for a while, but it didn't seem to fix it permanently. And the odd thing is that this is not limited to LLR, nor is it limited to the k=105 range that I was doing before; it even showed up wiht a BOINC LLR workunit that I was doing for Riesel Sieve, crashing after about 15 minutes of crunching.

I think I'm still producing good residuals--when the problem first showed up, I was doing LLR work for Primegrid that was all fine and dandy (and it didn't crash at all); but now it's almost like the problem has spread all over my system, even to Windows, which seems to be giving me more BSOD's then usual.

Does anyone have the slightest idea what my problem might be? These crashes are becoming a real bear, especially when they happen while I'm not at home (and they seem to do so almost every time I'm away)--and they sometimes cause 4-5 hours of CPU time, sometimes more, to be wasted.

If it helps at all, I haven't noticed this problem happening with any application other than LLR, Prime95, or LLRNet (the BOINC LLR application is actually just LLR.exe with a wrapper). I hadn't had any problems with any of these applications for the longest time, but now they're all starting to crash on me.

[paulunderwood]

Sounds like bad hardware. Please post all the "health status" readings from your BIOS.

Check the cpu fan and heatsink are free of dust.

Run memtest86+ to check you memory is ok.

[mdettweiler]

Originally Posted by paulunderwood

Sounds like bad hardware. Please post all the "health status" readings from your BIOS.

Check the cpu fan and heatsink are free of dust.

Run memtest86+ to check you memory is ok.

Okay. Where do I find health status readings from my BIOS? Also, where would I find a copy of memtest86+?

I'll check out my CPU fan and heatsink sometime ASAP, too. Thanks!

[paulunderwood]

It depends on your motherboard. Usually either the [del] or [F2] key needs to pressed as your machine boots -- just after the keyboard light flashes or comes on. Navigate the menus, not changing anything, read the health status and post it here for us to see. Press (usually) [F10] to exit the bios -- or use the menus.

Here is the memtest86+ wiki

You will have to be careful of static electricity when cleaning the heatsink and fan. I use an artist's brush and vacuum cleaner. Others recommend using a can of compressed air (from your local computer shop) or even a leaf blower

ps. The health reading that would be helpful for me are:

• all voltage readings
• all fan speeds
• all temperature readings

[mdettweiler]

Originally Posted by paulunderwood

It depends on your motherboard. Usually either the [del] or [F2] key needs to pressed as your machine boots -- just after the keyboard light flashes or comes on. Navigate the menus, not changing anything, read the health status and post it here for us to see. Press (usually) [F10] to exit the bios -- or use the menus.

Here is the memtest86+ wiki

You will have to be careful of static electricity when cleaning the heatsink and fan. I use an artist's brush and vacuum cleaner. Others recommend using a can of compressed air (from your local computer shop) or even a leaf blower

ps. The health reading that would be helpful for me are:

• all voltage readings
• all fan speeds
• all temperature readings

Okay, thanks. Actually, soon after I wrote asking where to find memtest86+, I found it on my own. Sorry if I caused you any unnecessary trouble.

As for the health readings, I can give you the temperature readings right off the bat, since I have a temperature readout on the front of my case. The CPU is 33 degrees Celsius (w/both hyperthreads busy running distributed computing apps), and the temperature inside the case is 27 degrees Celsius.

As for the voltage and fan readings, I've looked in the BIOS settings of my motherboard before, and haven't noticed naything remotely near voltage and fan readouts. I'll check it out later on, though, hopefully sometime today. And while I'm at it, I'll run memtest86 and then clean out my computer's various fans with a small computer vacuum that I've got.

I'll post back here after I've done so if either a)I can find the readouts for voltage and fan speed; b)The memtest86 program turns up any errors.

Oh, and by the way, I was running the Prime95 stress test on my system for a bit, and would you know, Prime95 crashed while running the stress test! So, I'll just have to try cleaning out the fans and running the memory test.

[paulunderwood]

As for the health readings, I can give you the temperature readings right off the bat, since I have a temperature readout on the front of my case. The CPU is 33 degrees Celsius (w/both hyperthreads busy running distributed computing apps), and the temperature inside the case is 27 degrees Celsius.

Those temperatures are cool I guess the fans are doing their job.

That leaves incorrect voltage and faulty memory to rule out as the cause of the problem... or it could be something else

[mdettweiler]

Originally Posted by paulunderwood

Those temperatures are cool I guess the fans are doing their job.

That leaves incorrect voltage and faulty memory to rule out as the cause of the problem... or it could be something else

Okay, so at least I know the fans are working properly. Would that rule out the possibility of dust in the fans causing the problem?

[paulunderwood]

I guess: yes