Prime95 version 29.3

[James Heinrich]

Diverging from the current theme here, I notice Prime95 now (presumably since I upgraded to v29.3, but I didn't really pay attention) outputs spurious datestamps at 21-hour intervals throughout the day, always 11.5 minutes after the hour:

[Mon Sep 25 14:11:31 2017]
[Wed Sep 27 08:11:30 2017]
[Thu Sep 28 05:11:30 2017]
[Fri Sep 29 02:11:29 2017]
[Fri Sep 29 23:11:30 2017]
[Sun Oct 01 17:11:29 2017]
[Mon Oct 02 14:11:30 2017]
[Tue Oct 03 11:11:30 2017]
[Wed Oct 04 08:11:29 2017]
[Thu Oct 05 05:11:31 2017]
[Fri Oct 06 02:11:30 2017]

It's always a datestamp followed by a blank line, then normal entries resume for the rest of the day until the next spurious datestamp/linefeed.

[rainchill]

I don't like how the new 29.3 error checking always outputs a bunch of extra lines into the worker windows. Could they go in the main communication window or an option be added to not include them at all?

[Prime95]

Originally Posted by rainchill

I don't like how the new 29.3 error checking always outputs a bunch of extra lines into the worker windows. Could they go in the main communication window or an option be added to not include them at all?

An option to disable Jacobi error-checking output will be added to 29.4

[rainchill]

Thanks George, I don't mind the error checking and feel it's probably best to leave it on. I just want the messages to not display unless there is an error.

[storm5510]

Originally Posted by James Heinrich

Diverging from the current theme here, I notice Prime95 now (presumably since I upgraded to v29.3, but I didn't really pay attention) outputs spurious datestamps at 21-hour intervals throughout the day, always 11.5 minutes after the hour:

[Mon Sep 25 14:11:31 2017]
[Wed Sep 27 08:11:30 2017]
[Thu Sep 28 05:11:30 2017]
[Fri Sep 29 02:11:29 2017]
[Fri Sep 29 23:11:30 2017]
[Sun Oct 01 17:11:29 2017]
[Mon Oct 02 14:11:30 2017]
[Tue Oct 03 11:11:30 2017]
[Wed Oct 04 08:11:29 2017]
[Thu Oct 05 05:11:31 2017]
[Fri Oct 06 02:11:30 2017]

It's always a datestamp followed by a blank line, then normal entries resume for the rest of the day until the next spurious datestamp/linefeed.

I have tried turning the results date/time stamp off by using this:

ResultsFileTimestampInterval=0

This does not seem to work.

The short paragraph in undoc.txt does not specify where to place this. I added it to prime.txt since this is where "TimeStamp=n" belongs.

[Prime95]

Originally Posted by James Heinrich

I notice Prime95 now outputs spurious datestamps at 21-hour intervals throughout the day, always 11.5 minutes after the hour:

I do not know why this is happening. I'm pretty sure it is due to auto-benchmark. I have a fix for the symptom you are seeing ready for version 29.4.

[Mark Rose]

I wonder if it's possible to run more Jacobi checks automatically?

[Work thread Oct 30 12:31] Iteration: 43070000 / 44924287 [95.87%], ms/iter: 4.109, ETA: 02:06:59
[Work thread Oct 30 12:31] Running Jacobi error check. Failed. Time: 14.096 sec.
[Work thread Oct 30 12:31] Iteration: 43072328/44924287, ERROR: Jacobi error check failed!
[Work thread Oct 30 12:31] Continuing from last save file.
[Work thread Oct 30 12:31] Setting affinity to run helper thread 1 on CPU core #2
[Work thread Oct 30 12:31] Setting affinity to run helper thread 2 on CPU core #3
[Work thread Oct 30 12:31] Setting affinity to run helper thread 3 on CPU core #4
[Work thread Oct 30 12:31] Running Jacobi error check. Failed. Time: 13.829 sec.
[Work thread Oct 30 12:31] Error reading intermediate file: p9J24287
[Work thread Oct 30 12:31] Renaming p9J24287 to p9J24287.bad1
[Work thread Oct 30 12:31] Trying backup intermediate file: p9J24287.bu
[Work thread Oct 30 12:31] Running Jacobi error check. Failed. Time: 13.966 sec.
[Work thread Oct 30 12:31] Error reading intermediate file: p9J24287.bu
[Work thread Oct 30 12:31] Renaming p9J24287.bu to p9J24287.bad2
[Work thread Oct 30 12:31] Trying backup intermediate file: p9J24287.bu2
[Work thread Oct 30 12:31] Running Jacobi error check. Failed. Time: 13.984 sec.
[Work thread Oct 30 12:32] Error reading intermediate file: p9J24287.bu2
[Work thread Oct 30 12:32] Renaming p9J24287.bu2 to p9J24287.bad3
[Work thread Oct 30 12:32] All intermediate files bad. Temporarily abandoning work unit.
[Work thread Oct 30 12:32] Setting affinity to run helper thread 1 on CPU core #2
[Work thread Oct 30 12:32] Setting affinity to run helper thread 2 on CPU core #3
[Work thread Oct 30 12:32] Setting affinity to run helper thread 3 on CPU core #4
[Work thread Oct 30 12:32] Starting primality test of M44940079 using FMA3 FFT length 2400K, Pass1=320, Pass2=7680, clm=4, 4 threads
[Work thread Oct 30 12:32] Iteration: 10000 / 44940079 [0.02%], ms/iter: 4.111, ETA: 51:18:05

Kind of sad to waste so much work. If a machine turns in a bad result or fails a Jacobi error check, might it make sense to set JacobiErrorCheckingInterval=1 automatically?

This machine went flaky once before and I hope I can resurrect it once again.

[Prime95]

Originally Posted by Mark Rose

I wonder if it's possible to run more Jacobi checks automatically?

Yes, see undoc.txt.

I'm surprised there wasn't a usable save file. pxxxxxxx, .bu, and .bu2 are the last 3 save files created at 30 minute intervals. Not surprising they were bad. .bu3 and .bu4 should have been created at the 12-hour Jacobi intervals. Any clues why .bu3 and .bu4 are not being created????

[Mark Rose]

Originally Posted by Prime95

Yes, see undoc.txt.

Yeah, I changed it to every hour after reading undoc.txt. But it may be a good idea to have it do that automatically.

I'm surprised there wasn't a usable save file. pxxxxxxx, .bu, and .bu2 are the last 3 save files created at 30 minute intervals. Not surprising they were bad. .bu3 and .bu4 should have been created at the 12-hour Jacobi intervals. Any clues why .bu3 and .bu4 are not being created????

I have no idea. I have no settings in prime.txt or local.txt related to backups, other than JacobiErrorCheckingInterval=1 which I added a couple of hours ago.

The machine has been working on exponents in the 45M range and it takes it about 48 hours to finish an LL test in that range.

I've removed half the RAM and reseated the last two sticks, so I may or may not be able to reproduce the results.

[Prime95]

Is anyone else seeing a lack of .bu3 and .bu4 files?

[GP2]

Originally Posted by Prime95

Is anyone else seeing a lack of .bu3 and .bu4 files?

No, I have lots of them.