Prime95 version 29.3

[Prime95]

Originally Posted by pepi37

At the end of PRP test always was We4, but now is Wf4

That is by design. The "e" and "f" in We4 and Wf4 is a version number (the 4 is an OS code). This is needed for results that are reported manually. The version number changed because the format of the 8 character error code is different.

[ATH]

Originally Posted by Prime95

EnableSetAffinity=0 disables all attempts at setting affinity. Essentially this is equivalent to "run on any core".

Jacobi error checks are for LL tests. This is implemented 29.3.

Gerbicz error checks are for PRP tests. This is not implemented yet.

Yeah, in my haste I mixed up the 2 latest ideas, brain freeze.

[bayanne]

Version 29.3 build 3 throws up the error message:

Dyld Error Message:
Library not loaded: /usr/local/lib/libgmp.10.dylib
Referenced from: /Applications/p95v293.MacOSX/Prime95.app/Contents/MacOS/Prime95
Reason: image not found

Where is this file so I can place it in the correct folder, or am I missing something obvious?

Date/Time: 2017-09-13 13:28:13.434 +0100
OS Version: Mac OS X 10.11.6 (15G1611)

[bayanne]

Originally Posted by bayanne

Version 29.3 build 3 throws up the error message:

Dyld Error Message:
Library not loaded: /usr/local/lib/libgmp.10.dylib
Referenced from: /Applications/p95v293.MacOSX/Prime95.app/Contents/MacOS/Prime95
Reason: image not found

Where is this file so I can place it in the correct folder, or am I missing something obvious?

Date/Time: 2017-09-13 13:28:13.434 +0100
OS Version: Mac OS X 10.11.6 (15G1611)

Version 28.9 works fine

[GP2]

Originally Posted by Prime95

... the 12-hour Jacobi timer ...

Sometimes the .bu3 and .bu4 files are separated by a time interval different from 12 hours.

I have about 60 instances running and as of this moment, here are the patterns seen:

* 5 hours: 18:55 and 00:00
* 12 hours: 18:55 and 06:55 or 06:55 and 18:55
* 9 hours: 02:55 and 12:00
* 4 hours: 02:55 and 06:55
* 1 hour: 02:55 and 03:55
* 10½ hours: 02:55 and 13:25 (only one occurrence of this)
* 12 hours: 15:55 and 03:55

The ":55" are approximate, sometimes they are 53 or 54 or 56, etc.

All of these instances were started at nearly the same time (within a minute or two), at about 06:55 on Sep 11.

Nearly all use a DiskWriteTime of 30 minutes, although a few use 10 minutes. Based on the track record of the servers (which have ECC memory), the Jacobi check would not be expected to fail ever, so that would not be the cause of any premature writing of a savefile.

[Prime95]

Originally Posted by bayanne

Version 29.3 build 3 throws up the error message:

Dyld Error Message:
Library not loaded: /usr/local/lib/libgmp.10.dylib
Referenced from: /Applications/p95v293.MacOSX/Prime95.app/Contents/MacOS/Prime95
Reason: image not found

Where is this file so I can place it in the correct folder, or am I missing something obvious?

Date/Time: 2017-09-13 13:28:13.434 +0100
OS Version: Mac OS X 10.11.6 (15G1611)

You are my Mac guinea pig.

The .dylib is included. Try renaming it from libgmp.dylib to libgmp.10.dylib.
If that does not work, try copying it to /usr/local/lib.

Let me know what works so we can update the readme file.

[Prime95]

Originally Posted by GP2

Sometimes the .bu3 and .bu4 files are separated by a time interval different from 12 hours.

Hmmm. Perhaps the workers are stopping and restarting for some reason. This would read a save file and run a Jacobi error check. This could happen if workers are restarted for different day/night memory settings, PauseWhileRunning, etc.

Might that be a plausible explanation?

[GP2]

Originally Posted by Prime95

Hmmm. Perhaps the workers are stopping and restarting for some reason. This would read a save file and run a Jacobi error check. This could happen if workers are restarted for different day/night memory settings, PauseWhileRunning, etc.

Might that be a plausible explanation?

No, mprime only stops/starts when the virtual machine boots up for the first time or is manually rebooted (or if I manually kill the mprime process for some reason, which I haven't done lately). These are cloud servers that do nothing but run mprime. There are no special settings in the configuration files for pausing. All of the instances mentioned here are running LL tests (double check).

They are spot instances, which can be interrupted when spot prices rise and then new instances launch later when spot prices fall again, and resume work on the work directories left behind by the defunct instances. The work directories are on EFS (network file system which persists indefinitely) rather than on local storage or EBS volumes.

Nearly all of these are c4.large instances, and the console shows they launched in late August and have been running ever since. They were manually rebooted at 06:55 on Sept 11, as part of switching them from 29.2 to 29.3. These all have DiskWriteTime of 30 minutes.

Five of them are c4.xlarge instances, which were rebooted at the same time as the other set, but later they were automatically terminated due to spot price fluctuations and then automatically relaunched at 09:35 on Sept 13. However, the timestamps for the .bu3 and .bu4 files actually predate that, and all of them have .bu3 at Sep 12 18:55 and .bu4 at Sep 13 0:00. These have DiskWriteTime of 10 minutes.

[Prime95]

BTW, you can turn off Jacobi checking, see undoc.txt. This might be a good idea for highly-reliable Amazon servers.

[Prime95]

Originally Posted by GP2

they were automatically terminated due to spot price fluctuations and then automatically relaunched

Does relaunching read from a save file? If so, prime95 runs a Jacobi check and will let that save file become a .bu3 or .bu4.

[Dubslow]

Originally Posted by Prime95

BTW, you can turn off Jacobi checking, see undoc.txt. This might be a good idea for highly-reliable Amazon servers.

Is it adjustable in frequency? For instance I have confidence in my own hardware so once every 12 hours seems excessive to me, but for instance once every 5 or 7 days sounds like a good tradeoff to me, and better than just leaving it off.