Prime95 v30.3

[Prime95]

Just built 32-bit versions (which I hope nobody is using!) and 64-bit NT service version.


This is the first time I've built these since my Mac died and I'm now on Windows with a new Visual Studio compiler.


A MacOSX port will be difficult. I'll have to steal the wife's computer.

[henryzz]

I have split the publication discussion into a new thread in the Math forum

[ric]

I've joined the v30 beta testing a few days ago, with v30.3 b2 on a colab instance, working on PRP-CF candidates. So far, so good, apart from the last candidate who has entered an endless loop of

[Work thread Aug 20 10:55] Iteration: 10488602 / 10489856 [99.98%].
[Work thread Aug 20 10:55] Gerbicz error check passed at iteration 10489826.
[Work thread Aug 20 10:55] Generating proof for M10489607.  Proof power = 8, Hash length = 64
[Work thread Aug 20 10:55] Root hash = D9FEACD9545125105C0A88E19768DF99EA43BF97B183A15F42454E8569216911
[Work thread Aug 20 10:55] hash0 = 772F19B5FDFDB1A5
[Work thread Aug 20 10:55] hash1 = FA82BB6558F585A1
[Work thread Aug 20 10:55] hash2 = FBA62EA2942D2DCF
[Work thread Aug 20 10:55] hash3 = C35FC946454B1C33
[Work thread Aug 20 10:55] hash4 = BABA2FED3B34B207
[Work thread Aug 20 10:55] hash5 = 8CCFECCFA556DE29
[Work thread Aug 20 10:56] MD5 error reading PRP proof interim residues file.
[Work thread Aug 20 10:56] Waiting 5 minutes to try proof generation again.
[Work thread Aug 20 10:56] Waiting five minutes before restarting.

probably because the colab session expired a few iterations before completion, and maybe left some file in an inconsistent state.

results.txt contains a bunch of

MD5 error reading PRP proof interim residues file.

every 5 minutes for the last couple hours. There's no entry in results.json.txt so far for that exponent.

How do I get out of here? A restart from scratch?

TIA

[kriesel]

Recently got a TF assignment listed as assigned in a colab mprime instance, unexpectedly, in .https://www.mersenne.org/workload/.

Factor=(AID),242692117,71,72

I'll probably throw that minnow back. Especially since it does not appear in an actual worktodo file.
The same instance allegedly had been given a PRP assignment that does not appear in worktodo.

PRP=(AID),1,2,98497291,-1,77,0,3,

There's no record of those assignments in prime.log.
Both indicate as assigned 2020-08-19. Had another internet outage yesterday, but I would not expect that to affect colab instance interaction with PrimeNet.

All my colab mprime work preferences are set to 150 (PRP).

Originally Posted by Prime95

Just built 32-bit versions (which I hope nobody is using!)

Not currently, but might fire up some smaller silicon space heaters with whatever is the current 32bit version in a few months. Certs might be just the right size for those.

[Prime95]

Originally Posted by ric

How do I get out of here? A restart from scratch?

Set MaxProofgenWaits=1 in prime.txt

The default is to try proof generation for 2 days before giving up (in case the file is on an NFS mount that is offline).
It is on my to-do list to tailor the wait time to the error. A file open error might wait two days, while your MD5 error is
unlikely to get better - maybe try just twice.

[Awesomeotts]

Originally Posted by kriesel

It will take a long time to get the bulk of the clients updated. Early adopters of prime95/mprime v30.x are bearing the brunt of CERT for both mprime/prime95 and gpuowl production. (Either curtisc or Ben Delo updating a fraction of their fleet would help a lot. But like for everyone in this all-volunteer project, their kit, their call. And if they had started already, we wouldn't know without doing some checking.)

I think Ben Delo has, https://www.mersenne.org/report_expo...8664553&full=1

[pinhodecarlos]

Got a new MD5 error: MD5 of downloaded starting value doesn’t match, will retry later blablabla, and then aborting. Using latest version of prime95.

[Uncwilly]

I am seeing something that looks odd. Every hour Prime95 stops a worker and restarts "to do priority work". That would be great if there was any in the queue, but there isn't.
It is only checking in with Primenet for priority work (cert) once every 6 hours.

[Aug 21 05:50] Iteration: 22461840 / xxxxxxxx [42.52%], ms/iter:  7.053, ETA: 59:29:13
[Aug 21 06:03] Restarting worker to do priority work.
[Aug 21 06:03] Stopping primality test of Mxxxxxxxx at iteration 22574063 [42.73%]
[Aug 21 06:03] Setting affinity to run helper thread 1 on CPU core #2
[Aug 21 06:03] Setting affinity to run helper thread 3 on CPU core #4
[Aug 21 06:03] Setting affinity to run helper thread 2 on CPU core #3
[Aug 21 06:03] Running Jacobi error check.  Passed.  Time: 14.649 sec.
[Aug 21 06:04] Resuming primality test of Mxxxxxxxx using FMA3 FFT length 2800K, Pass1=448, Pass2=6400, clm=1, 4 threads
[Aug 21 06:04] Iteration: 22574064 / xxxxxxxx [42.73%].
[Aug 21 06:09] Iteration: 22617825 / xxxxxxxx [42.81%], ms/iter:  7.036, ETA: 59:02:12
[Aug 21 06:27] Iteration: 22773810 / xxxxxxxx [43.11%], ms/iter:  7.054, ETA: 58:52:35
[Aug 21 06:47] Iteration: 22929795 / xxxxxxxx [43.40%], ms/iter:  7.672, ETA: 63:42:36
[Aug 21 07:03] Restarting worker to do priority work.
[Aug 21 07:03] Stopping primality test of Mxxxxxxxx at iteration 23051629 [43.63%]
[Aug 21 07:03] Setting affinity to run helper thread 1 on CPU core #2
[Aug 21 07:03] Setting affinity to run helper thread 3 on CPU core #4
[Aug 21 07:03] Setting affinity to run helper thread 2 on CPU core #3
[Aug 21 07:03] Running Jacobi error check.  Passed.  Time: 15.611 sec.
[Aug 21 07:04] Resuming primality test of Mxxxxxxxx using FMA3 FFT length 2800K, Pass1=448, Pass2=6400, clm=1, 4 threads
[Aug 21 07:04] Iteration: 23051630 / xxxxxxxx [43.63%].
[Aug 21 07:08] Iteration: 23085780 / xxxxxxxx [43.70%], ms/iter:  7.368, ETA: 60:51:54
[Aug 21 07:28] Iteration: 23241765 / xxxxxxxx [43.99%], ms/iter:  7.608, ETA: 62:31:10
[Aug 21 07:49] Iteration: 23397750 / xxxxxxxx [44.29%], ms/iter:  8.068, ETA: 65:56:52
[Aug 21 07:58] Running Jacobi error check.  Passed.  Time: 14.077 sec.
[Aug 21 08:03] Restarting worker to do priority work.
[Aug 21 08:03] Stopping primality test of Mxxxxxxxx at iteration 23510657 [44.50%]
[Aug 21 08:03] Setting affinity to run helper thread 1 on CPU core #2
[Aug 21 08:03] Setting affinity to run helper thread 3 on CPU core #4
[Aug 21 08:03] Setting affinity to run helper thread 2 on CPU core #3
[Aug 21 08:03] Running Jacobi error check.  Passed.  Time: 14.755 sec.
[Aug 21 08:04] Resuming primality test of Mxxxxxxxx using FMA3 FFT length 2800K, Pass1=448, Pass2=6400, clm=1, 4 threads
[Aug 21 08:04] Iteration: 23510658 / xxxxxxxx [44.50%].
[Aug 21 08:09] Iteration: 23553735 / xxxxxxxx [44.58%], ms/iter:  7.748, ETA: 62:59:48
[Aug 21 08:28] Iteration: 23709720 / xxxxxxxx [44.88%], ms/iter:  7.344, ETA: 59:23:21
[Aug 21 08:47] Iteration: 23865705 / xxxxxxxx [45.18%], ms/iter:  7.276, ETA: 58:31:44
[Aug 21 09:03] Restarting worker to do priority work.
[Aug 21 09:03] Stopping primality test of Mxxxxxxxx at iteration 23999395 [45.43%]
[Aug 21 09:03] Setting affinity to run helper thread 1 on CPU core #2
[Aug 21 09:03] Setting affinity to run helper thread 3 on CPU core #4
[Aug 21 09:03] Setting affinity to run helper thread 2 on CPU core #3
[Aug 21 09:03] Running Jacobi error check.  Passed.  Time: 22.355 sec.
[Aug 21 09:04] Resuming primality test of Mxxxxxxxx using FMA3 FFT length 2800K, Pass1=448, Pass2=6400, clm=1, 4 threads
[Aug 21 09:04] Iteration: 23999396 / xxxxxxxx [45.43%].
[Aug 21 09:07] Iteration: 24021690 / xxxxxxxx [45.47%], ms/iter:  7.655, ETA: 61:14:43

[Prime95]

Originally Posted by Uncwilly

I am seeing something that looks odd. Every hour Prime95 stops a worker and restarts "to do priority work". That would be great if there was any in the queue, but there isn't.
It is only checking in with Primenet for priority work (cert) once every 6 hours.

Can you send me the worktodo.txt file? THanks.

[DrobinsonPE]

I am still running 30.3 build 2 on my computers so I need to upgrade them all to build 3 this weekend. Today one of my computers got a MD5 error.

This is when it started:

[Work thread Aug 21 06:50] Restarting worker to do priority work.
[Work thread Aug 21 06:50] Stopping primality test of M54082711 at iteration 39137183 [72.36%]
[Work thread Aug 21 06:50] Setting affinity to run helper thread 1 on CPU core #2
[Work thread Aug 21 06:50] Setting affinity to run helper thread 3 on CPU core #4
[Work thread Aug 21 06:50] Setting affinity to run helper thread 2 on CPU core #3
[Work thread Aug 21 06:50] Starting certification of M1158401 using type-1 FFT length 56K, Pass1=224, Pass2=256, clm=4, 4 threads
[Work thread Aug 21 06:50] MD5 of downloaded starting value does not match.
[Work thread Aug 21 06:50] Aborting processing of this work unit -- will try again later.

Is there a file you want me to send or should I just upgrade to build 3 and start up again to see if that fixes it?

Strange thing is... this exponent does not show up in my assignments list but it does show up in my worktodo.

Cert=UID,1,2,1158401,-1,18101

[Prime95]

Originally Posted by DrobinsonPE

I am still running 30.3 build 2 on my computers so I need to upgrade them all to build 3 this weekend. Today one of my computers got a MD5 error.

This is when it started:

[Work thread Aug 21 06:50] MD5 of downloaded starting value does not match.
[Work thread Aug 21 06:50] Aborting processing of this work unit -- will try again later.

Is there a file you want me to send or should I just upgrade to build 3 and start up again to see if that fixes it?

Strange thing is... this exponent does not show up in my assignments list but it does show up in my worktodo.

Cert=UID,1,2,1158401,-1,18101

Do upgrade to build 3.

The error message is poor. The problem really is that prime95 is unable to get the starting value because the assignment is invalid. I've fixed the error message in build 4.

In build 2, this error will continue forever. In build 3, this error will happen 8 or 10 times and then the assignment will be removed from worktodo.txt.

I don't know why the assignment didn't "stick". It apparently was later assigned to me and I've already finished it.