mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Information & Answers (https://www.mersenneforum.org/forumdisplay.php?f=38)
-   -   How do I get the ~133 mb proof file delivered to GIMPS in the manual environment? (https://www.mersenneforum.org/showthread.php?t=26191)

tsc 2020-11-14 22:20

How do I get the ~133 mb proof file delivered to GIMPS in the manual environment?
 
Got this error message today:

=================================================================
Your software is outdated.
Your L-L result has been accepted for now, but please upgrade your software to at least v30.3 as soon as possible.
=================================================================

The sent json file looks like this:

=================================================================
[Wed Nov 11 08:37:23 2020]
{"status":"C", "exponent":105872743, "worktype":"LL", "res64":"hex", "fft-length":5734400, "shift-count":18291950, "error-code":"00000000", "security-code":"hex", "program":{"name":"Prime95",[SIZE="5"]"version":"30.3"[/SIZE] "build":6, "port":4}, "timestamp":"2020-11-11 07:37:23", "user":"tsc", "computer":"w10", "aid":"hex"}
=================================================================

The LL in question was finished using ver. 30.3. (3/4 was done by a version 29.8).
So maybe the error is a little overkill.

I discarded the LL's I had in pipeline and replaced it with PRP's.
Maybe this is no problem for internet connected clients, but it could be a problem for manual assignments. Without my intervention I could have done 4 LL's more.


I wonder how do I get the ~133 mb proof file delivered to GIMPS in the manual environment?
On the page [url]https://www.mersenne.org/manual_result/[/url]
the limit is ~2 mb.


br tsc


tnx to James Heinrich for solving the P-1 problem I had some weeks ago.

kriesel 2020-11-15 11:15

[url]https://www.mersenneforum.org/showpost.php?p=553120&postcount=26[/url]

tsc 2020-11-15 20:11

windows upload tool
 
tnx!
Slightly more complicated than just downloading a prime95 version. :-)
Took some reading before I found the windows upload tool, which is straight forward.
I look forward to use the tool.
I wont miss the DC work.
Any estimation to how many years DC's will still be executed?
Has a DC ever found a prime?
br tsc

Uncwilly 2020-11-15 22:37

[QUOTE=tsc;563311]Any estimation to how many years DC's will still be executed?
Has a DC ever found a prime?[/QUOTE]It will be years (12.25 or so) before DC catches up to the area that has the PRP's with certs.

This is based upon adding up all of the values in the [C]Status LL/PRP[/C] column up to the 109M level and dividing it by the DC rate from July capture of [url]http://hoegge.dk/mersenne/GIMPSstats.html[/url] at archive.org (DC's completed in the last 365) doing the maths.

No Prime has been found in the computer era by DC.
There may have been in the pencil and paper era.

Prime95 2020-11-16 00:36

[QUOTE=Uncwilly;563345]No Prime has been found in the computer era by DC.
There may have been in the pencil and paper era.[/QUOTE]

Luke Welsh might disagree.

Uncwilly 2020-11-16 01:12

One of the documents referenced about the discovery (on wikipedia) is not available (even on the wayback machine.) I had done articles for the Mersenne wiki, but did not do that one. Didn't Slowinski skip some exponents, and thus missed M29? If skipped that was not a DC. Out of order sure.

Prime95 2020-11-16 04:14

[QUOTE=Uncwilly;563354]One of the documents referenced about the discovery (on wikipedia) is not available (even on the wayback machine.) I had done articles for the Mersenne wiki, but did not do that one. Didn't Slowinski skip some exponents, and thus missed M29? If skipped that was not a DC. Out of order sure.[/QUOTE]

I don't believe Slowinski definitively stated the exponent was skipped vs. tested incorrectly. When dealing with him back in the day he was very reluctant to share residues for fear of bad PR from a "Cray supercomputer produces incorrect results" story. Thus, the vague "skipped some exponents" cover story.

That said, I have no knowledge one way way or the other.

tsc 2020-11-20 20:10

I have this fine proof for GIMPS:
2020-11-18 15:21 132.833.108 p106266437.proof

The upload.exe program seems to be written for x64.
My uploader machine is a win7 32-bit.

Any suggestions?

br tsc

Prime95 2020-11-20 21:16

Why can't prime95 send the proof? If the file is in the directory, prime95 should upload it.

tsc 2020-11-23 23:23

[QUOTE=Prime95;563876]Why can't prime95 send the proof? If the file is in the directory, prime95 should upload it.[/QUOTE]

It did.
I delete and forget about the upload tool suggested.

Shouldnt PRP be more rewarded than just a LL?
Somebody might give a damn about the difference. LL is fastest.
But not the best choice for the project.
Besides it would look better at the workload page if PRP will be rewarded accordingly.
Rewarded equally we might get many more DC's to deal with in the future.

Will LL expire?
Will PRP be given slightly more reward?

br tsc

Uncwilly 2020-11-23 23:34

PRP and LL get rewarded for the amount of work they require.
PRP takes a little more work on the same exponent, because of the tiny bit of error checking and the little bit of work for proof generation. Proof generation and running of the Cert eliminate the need for a DC. So PRP with proofs are vastly preferred over LL now, for first time checks. LL should be used to DC numbers that had a first time check with LL.

Ideally by the time all of the DC's that currently need to be done are done, no more LL's will be done for first time checks. So it will be phased out for normal first time checks. It is likely that the server will stop handing out LL for FTC and only hand it out for DCs.


All times are UTC. The time now is 18:21.

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