processed dc and tc posts

[zircon]

Took:
CERT=00000000000000000000000013CD5897,1,2,332224663,-1,1297753

[kriesel]

ETA ~6 hours:
CERT=0000000000000000000000001D070E8F,1,2,487001743,-1,475588

[kriesel]

Originally Posted by Uncwilly

...
PRP=1,2,108913897,-1,76,0 to get Assignment ID use DoubleCheck=108913897,76,1
.[/CODE]
Suggestion: Copy both entries into your Worktodo, comment out the PRP entry by preceding it with a *, then do a coms with PrimeNet to get an AssignmentID(s), copy AID to the PRP entry (between the = and the 1 {using a comma of course}), then remove or comment out the DC line and remove the comment out on the PRP line.

That seems excessively complicated and unnecessary and error-prone.
Instead:
copy the exponent string only.
Go to https://www.mersenne.org/manual_assignment/ and make sure you're logged in;
paste the exponent into both the optional exponent range boxes;
select "Double check using PRP with proof" work type, click "get assignments".
Copy the returned assignment from there, paste into the worktodo.txt of whatever application is being used; prime95, mprime, gpuowl, and run it.
Zero manual edits of the worktodo line means zero manual editing errors.

[kriesel]

Took:
CERT=00000000000000000000000013CD0761,1,2,332203873,-1,1297672
CERT=00000000000000000000000013CD42DD,1,2,332219101,-1,1297731
CERT=000000000000000000000000153BD30D,1,2,356242189,-1,1391572
CERT=0000000000000000000000001615073B,1,2,370476859,-1,1447176

[JHagerson]

Taken:
CERT=00000000000000000000000013CD5883,1,2,332224643,-1,1297753

[kriesel]

Taken:
CERT=00000000000000000000000013CD5AED,1,2,332225261,-1,648878
CERT=00000000000000000000000013CD5B29,1,2,332225321,-1,1297756
CERT=00000000000000000000000013CD5BDF,1,2,332225503,-1,1297756
CERT=00000000000000000000000013CD752F,1,2,332231983,-1,1297782
CERT=00000000000000000000000013CD7953,1,2,332233043,-1,1297786
CERT=00000000000000000000000013CD8E71,1,2,332238449,-1,648904

[kriesel]

And (leaving 3):
CERT=00000000000000000000000013CD8ED9,1,2,332238553,-1,1297807
CERT=00000000000000000000000013CDEE89,1,2,332263049,-1,1297903
CERT=00000000000000000000000013CE7709,1,2,332297993,-1,1298040
CERT=00000000000000000000000013CE7C2D,1,2,332299309,-1,1298045
CERT=00000000000000000000000013CE80D1,1,2,332300497,-1,1298049
CERT=00000000000000000000000013D0344F,1,2,332411983,-1,649243

[kriesel]

Took the p>500M certs; ETAs Mar 2 - 12:
CERT=00000000000000000000000026FB3FBB,1,2,654000059,-1,1277344
CERT=0000000000000000000000002FAF07FF,1,2,799999999,-1,1562500
CERT=00000000000000000000000030DC6895,1,2,819751061,-1,800539
CERT=000000000000000000000000380D6719,1,2,940402457,-1,1836724
CERT=0000000000000000000000003B0244C7,1,2,990004423,-1,966802
CERT=0000000000000000000000003B9AAEA7,1,2,999992999,-1,1953112

[Andrew Usher]

I guess this is as good a thread as any to post this, but I just saw a very remarkable result from the previous poster: C-PRP on exponent 1,000,000,007 (and the exponent column widened automatically to accomodate it). I believe he had to submit it here rather than to mersenne.ca because James can't certify a PRP - as I understand it, the key step of converting the proof into a cert is kept secret, not included in Prime95 itself, but can only be done by the server. (Given his strongly expressed opinion on the matter, I doubt he did it without a proof, and if he did James could have taken it directly).

I did not believe any result with an exponent over 1G would be accepted; is this a change? Can any exponent up to MAXINT be submitted and the results appear on mersenne.ca at least? Or is this done merely to create a (very) 'special' CERT that will be entirely manually processed?

[Prime95]

Originally Posted by Andrew Usher

C-PRP on exponent 1,000,000,007 (and the exponent column widened automatically to accomodate it). I believe he had to submit it here rather than to mersenne.ca because James can't certify a PRP - as I understand it, the key step of converting the proof into a cert is kept secret, not included in Prime95 itself, but can only be done by the server.

I did not believe any result with an exponent over 1G would be accepted; is this a change? Can any exponent up to MAXINT be submitted and the results appear on mersenne.ca at least? Or is this done merely to create a (very) 'special' CERT that will be entirely manually processed?

Anyone wanting to do a PRP test on an exponent above 1e9 can let me know and I'll add the exponent to the database. Ken, is testing out what web pages might break due to the large exponent. I will not be expanding the database to track TF to low bounds.

Code for turning a proof into a cert can be made public without compromising certification integrity.

All proofs above ~590M must be handled manually. This is because the server is limited by the maximum FFT size for SSE2 FFTs.

[kriesel]

Originally Posted by Prime95

Anyone wanting to do a PRP test on an exponent above 1e9 can let me know and I'll add the exponent to the database. Ken is testing out what web pages might break due to the large exponent.

Presumably, any such runs should be p < 2³¹ to avoid the necessity of changing the exponent representation in the database from signed int32 to unsigned 32-bit or 64-bit, and p < 1169M to avoid a necessity of creating longer AVX512 fft lengths than 64M for proof processing and cert performance. PRP, with GEC, and proof generation, to near optimal power for overall run time (PRP, proof generation, server processing, and cert), ought be required, required, required, and strongly recommended, respectively.
Exponents ~1G on Gpuowl, Radeon VII have taken 4.5 to 6.5 months or so each, depending on how much they are interrupted for other faster work, system hangs or maintenance, etc.
The power 10 proof file for 1G7 was ~1.35 GB so a nontrivial upload. Power 9 proof temporaries were ~64. GB so presumably 1G7's power ten required ~130. GB temporary space. That should be local disk space for reliability and performance.
I only attempted M1000000007 because it was an old absurdly high exponent Mp#52 prediction I wanted to clear, and it had survived adequate TF and P-1. https://mersenneforum.org/showpost.p...&postcount=230
It would have made more sense as a guess for Mp#57.
Such large exponents have trivially small odds of finding a prime, orders of magnitude lower than at the first test wavefront per GPU-year expended.

Personal results summaries https://www.mersenne.org/results/ omit the 1G+ for manually submitted and accepted TF, P-1, or PRP results;
explicit exponent status lookup https://www.mersenne.org/report_expo...ll=1&expired=1 refuses entry of exponents >1G;
clicking on a >1G exponent entry in a recent cleared listing https://www.mersenne.org/report_recent_cleared/ results in a 404 error.
Checking on proof status https://www.mersenne.org/report_proof/ also refuses entry of exponents >999999937.
There may be more.

Back migrating previous TF, P-1 etc already listed at mersenne.ca to mersenne.org for any added >1G exponents is a possible low priority future project for the server admins.

This 1G7 PRP run was done by prior agreement by PM that George would add an entry for the exponent to the mersenne.org database when a test result was available. It was not possible to obtain an AID from mersenne.org, or to reserve PRP at p > 1G at mersenne.ca, at the time.

Its proof upload required some tweaking of the server scripting. After that the manual proof upload web page succeeded. Before that tweak, neither web page upload nor prime95 upload worked.