Predict M52

[Uncwilly]

Originally Posted by masser

Add a digit somewhere after the 3 and a reasonable date. Then it will make a nice entry on uncwilly's graph.

The graph has set bounds. There are guesses beyond the bounds.

Originally Posted by Uncwilly

Absurd guesses may be ignore[d] it any statistical reports or graphs.

[masser]

Originally Posted by Uncwilly

The graph has set bounds. There are guesses beyond the bounds.

I know. I think sweety439 should amend their guess to something like:

135029747 - Christmas Day, 2024

That guess would land right in a huge (but realistic!) open patch on your graph.

...AND 135029747 is13D0D13D in base 14! Isn't that awesome?

[tuckerkao]

Originally Posted by masser

I know. I think sweety439 should amend their guess to something like:

135029747 - Christmas Day, 2024

The closest large-enough prime exponent 132,029,749 already has a factor. All other 1_3,029,749 and 13_,029,749 are composite numbers.

sweety439 could have guessed 130,629,749 or 130,299,749 or 130,297,949 -> those numbers may have higher chances for the factors to show up between 2^73 to 2^77 when being trialed as there aren't very many other neighbor primes.


The number M103,305,173 I've passed on to Kriesel which he's currently performing the PRP tests still have many other un-factored look-alikes: M103,305,133, M103,307,153, M103,307,173, M103,308,173, M103,505,173

[kriesel]

Originally Posted by tuckerkao

The number M103,305,173

verified composite (Cert by Ben Delo). The other recently posted M52 predictions have also been primality tested (but not the suggestions for Sweety439's future choices), except for 103037743 which will complete in a few days.

[Uncwilly]

Originally Posted by kriesel

The other recently posted M52 predictions have also been primality tested (but not the suggestions for Sweety439's future choices), except for 103037743 which will complete in a few days.

Your PRP on 102685711 and 102685769 show as suspect. You might get someone else to do them again.

[kriesel]

These were originally done on gpuowl v7.1-1 and v7.1-11 on a system with nonECC ram.
"Please upgrade to v7.2 which fixes a proof generation bug." https://www.mersenneforum.org/showpo...&postcount=154
Sometime soon, will update additional workers to gpuowl v7.2-21.
I don't think PrimeNet will accept a fixed proof submission from the original runs, and it's moot because the inputs were not saved. (Too bad, because they take about a day to PRP and minutes to verify on radeon vii.)
I didn't think I needed to save them, because these were self-verified at the ends of the runs:

2020-11-26 01:10:15 asr2/radeonvii3 102685769 proof verification: doing 200559 iterations
...
2020-11-26 01:13:08 asr2/radeonvii3 102685769 proof: 102685769 proved composite
2020-11-26 01:13:08 asr2/radeonvii3 102685769 Proof self-verification OK
2020-11-26 01:13:08 asr2/radeonvii3 102685769 Proof 'proof\102685769-9.proof' generated

2020-11-26 11:52:08 asr2/radeonvii2 102685711 proof verification: doing 200559 iterations
...
2020-11-26 11:55:04 asr2/radeonvii2 102685711 proof: 102685711 proved composite
2020-11-26 11:55:04 asr2/radeonvii2 102685711 Proof self-verification OK
2020-11-26 11:55:04 asr2/radeonvii2 102685711 Proof 'proof\102685711-9.proof' generated

https://www.mersenneforum.org/showpo...&postcount=111
Thought I was being conservative by self-verifying, and ironically, that is what triggers the proof error. https://mersenneforum.org/showthread.php?t=26152

So falling back to traditional DC approach:
102685769 was already queued in prime95 v30.3b6 on a dual-xeon workstation with ECC ram and current ETA is Dec 26.
102685711 was already queued in prime95 v30.3b6 on a somewhat faster dual-xeon workstation with ECC ram and current ETA is Dec 22.

[tuckerkao]

Looks like the exponents from 102 to 103 millions just opened to everyone today. Can someone other than Kriesel reserve and start to test 102,374,113?

Especially when a specific exponent looks similar to the ones I've guessed, I want to see the results of it faster.

[Uncwilly]

Originally Posted by tuckerkao

Can someone other than Kriesel reserve and start to test 102,374,113?

Test your own number. This is not the thread to solicit help in testing your guesses. If you really think it is prime, test it yourself to get the prize. Put your hardware where your mouth is.

[tuckerkao]

The exponent which I ran the LL test during Oct 2020, I tried the PRP on it too, looked like everything went okay.

https://www.mersenne.org/report_expo...exp_hi=&full=1

[kriesel]

Originally Posted by tuckerkao

Looks like the exponents from 102 to 103 millions just opened to everyone today. Can someone other than Kriesel reserve and start to test 102,374,113?

Especially when a specific exponent looks similar to the ones I've guessed, I want to see the results of it faster.

You could consider getting yourself a Xeon Phi 7250 which can test 4 such guesses simultaneously, completing in 11 days, for a throughput of about 2.5 per week. ~$700. https://www.mersenneforum.org/showpo...4&postcount=11

Not fast enough? Get a used radeon vii gpu, which can test 1 in about a day, for ~$800. Or both, or more, with correspondingly higher cost. But patience is less costly.

All we have to do is fit the assignment rules. https://www.mersenne.org/thresholds/ Within that, I run the mix I choose to run. I think it's peculiar that you seem to mistakenly believe you have a right to tell others what they can or can not test. Perhaps I misunderstood.

Patience and commitment are powerful.

[tuckerkao]

I could finish a PRP test with the result certified, but wasn't able to reserve it when I started on that one, the server required me to finish up 5 PRPs prior to it.

The server even showed my result was reliable before the final certification from another user.