New ECPP record (currently: 59,798 digits)

[sweety439]

Originally Posted by paulunderwood

R49081 also tops the top20 Unique primes: https://primes.utm.edu/top20/page.php?id=62

Thanks to Ray Chandler for pointing it out.

2^82589933-1 and (2^95369+1)/3 are also generalized unique primes (in base 2)

[paulunderwood]

R49081's certificate has been independently verified by factorDB

[ixfd64]

For those who don't follow the "And now for something completely different" forum, here's another new ECPP record as well as a new open source ECPP implementation that's faster than Primo: https://mersenneforum.org/showthread.php?t=27775

[paulunderwood]

Congrats to Andreas Enge for using his own ECPP software CM to prove tau 199^4518 at 57,125 digits

In factorDB - http://factordb.com/index.php?id=1100000003886300622

[frmky]

125330^3+3^125330, at 59,798 digits, is prime.

[paulunderwood]

Originally Posted by frmky

125330^3+3^125330, at 59,798 digits, is prime.

Congrats!!!!!! https://primes.utm.edu/primes/page.php?id=134631

Pray tell what hardware you used and how long the proof took.

[frmky]

Originally Posted by paulunderwood

Congrats!!!!!! https://primes.utm.edu/primes/page.php?id=134631

Pray tell what hardware you used and how long the proof took.

It was completed with CM + GWNUM for the stage 1 primality step. Stage 1 was run on 6 nodes, each with two Xeon E5-2670 v2 10-core CPUs, totaling 120 cores, and took just over two months to complete. Stage 2 was less organized because some steps took longer than the 100 hour runtime limit imposed by the cluster queue. Thus it was inefficiently shuttled around to available machines as needed, and took about three calendar months to complete.

[chalsall]

Originally Posted by frmky

Thus it was inefficiently shuttled around to available machines as needed, and took about three calendar months to complete.

Ummm... Wow... 8-)

[frmky]

Originally Posted by chalsall

Ummm... Wow... 8-)

Should I have said the methodology was adapted to optimize use of available resources?

[Batalov]

I can relate.

I recently had a test where Stage 2 "test #0" was not finished when 90% of the other tests were already finished. If that process gets killed - you have to start from scratch because no data in the state file (.cert2) are written that are not in sequence from #0 to #tests. Maybe this can be refactored (thinking about it)?

So if you had the same then you are stuck on redoing all the same things again and again until the resources are enough for compound Stage 2 test.

I think even if suboptimal - the Primo strategy of writing finished <base>.t#n files are written when the are finished. They do not have to be a "single flat file" database; they can be (fwiw) a flat fileset database

[paulunderwood]

Originally Posted by Batalov

I recently had a test where Stage 2 "test #0" was not finished when 90% of the other tests were already finished. If that process gets killed - you have to start from scratch because no data in the state file (.cert2) are written that are not in sequence from #0 to #tests. Maybe this can be refactored (thinking about it)?

That would be a nice feature. Duplicated work is not welcomed. Maybe someone can tell the author.