Why are exponents not expired when nobody is working on them?

[drkirkby]

Looking at some exponents, eg this one

https://www.mersenne.org/report_expo...2193751&full=1

I can't help feel that should be not be held any longer. An anonymous person got assigned it on 14th May last year, has done 0%, and the server was last updated on 17th May last year. That's more than 9 months since they have connected to the server.

An even worse example is this one

https://www.mersenne.org/report_expo...2195761&full=1
which was assigned to an anonymous person in 2018, and was last updated in 2018!

There seems an argument for expiring exponents if people have not done anything with them for a "reasonable" period of time. I would have thought if someone has not accessed the server for 3 months or so, they are not working on it. In some cases its more than 2 years, but still the exponent is assigned to them.

[Chuck]

These exponents (332M) are so far ahead of the current wavefront that they are of little concern.

[lycorn]

See this link for the assignment rules: https://www.mersenne.org/thresholds/

In principle, even cat 4 exponents are recycled after one year.
But the examples you posted are way beyond the wave front for 1st time tests and, to the best of my knowledge, they are not subject to the assignment rules depicted above, although that´s not mentioned on the web page.

Cross posted...

[drkirkby]

Originally Posted by Chuck

These exponents (332M) are so far ahead of the current wavefront that they are of little concern.

But the 332 million exponents are more than 100 million digits, so attract a substantial prize. I am doing a similar-sized exponent. It should take 92 days. So it's not a ridiculous time. However, many of the exponents are reserved by people not working on testing to see if the result is prime or not.

Dave

[PhilF]

Originally Posted by drkirkby

But the 332 million exponents are more than 100 million digits, so attract a substantial prize. I am doing a similar-sized exponent. It should take 92 days. So it's not a ridiculous time. However, many of the exponents are reserved by people not working on testing to see if the result is prime or not.

Dave

Let's say someone thought they were lucky enough that one of the (10) 100-million digit exponents they had picked out in their mind was prime. They could then reserve all 10 of them, and even though the website initially shows 60 day expiration times, that applies only if the machine that has them reserved does not check in at least every 60 days. When it does check in, that 60 day expiration time gets reset.

So, on a slow machine, it might take years to test all 10 of them, with only the one currently being worked on showing any progress. The others would appear "stuck".

[kriesel]

Gpu programs don't have PrimeNet API interface code like mprime /prime95 has.
Progress reporting from Mlucas is a very new development.
The manual web interface does not support submitting progress indications.
I've requested that functionality be added.

Cpus without internet access also can't report progress.

Lack of progress reports is not necessarily lack of progress. I'm not alone in having gpu runs that take hours, days, weeks, or months.

There are plenty of exponents to go around. (nearly 14 million left needing TF to adequate level, P-1 to adequate bounds, then PRP above 332M: http://hoegge.dk/mersenne/GIMPSstats.html)

And the odds of hitting a prime above 332M are quite poor per unit effort. https://www.mersenneforum.org/showpo...4&postcount=16

[drkirkby]

Originally Posted by kriesel

Gpu programs don't have PrimeNet API interface code like mprime /prime95 has.
Progress reporting from Mlucas is a very new development.
The manual web interface does not support submitting progress indications.
I've requested that functionality be added.

Cpus without internet access also can't report progress.

Lack of progress reports is not necessarily lack of progress. I'm not alone in having gpu runs that take hours, days, weeks, or months.

There are plenty of exponents to go around. (nearly 14 million left needing TF to adequate level, P-1 to adequate bounds, then PRP above 332M: http://hoegge.dk/mersenne/GIMPSstats.html)

And the odds of hitting a prime above 332M are quite poor per unit effort. https://www.mersenneforum.org/showpo...4&postcount=16

I take your point, but in the case of

https://www.mersenne.org/report_expo...2195761&full=1

the work done (4.8% in 6 weeks during 2018), has been reported to the server. But since then nothing. I guess it's possible some computer has had no internet access for nearly 3 years. However, reading

https://www.mersenne.org/various/works.php

it states as one of the requirements "An Internet connection available at least once every two months – more often is preferred, weekly or daily is ideal."

Yes, there are millions of exponents to go around, but some in the 332 million range have been tested more than others. If a smaller exponent has been tested for factors up to 2^82, and a larger exponent has been tested for factors up to 2^77, the first is obviously more attractive to test.

This is the first, and probably last very large exponent I will run a PRP test on. The expected completion date is my wedding anniversary, so perhaps it will work out.

Dave