Possible extension of expiry dates because of COVID19

[Uncwilly]

I think that there needs to be an extension of assignments for machines available.
There are a number of folks who aren't going into work and can't tend machines there. And some businesses have shut the facilities. So machines that were producing consistently will stop reporting in and will fail to meet deadlines. It would be a shame if machines that were reliable until say March 1 and were doing DC's get their assignments reassigned and done. Then when the user gets the machine back up and running, they complete a DC that is not needed. The date could be set for Feb 1 for Italian IP's. Don't have date for other regions.

In a PM with a user with soon to be expiring exponents, at least one of their machine was like this. One of the super-mods can't get to their machines located at work.

Anyone in this situation please say so, so that we can get a feel if this might be an issue.
Also, if you are working from home and can' access your machines, check their assignment status on you PrimeNet user page. https://www.mersenne.org/workload/ You may see that some haven't checked in since you left.

[kladner]

Lots of other commitments are going on hold: rent, mortgage, loans....
It seems the right thing to do here.

[kriesel]

My inclination is to let DCs extend, except for cat 0 that delay milestones.

The current COVID19 situation, of widespread disruption due to pandemic and lack of a proven vaccine, is expected to continue for more than a year. There will be some breaks in the shutdowns of most direct human contact. These breaks will probably not be synchronized from one nation to the next or even across individual large nations. Opportunities to reconfigure for less delay of progress during shutdown periods ought be taken advantage of when practical. Where policies permit, remote access may be configured during local access periods. (My systems are mostly configured with redundant remote access provisions.) Contingency plans for the GIMPS participant being felled are also to be considered, whether occurring by pandemic, traffic collision or the myriad other possibilities. Perhaps by a discussion with and documentation memo to a relative or coworker who agrees to step in to conclude things GIMPS-wise.

Work-from-home scenarios would typically involve a VPN to the office. Remote desktop around the LAN from that beachhead is the other piece.
The inherent remote access of cloud computing is an advantage of Google Colaboratory and other platforms. Consider switching assignment patterns to exploit that.

The bulk of known Mersenne primes have been found during epidemics or pandemics.
HIV is thought to have originated in a jump from a chimpanzee with SIV in 1920. It came into general awareness around 1981, same year as the introduction of the IBM PC. The run of HIV in humans continues without a cure today, covering the span of discovery of M13 = M(521) and higher, now up to M51. HIV general awareness predates the creation of GIMPS by 15 years. SARS in 2003 coincides with M(20996011), and the SARS recurrence in 2013 with M(57885161), which remains the lowest Mersenne prime not to have been reached by exhaustive doublechecking.
https://www.history.com/topics/middl...emics-timeline

Looking at the list of Mersenne numbers at https://www.mersenne.org/primes/, it seems to me they ought to have been called Cataldi numbers. Cataldi found two by the year Mersenne was born; Mersenne actually found none.

[Uncwilly]

Originally Posted by kriesel

My inclination is to let DCs extend, except for cat 0 that delay milestones.

I think that the lowest DC outstanding might be on a machine that is inaccessible due to restrictions.
I know that 50763169 and 50763341 are on untouchable machines.

There are a few other really low ones that have dates that show assignments before stay at home orders and checks before, then nothing after.

[chalsall]

Originally Posted by Uncwilly

There are a few other really low ones that have dates that show assignments before stay at home orders and checks before, then nothing after.

IMO, now is *not* the time to be worried about milestones!!!

[Uncwilly]

Originally Posted by chalsall

IMO, now is *not* the time to be worried about milestones!!!

We are in violent agreement.

[Prime95]

Cat 0,1,2 DC expirations have been put on hold. Someone please remind me to turn them back on at a later date :)

I also changed the "do what makes the most sense" work preference to get a DC 50% of the time -- up from 20%.

[Uncwilly]

Originally Posted by Prime95

Cat 0,1,2 DC expirations have been put on hold. Someone please remind me to turn them back on at a later date :)

I put a reminder in my phone for May 9. Let's see if that makes sense then.

[kriesel]

Originally Posted by Prime95

Cat 0,1,2 DC expirations have been put on hold. Someone please remind me to turn them back on at a later date :)

I also changed the "do what makes the most sense" work preference to get a DC 50% of the time -- up from 20%.

Now that last line, I like. M57885161 has remained in uncertain status for over seven years already. I stepped up my own volume of LL DC on gpus a while back. Shifting what "most sense" means will help limit the rate at which the LL DC is falling even further behind first-tests.

What is the rationale for continuing to expire the rather numerous Cat3 and cat 4 DC but not the Cat 0-2; the expectation that there will be many abandoned Cat 3 & 4?

[Prime95]

Cat 3 & 4 were given a long period of time to start with. Just a hunch that if 9 months was not enough time another 6 weeks probably isn't enough time either.

[ric]

Originally Posted by Uncwilly

I think that there needs to be an extension of assignments for machines available.

On a similar page, is there a way to override the

NoMoreWork=1

setting from the server side? I've a slow machine, so far running smoothly despite being left alone for a few weeks, which is running out of work shortly.