Newer milestone thread

[petrw1]

Originally Posted by kriesel

Under what name? I did a search just now and found no posts under the name "David Barina".Same here. .

I recognized the picture and looked until I found a post with the same picture.

dabler

[rudy235]

LOOKS VERY PROMISING. Things have changed notably!

There are 13 assignments

46753351 D LL, 86.60% 7 4 2019-02-04 2019-03-28 2019-03-31 2019-04-02 BigBrother
46769131 D LL, 78.80% 9 7 2019-02-06 2019-03-28 2019-03-31 2019-04-05 BigBrother

46810213 D LL, 93.90% 10 1 2019-02-07 2019-03-29 2019-03-30 2019-03-30 David Barina
46833559 D LL, 92.00% 11 1 2019-02-08 2019-03-29 2019-03-30 2019-03-30 David Barina

46837667 D LL, 88.20% 11 1 2019-02-08 2019-03-29 2019-03-30 2019-03-30 David Barina
46838347 D LL, 87.60% 12 1 2019-02-09 2019-03-29 2019-03-30 2019-03-30 David Barina

46868161 D LL, 66.70% 16 4 2019-02-13 2019-03-29 2019-03-30 2019-04-02 David Barina
46868317 D LL, 66.80% 16 4 2019-02-13 2019-03-29 2019-03-30 2019-04-02 David Barina

46891193 D LL, 56.70% 19 5 2019-02-16 2019-03-29 2019-03-30 2019-04-03 David Barina
46891237 D LL, 57.20% 19 5 2019-02-16 2019-03-29 2019-03-30 2019-04-03 David Barina

46952509 D LL, 26.90% 26 8 2019-02-23 2019-03-29 2019-03-30 2019-04-06 David Barina

46952617 D LL, 26.40% 26 8 2019-02-23 2019-03-29 2019-03-30 2019-04-06 David Barina

46999171 D LL, 3.40% 3 -21 2019-02-27 2019-03-02 2019-03-03 2019-03-08 Craig Chadwick Esq

[Madpoo]

Originally Posted by kriesel

Received a gracious email reply indicating they are resumed.

Cool. Sometimes machines reboot or hang or whatever and the person may not realize it.

I was doing a stress test on a new desktop and had Prime95 cooking away on it. It was a few days before I realized it had rebooted to install some updates or whatever (I wasn't running it as a service or anything).

[kriesel]

Originally Posted by Madpoo

Cool. Sometimes machines reboot or hang or whatever and the person may not realize it.

I was doing a stress test on a new desktop and had Prime95 cooking away on it. It was a few days before I realized it had rebooted to install some updates or whatever (I wasn't running it as a service or anything).

Or systems get upgraded, and the focus can be on getting that done right and back in service.

One of David Barina's ten has already completed, and 3 more are expected to complete today. So it seemed well worth the minor effort to contact him.
Expiration of Craig Chadwick's 3.4% complete does not concern me. This milestone should go pretty straightforwardly and quickly from here, if the one expiration does not land on too slow a machine.

[kriesel]

6 of which are projected to be done in 5 days.
https://www.mersenne.org/assignments...tf=1&exfirst=1

[Dr Sardonicus]

I was wondering if there was an ETA for double-checking the remaining exponents up to 57885161 (the exponent of the 48th known Mersenne prime, but not yet known to be the 48th Mersenne prime because of exponents not yet double-checked).

By looking at the Exponent Status Distribution page, I did a BOTE calculation. I tried adding (approximately, since I was doing it by sight) the assigned and available DC's, and dividing by the current rate of clearing 175 DC's per day.

I got something around two and a half years. I might have screwed up the arithmetic, but is the method reasonable?

[kriesel]

Originally Posted by Dr Sardonicus

I was wondering if there was an ETA for double-checking the remaining exponents up to 57885161 (the exponent of the 48th known Mersenne prime, but not yet known to be the 48th Mersenne prime because of exponents not yet double-checked).

By looking at the Exponent Status Distribution page, I did a BOTE calculation. I tried adding (approximately, since I was doing it by sight) the assigned and available DC's, and dividing by the current rate of clearing 175 DC's per day.

I got something around two and a half years. I might have screwed up the arithmetic, but is the method reasonable?

Bar-napkin check:
In a month or two we'll pass 47M DC completely, 57885161 ~58M, delta=11M;
https://www.mersenne.org/report_milestones/ see DC section, 2014 thru 2018 averaged 4M annually of DC milestones; 11M/4M=2.75 years; early 2022 to 58M. ~115K/4000K * 365 days ~ 10 days rounding error to 58M.
Looks reasonable, and actually pretty good agreement.

For your same method, what do you get for first-time checks through 100M?
Second napkin:
first test milestone progression 6M/year in 2017-2018*; 82M through end of 2018; rough est 18/6=3 years, end of 2021 for 100M first test milestone.

(* last 2018 milestone was accelerated a bit to complete 12/30/18)

[ATH]

Originally Posted by Dr Sardonicus

I was wondering if there was an ETA for double-checking the remaining exponents up to 57885161 (the exponent of the 48th known Mersenne prime, but not yet known to be the 48th Mersenne prime because of exponents not yet double-checked).

By looking at the Exponent Status Distribution page, I did a BOTE calculation. I tried adding (approximately, since I was doing it by sight) the assigned and available DC's, and dividing by the current rate of clearing 175 DC's per day.

I got something around two and a half years. I might have screwed up the arithmetic, but is the method reasonable?

http://hoegge.dk/mersenne/GIMPSstats.html

Last 30 days we did 5854 LL-D under 100M or 195 per day, while the last 365 days we did 61447 or 168 per day.

https://www.mersenne.org/report_milestones/

There are 148,976 left until 57885161, so probably somewhere between 148976/195 ~ 764 days (2021-05-03) and 148976/168 ~ 887 days (2021-09-03).

[kriesel]

Originally Posted by ATH

http://hoegge.dk/mersenne/GIMPSstats.html

Last 30 days we did 5854 LL-D under 100M or 195 per day, while the last 365 days we did 61447 or 168 per day.

https://www.mersenne.org/report_milestones/

There are 148,976 left until 57885161, so probably somewhere between 148976/195 ~ 764 days (2021-05-03) and 148976/168 ~ 887 days (2021-09-03).

Excellent, a third estimate. It seems a bit optimistic, since the wavefront has width of several million in exponent value at any given time, with much of the activity at any given moment happening several million higher than the million-bin for the next minor milestone. So call it a lower bound of 2.1-2.4 years.

In Feb 2018, DC activity was spread out over ~10M; in Nov 2018, also 10M.
https://www.mersenneforum.org/showpo...73&postcount=2
It's also currently around 10M width. https://www.mersenne.org/primenet/
Milestones are accomplished when the trailing edge of the wavefront passes the significant point. Time skew between the DC wavefront's median active exponent and trailing edge is ~1.6 years. Median remaining DC below 58M is ~53.5M now.

Typical DC candidate count ahead of the wavefront is ~19154/M in the range of interest.
The wavefront will advance, based on ATH's 30-day rate, at 19154/5854 * 30d/M or 98.16d/M.
(57.885-46.753)M * 98.16/M =1093 days = 2.99 years. The acceleration implied in the 30 day and 365 day numbers if it continues could take a few months off that.

[Uncwilly]

Since the assignment rules change, I believe that the rate of ticking off milestones for the DC's has gone up. I don't have the data at hand. I will have to look at it when I get to the machine that the data live on.

[Dr Sardonicus]

Originally Posted by kriesel

<snip>
For your same method, what do you get for first-time checks through 100M?
<snip>

I wouldn't hazard a guess. Once you get past 57000000 the number of assigned DC's drops off precipitously, and the number of "available" goes up correspondingly.

I reckon there's a "wave" of DC's closing in on the next known Mersenne prime (exponent 57885161), and a "wave" of (first) LL's between about 83000000 and 91000000.

There are also "waves" of TF and P-1 activity that start around 90000000, which I reckon will mostly run their courses before the "waves" of LL and DC activity go through. I suppose you could use clearance rates for these activities to get estimates. I imagine, however, one would have to juggle such variables as (1) the required effort increasing with the size of the exponents, (2) the speed of computing hardware/software, and (3) the amount of hardware being thrown at the project. Too much for my pea brain


For the moment, I'm quite pleased to have some idea how long it will be until we know to DC certainly the next Mersenne prime.