20210831, 06:05  #3455 
Romulan Interpreter
"name field"
Jun 2011
Thailand
10011001001110_{2} Posts 

20210902, 16:57  #3456 
Xebeche
Apr 2019
πΊππΊ
2·3·73 Posts 
A week ago I corrected my prediction regarding All tests below M57885161 moment from Nov 22 to Nov 16 by 6 days.
https://www.mersenneforum.org/showpo...1&postcount=31 What we have at the moment looks like this: The difference between kriesel's forecast and mine is more than a month. The intrigue is growing :) Last fiddled with by greenskull on 20210902 at 17:03 
20210902, 19:48  #3457 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest
3×11×179 Posts 
Sorting https://www.mersenne.org/assignments...execm=1&extf=1 by estimated completion,
1) look toward the end: latest completion date estimate for which expire time is after ETA (indicating probable completion before reassignment) is 20210926. 2) look toward the beginning: latest expiration date for those that have stopped updating days or weeks ago is 24 days from now, so also 20210926 (57836003) to expiration. It and similar will be assigned someone with a track record of completing within 10 days, so unless poached (providing earlier completion up to 57885161), or abandoned by the new recipient delaying it with another expiration/reassignment, it will take until ~Oct 7 or sooner, probably ~Oct 2. (Drkirby and some others have hardware that can do such an exponent in ~1 day.) 3) Likely completion date of all remaining exponents below Mp#48* is the later of the dates from the two points above. 4) All these exponents are cat 0 now; recycled by the server if not started within 10 days or completed within 30. The above is quite consistent with kriesel's previous projections produced by a completely different method (currently indicating ~20210928). Greenskull's revised projection is Nov 16, 40 days later than Oct 7. For greenskull's projection to be accurate or optimistic, multiple expirations or recycles must occur on the same remaining exponent(s), and no poaches that offset them. For example, 20210926 expiration + 30 days second expiration is 20211026, still short of Nov 16 by 3 weeks. The completion rate in the final ~12 days of reaching 56M was ~1.7/day. There are now 52 left to Mp#48*; 52/1.7 ~ 30 days > ~Oct 2. (Daily rate in the month before the final ~2 weeks was ~2.34/day. Averaging for ~4 weeks, ~2/day, gives ~26 days, ~Sep 28.) Latest expiration is now Sep 29, and that one is estimated to complete on Sep 4. Last fiddled with by kriesel on 20210902 at 19:56 
20210909, 09:18  #3458 
Xebeche
Apr 2019
πΊππΊ
2·3·73 Posts 
In theory, of course, the situation could develop along a solid green curve that reflects kriesel's vision.
But it seems to me that such an outcome is unlikely. Since we are running between two threads, I will leave here a link to my comment in another thread, which I wrote half an hour earlier: https://www.mersenneforum.org/showpo...0&postcount=33 Last fiddled with by greenskull on 20210909 at 09:28 
20210909, 11:23  #3459 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest
3·11·179 Posts 

20210909, 21:27  #3460 
Einyen
Dec 2003
Denmark
3,191 Posts 
Code:
All exponents below 104β286β673 have been tested at least once. Last fiddled with by ATH on 20210909 at 21:27 
20210909, 21:32  #3461 
Xebeche
Apr 2019
πΊππΊ
2·3·73 Posts 

20210914, 17:47  #3462 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest
3×11×179 Posts 
In about 5 days, the lowest unverified exponent value will jump by about 1/3 million, with the completion of 57223591 DC, assuming the ETAs of the lowest few exponents at https://www.mersenne.org/assignments...tf=1&exfirst=1 are what occurs.
Current projection from the 56Mhistory analogy is ~Sept 28, while the alternate method gives ~Oct 3, for completing verification up to M57885161. Last fiddled with by kriesel on 20210914 at 17:47 
20210915, 08:24  #3463 
Xebeche
Apr 2019
πΊππΊ
2×3×73 Posts 

20210916, 09:50  #3464 
Xebeche
Apr 2019
πΊππΊ
2×3×73 Posts 
What we have at the moment. We have nothing special.
I still stick to my early November 16th prediction. I admit that the cherished moment may come a little earlier (let's say a couple of days), but for now I will not catch fleas :) 
20211004, 12:56  #3465 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest
3·11·179 Posts 
2 left to 105M first test milestone

Thread Tools  
Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
Newer X64 build needed  Googulator  Msieve  73  20200830 07:47 
Performance of cudaecm on newer hardware?  fivemack  GMPECM  14  20150212 20:10 
Cause this don't belong in the milestone thread  bcp19  Data  30  20120908 15:09 
Newer msieves are slow on Core i7  mklasson  Msieve  9  20090218 12:58 
Use of large memory pages possible with newer linux kernels  Dresdenboy  Software  3  20031208 14:47 