mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Data (https://www.mersenneforum.org/forumdisplay.php?f=21)
-   -   COMPLETE!!!! Thinking out loud about getting under 20M unfactored exponents (https://www.mersenneforum.org/showthread.php?t=22476)

petrw1 2022-03-29 19:11

[QUOTE=kruoli;602784]With my current procedure (hitting 7.8M with [I]really[/I] high bounds), I will take likely longer. What do you suggest:[LIST][*]Giving up my high bounds and prefer clearing the range?[*]Proceed as is?[*]Since my stage 1 is currently a lot more efficient: provide save files such that other can help with my high bounds?[/LIST]
Any input/suggestions is/are appreciated![/QUOTE]

I won't tell you what to do....your PC your choice.

I did a quick test.
My fastest computer (a 3 year old 7820X) would take about 9 hours to do one test to your bounds (25M/40B)
and statistically, at that rate it would take 2 or 3 days per factor.
I have some decent DDR4-3600 RAM so I might be more useful to help with Stage 2.

I could also grab a couple hundred to somewhat lower bounds to finish this project, then you cold continue to run the entire range to your bounds. I could send you my savefiles to save some time if you choose to re-do them to your bounds.

Maybe someone else with a "beast" wants to help and honor your chosen bounds? @luminescence (and others) has some impressive gear. I leave it to them whether they want to respond.

chalsall 2022-03-29 20:19

[QUOTE=kruoli;602784]Any input/suggestions is/are appreciated![/QUOTE]

An additional option... Although it's really expensive down there, I know there are several who don't mind doing TF at the low ranges.

Let me know if you want me to bring back in -- let's say a thousand -- to work; race conditions aren't really all that much of a problem (statistically; as in, not /that/ much toe stepping as the odds are low for both effort types).

kruoli 2022-03-29 20:22

[QUOTE=petrw1;602817]I won't tell you what to do....your PC your choice.[/QUOTE]
Of course, I never thought that you wanted to compel me to do something. I just wanted thoughts of others. :smile:

[QUOTE=petrw1;602817]I could also grab a couple hundred to somewhat lower bounds to finish this project, then you cold continue to run the entire range to your bounds.[/QUOTE]
As of now, it is not my plan to run the entire range with this settings since it would take months on the system I am running it on. I am applying the strategy of running exponents with the highest number of remaining candidates in the subranges first and will stop when I have all subranges under 200 (or if running a whole subrange does not clear it). With my high bounds, it is really unlikely that anyone else needs/wants to re-do my exponents en masse to even higher bounds, but the exponents I have not touched can be P-1'ed to everyone's likings.

kruoli 2022-03-29 20:26

[QUOTE=chalsall;602822]Let me know if you want me to bring back in -- let's say a thousand -- to work; race conditions aren't really all that much of a problem (statistically; as in, not /that/ much toe stepping as the odds are low for both effort types).[/QUOTE]

If there is nothing else that needs thatcomputational power more urgently, I will gratiously accept your offer, thank you! :smile:
What about 500 exponents for a start? I can PM you a list of exponents, if that is easy to implement with your system. Otherwise, feel free to choose them yourself (of course :razz:).

petrw1 2022-03-29 21:23

[QUOTE=kruoli;602823], but the exponents I have not touched can be P-1'ed to everyone's likings.[/QUOTE]

I'll get back to you in a about a week and help then.

To stay out of your way I'll work on sub-ranges already under 200 where I suspect you are not??

axn 2022-03-30 13:25

Current range stats:
[CODE]Ranges Remaining: 15. Less than 20 To Go: 8. 20-39: 4. 40-59: 2. 60 or more: 1.[/CODE]

Worthy of Mersenne himself :smile:

masser 2022-03-30 14:01

[QUOTE=axn;602861]Current range stats:
[CODE]Ranges Remaining: 15. Less than 20 To Go: 8. 20-39: 4. 40-59: 2. 60 or more: 1.[/CODE]

Worthy of Mersenne himself :smile:[/QUOTE]

:cool:

I was thinking we should change the categories to:

- less than 8
- 8-15
- 16-31
- 32 or more

chalsall 2022-03-30 15:29

[QUOTE=masser;602866]I was thinking we should change the categories to:[/QUOTE]

LOL... How about a compromise with a [URL="https://www.gpu72.com/twok/"]better "spread" of the buckets[/URL]...

[C]Ranges Remaining: 15. Less than 16 To Go: 6. 16-31: 3. 32-47: 5. 48 or more: 1.[/C]

masser 2022-03-30 16:41

[QUOTE=chalsall;602872]LOL... How about a compromise with a [URL="https://www.gpu72.com/twok/"]better "spread" of the buckets[/URL]...

[C]Ranges Remaining: 15. Less than 16 To Go: 6. 16-31: 3. 32-47: 5. 48 or more: 1.[/C][/QUOTE]

:tu:

petrw1 2022-03-31 16:31

If I can offer a TF opinion...
 
15.6M to TF73 and
15.2M to TF74
would help the most.

chalsall 2022-03-31 17:07

[QUOTE=petrw1;602938]15.6M to TF73 and 15.2M to TF74 would help the most.[/QUOTE]

The former just started (by the Colab TF'ers) and the latter is just about complete.

BTW... Does anyone have a block of say 0.01M that I could give to the Colab P-1'ers? We've been surprisingly successful in the last 24 hours! :tu:


All times are UTC. The time now is 10:54.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2022, Jelsoft Enterprises Ltd.