mersenneforum.org > Data Thinking out loud about getting under 20M unfactored exponents
 Register FAQ Search Today's Posts Mark Forums Read

2021-12-10, 06:05   #991
petrw1
1976 Toyota Corona years forever!

"Wayne"
Nov 2006

52·199 Posts

Quote:
 Originally Posted by Luminescence I used to register with Pfactor and then just rewrote the assignments with the „find and replace“ function to the correct Pminus1 line. I’ve stopped doing that since it is so tedious with three or four machines. I wonder why you can’t use Pminus1
Just a bit of a guess but when you register a Pfactor such as this
Quote:
 Pfactor=1,2,10000157,-1,64,2.0
can you choose the last parameter (tests saved) high enough to get the desired B1/B2?

 2021-12-10, 09:12 #992 kruoli     "Oliver" Sep 2017 Porta Westfalica, DE 33E16 Posts Unfortunately, 10 tests_saved is the maximum. This is usually not enough. A few data points: Code: [Work thread Dec 10 09:58] Assuming no factors below 2^64 and 10 primality tests saved if a factor is found. [Work thread Dec 10 09:58] Optimal bounds are B1=554000, B2=33719000 [Work thread Dec 10 09:58] Chance of finding a factor is an estimated 9.14% [Work thread Dec 10 10:01] Assuming no factors below 2^74 and 10 primality tests saved if a factor is found. [Work thread Dec 10 10:01] Optimal bounds are B1=358000, B2=14042000 [Work thread Dec 10 10:01] Chance of finding a factor is an estimated 2.91% With M10000169, I registered it as Pfactor= and after completion (without changing to Pminus1=), it showed up as expired. Weird.
2021-12-10, 09:23   #993
Luminescence

Oct 2021
Germany

41 Posts

Quote:
 Originally Posted by kruoli With M10000169, I registered it as Pfactor= and after completion (without changing to Pminus1=), it showed up as expired. Weird.
What? I thought that was a quirk of changing Pfactor= to Pminus1= in the worktodo file. Uhhhhh… very weird.

2021-12-10, 14:39   #994
petrw1
1976 Toyota Corona years forever!

"Wayne"
Nov 2006

136F16 Posts

Quote:
 Originally Posted by kruoli Unfortunately, 10 tests_saved is the maximum. This is usually not enough.
@George: In light of 30.8's speed could this parm limit be increased?

Or could PrimeNet start to allow/register PMinus1?

Thanks

2021-12-10, 15:04   #995
axn

Jun 2003

122208 Posts

Quote:
 Originally Posted by petrw1 @George: In light of 30.8's speed could this parm limit be increased?
Bad idea. That parameter is a crutch (and a bad one at that) to make P-1 do what we want it to do.

Quote:
 Originally Posted by petrw1 Or could PrimeNet start to allow/register PMinus1?
Good idea. This is the right way for our purpose.

2021-12-10, 16:01   #996
chalsall
If I May

"Chris Halsall"
Sep 2002

1018110 Posts

Quote:
 Originally Posted by axn Good idea. This is the right way for our purpose.
+1.

It would *really* help with the Colab P-1 work.

 2021-12-10, 20:51 #997 Uncwilly 6809 > 6502     """"""""""""""""""" Aug 2003 101×103 Posts 13·787 Posts Can the GPU72 colab sessions get upgraded to v30.7 at least?
2021-12-10, 22:19   #998
Prime95
P90 years forever!

Aug 2002
Yeehaw, FL

22·3·647 Posts

Quote:
 Originally Posted by petrw1 and many others @George: In light of 30.8's speed could this parm limit be increased? Or could PrimeNet start to allow/register PMinus1?
Quote:
 Originally Posted by chalsall It would *really* help with the Colab P-1 work.

Quote:
 Originally Posted by axn Good idea. This is the right way for our purpose.
OK, I tried a hack. Prime95 ought to now register these as PM1-S (same as a Pfactor assignment). I assume they have the same expiration rules (I don't know what they are).

You cannot get P-1 assignments from the server.

Hope that helps (and works).

2021-12-10, 22:58   #999
chalsall
If I May

"Chris Halsall"
Sep 2002

237058 Posts

Quote:
 Originally Posted by Uncwilly Can the GPU72 colab sessions get upgraded to v30.7 at least?
It's a bit complicated...

Whenever there is an incompatibility between versions of mprime I have to do some work to ensure "sane transition" of the checkpoint files.

It's a bad Cx when someone loses a large amount of compute.

@George... Thanks for the alpha hack on Primenet accepting PM1-S via the API. I'll test and "sniff the wire" over the weekend on this.

Just to share...

Few appreciate just how well this collaborative, multi-disciplinary, and agile ad-hock team works.

I sometimes present this as an example of what /should/ be standard MO.

When I do this, many people run away with their eyes closed and their hands over their ears...

Seriously.

Last fiddled with by chalsall on 2021-12-10 at 23:11 Reason: s/stiff the wire/sniff the wire/; # Very different things...

 2021-12-11, 18:19 #1000 petrw1 1976 Toyota Corona years forever!     "Wayne" Nov 2006 Saskatchewan, Canada 52·199 Posts Looking ahead to below 20M 2xM is about 3 months from complete ... holy carp is the new P-1 fast!!!! So I'm looking ahead to 1xM and under: With TF getting less efficient as the ranges get lower and P-1 more efficient and so bloody fast now the balance of power is changing. I'm not sure the TF'ers can keep up to the P1'ers going forward barring a big TF farm surfacing soon. I can't predict the balance of TF vs. P-1 going forward but if I use my farm as a *rough* gauge: - My PC farm of 5 will be able to clear about 10 P-1 factors a day or a range with 70 ToGo in 1 week. - My 2 GPUs would need more than a month to advance 1 range to 73 bits and 2+ months to 74 bits. I've noticed that recently the bulk of TF work below 20M has been in ranges with less than 50 ToGo. I'd like to encourage the TFers to instead focus their efforts on the *worst* ranges. And to encourage P-1'ers to take ranges with the least ToGo and we'll meet in the middle. I think there is a reasonable chance this project comes to an end in 2022. I wish I could say the same for Covid-19. Thanks for your time and consideration. And for everyone's help to date.
2021-12-11, 23:30   #1001
Luminescence

Oct 2021
Germany

41 Posts

Quote:
 Originally Posted by petrw1 2xM is about 3 months from complete ... holy carp is the new P-1 fast!!!! So I'm looking ahead to 1xM and under: With TF getting less efficient as the ranges get lower and P-1 more efficient and so bloody fast now the balance of power is changing. I'm not sure the TF'ers can keep up to the P1'ers going forward barring a big TF farm surfacing soon. I can't predict the balance of TF vs. P-1 going forward but if I use my farm as a *rough* gauge: - My PC farm of 5 will be able to clear about 10 P-1 factors a day or a range with 70 ToGo in 1 week. - My 2 GPUs would need more than a month to advance 1 range to 73 bits and 2+ months to 74 bits.
I also thought of TF in the lower ranges, there is no way they can keep up. 30.8 is leaving GPUs completely behind.

26.5M is in it's final moments and I don't like to see
Code:
Ranges remaining:
100 or more: 9
Mind if I take down 12.3M? Do you have any probability I should be aiming for?

 Similar Threads Thread Thread Starter Forum Replies Last Post jschwar313 GPU to 72 3 2016-01-31 00:50 Batalov Factoring 6 2011-12-27 22:40 jasong jasong 1 2008-11-11 09:43 devarajkandadai Math 4 2007-07-25 03:01 WraithX GMP-ECM 1 2006-03-19 22:16

All times are UTC. The time now is 11:03.

Mon Jan 17 11:03:00 UTC 2022 up 178 days, 5:31, 0 users, load averages: 1.38, 1.01, 0.98