mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > PrimeNet

Reply
 
Thread Tools
Old 2021-01-24, 18:56   #111
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

52·311 Posts
Default

@Kriesel: Please consider prime95 version 30.4 for your P-1 efforts

Quote:
Originally Posted by lisanderke View Post
Oops, it seems like I did what the OP of this thread didn't want to see, I used the manual GPU assignment tool to get Cat1 DC TF assignments... I grabbed about 70 exponents and I should be able to TF all of them within a matter of days (3, perhaps 4).
You are good. Any work you do with a 4 day turnaround won't bother anyone. The project needs LL TF more than DC TF, but do whatever makes you happy.

Quote:
Originally Posted by Runtime Error View Post
What amount of B2 will Prime95 run with the default 0.3 GB of memory for second stage P-1?
No B2. But gpuowl will use good bounds and Ben Delo is using good bounds, so it isn't a huge problem.
Prime95 is online now   Reply With Quote
Old 2021-01-25, 03:52   #112
LaurV
Romulan Interpreter
 
LaurV's Avatar
 
"name field"
Jun 2011
Thailand

2·3·5·7·47 Posts
Default

I take all my P-1 assignments through GPU72, only for the vanity of having the "credit" registered two times (once in Gimps, once in Gpu72). I "hacked" MISFIT in helping me organize them and split them between cards, but the assignments are taken manually, therefore you won't expect me to reserve less than few hundreds of them at a time. The cards go through them quite fast, and I have no time to reserve P-1 work every 10 minutes... At least, till Scott will fix MISFIT to handle P-1 beside of TF (he promised some time ago, but I didn't see any announcement of a new version, most probably he was busy, or in holiday).

A week or two ago I took 400 P-1 at the front wave, from GPU72, of which there are about 90 left to go. Note that usually I raise the default B1/B2 from 1e6/30x to 1.5e6/30x, but few days ago I went back to 1.2e6/30x for one of the cards which was much slower in stage 2. With the default B1/B2 selected by the owl, they would go even faster, but I found out that the best spot for credit per time unit (wall clock) is somewhere at B1=1.2e6, while the most efficient point is somewhere at B1=1.5e6 (i.e. where my chance of finding factors, therefore eliminating exponents is the highest per wall clock time).

Therefore, because I take the P-1 assignments not from PrimeNet, they do not appear in the list as assigned to me. I mean, you can't see a LaurV guy hoarding lots and lots of P-1 exponents. This raise the question what to do about users who take lots of assignments from other sources than PrimeNet (as GPU72) and "block" them? Are there such users? (no idea, just asking).

Last fiddled with by LaurV on 2021-01-25 at 04:06
LaurV is offline   Reply With Quote
Old 2021-01-25, 15:29   #113
ATH
Einyen
 
ATH's Avatar
 
Dec 2003
Denmark

3·5·7·31 Posts
Default

What about these P-1 assignments?
https://www.mersenne.org/assignments...hk=1&exfirst=1

10 from "Albrecht Heeffer" Nov 18th
17 from "OneAfterlife" Nov 24th
20 from "Wentao Huang Nov 29th
10 from "Chad Steenerson" Dec 11th-13th
20 from "ANONYMOUS" Dec 19th.

No visible progress on them, and they were all in Cat 1 when assigned and are now 37-68 days old.
How many days should Cat 1 P-1 assignments be allowed to finish?

Last fiddled with by ATH on 2021-01-25 at 15:30
ATH is online now   Reply With Quote
Old 2021-01-29, 21:19   #114
Uncwilly
6809 > 6502
 
Uncwilly's Avatar
 
"""""""""""""""""""
Aug 2003
101×103 Posts

37·277 Posts
Default

More than 1/3 of all the assignments in the 101M-102M range are P-1's that are a month old (105 of 300 total). 45 are 2 or more months old
https://www.mersenne.org/assignments...rst=1&excert=1

Last fiddled with by Uncwilly on 2021-01-29 at 21:19
Uncwilly is offline   Reply With Quote
Old 2021-01-29, 23:19   #115
chalsall
If I May
 
chalsall's Avatar
 
"Chris Halsall"
Sep 2002
Barbados

3×43×79 Posts
Default

Quote:
Originally Posted by Uncwilly View Post
More than 1/3 of all the assignments in the 101M-102M range are P-1's that are a month old (105 of 300 total). 45 are 2 or more months old
George et al...

This is now becoming a formal policy decision discussion.

Some think it doesn't make sense. The code allows it.

Convergence is...?
chalsall is online now   Reply With Quote
Old 2021-01-29, 23:35   #116
ATH
Einyen
 
ATH's Avatar
 
Dec 2003
Denmark

3×5×7×31 Posts
Default

I'm currently poaching the ones from "Albrecht Heeffer" and "OneAfterlife" from November, and I do not feel bad at all about it.

There are no progress on them and no expiration rules and they over 2 months old for a Cat1 P-1 assignment. That is unacceptable and no one seems to be doing anything about it.

On the P100 and V100 on Colab Pro I'm probably finishing PRP tests in not much longer time than others would take to do a P-1 test. Seems kinda wasteful these days spending much time on P-1 when only doing 1 PRP test + Cert.

Last fiddled with by ATH on 2021-01-29 at 23:40
ATH is online now   Reply With Quote
Old 2021-01-29, 23:47   #117
chalsall
If I May
 
chalsall's Avatar
 
"Chris Halsall"
Sep 2002
Barbados

3×43×79 Posts
Default

Quote:
Originally Posted by ATH View Post
I'm currently poaching the ones from "Albrecht Heeffer" and "OneAfterlife" from November, and I do not feel bad at all about it.
Nor should you.

To share, I expended some compute to clear out the 104M range where a FT was done without a P-1.

We do what we do. With what amuses our sorry little butts.

Please do try to keep up...
chalsall is online now   Reply With Quote
Old 2021-01-29, 23:54   #118
chalsall
If I May
 
chalsall's Avatar
 
"Chris Halsall"
Sep 2002
Barbados

100111110011112 Posts
Default

Quote:
Originally Posted by chalsall View Post
Please do try to keep up...
Sorry ATH. That was not directed to you.

What I /meant/ to say is several of us are working this space, and go out of our way to not step on each other's toes.

Human is a difficult language to Grok. So many different variants.
chalsall is online now   Reply With Quote
Old 2021-01-30, 16:36   #119
mathwiz
 
Mar 2019

13·17 Posts
Default

I don't see why the server can't just auto-expire these after a month. That would put an end to all this...
mathwiz is online now   Reply With Quote
Old 2021-01-30, 17:32   #120
Uncwilly
6809 > 6502
 
Uncwilly's Avatar
 
"""""""""""""""""""
Aug 2003
101×103 Posts

37×277 Posts
Default

Quote:
Originally Posted by mathwiz View Post
I don't see why the server can't just auto-expire these after a month. That would put an end to all this...
Maybe do something like this:
FTC Cat 0 expiry rules are applied to P-1 in Cat 1 range and TF in Cat 2 range
FTC Cat 1 expiry rules are applied to P-1 in Cat 2 range and TF in Cat 3 range
FTC Cat 2 expiry rules are applied to P-1 in Cat 3

Any Cat 2 that doesn't have P-1 already done, gets assigned with P-1 as part of the FTC.

Same pattern in the DC range.

And like others have pointed out, P-1 can be less well done that in the past, as we are not saving 2 full primality checks.

Uncwilly is offline   Reply With Quote
Old 2021-01-30, 18:02   #121
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

25×191 Posts
Default

Or maybe:
Cat 0 expiration rules apply equally to all assignment types on PrimeNet API issued FTC Cat 0 exponents, and are applied promptly
Cat 1 expiration rules apply equally to all assignment types on PrimeNet API issued FTC Cat 1 exponents, and are applied promptly
Cat 2 expiration rules apply equally to all assignment types on PrimeNet API issued FTC Cat 2 exponents, and are applied promptly
That has the advantages of conceptual simplicity, moving wavefront/milestone exponents toward completion rapidly, and moving exponents toward completion in an orderly way, especially focused on prime95 set and forget installations.
(Good luck to whoever scripts that on the server.)

A FTC assignment in mprime / prime95 will first do P-1 on an exponent if it is not adequately done already, and memory available is adequate, and the exponent could get stalled or abandoned while in P-1 or LL or PRP.
A TF assignment in whatever can get delayed. Such an assignment ought not delay progress on a wavefront exponent indefinitely, especially when it is an unneeded, past-optimal-bit-depth, manual TF.

Also it's worth considering an end to issue of first-time LL via PrimeNet or in general.

Last fiddled with by kriesel on 2021-01-30 at 18:15
kriesel is online now   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Trial Factoring - Factor Confirmation? butera PrimeNet 6 2021-04-02 21:34
possible overlapping Fermat factor ranges MattcAnderson FermatSearch 3 2021-01-05 14:34
Trial Factor Bit Depth lavalamp Operation Billion Digits 8 2010-08-02 18:49
trial division over a factor base Peter Hackman Factoring 7 2009-10-26 18:27
Shortest time to complete a 2^67 trial factor (no factor) dsouza123 Software 12 2003-08-21 18:38

All times are UTC. The time now is 20:02.


Mon Jan 24 20:02:23 UTC 2022 up 185 days, 14:31, 1 user, load averages: 1.23, 1.51, 1.53

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

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.

≠ ± ∓ ÷ × · − √ ‰ ⊗ ⊕ ⊖ ⊘ ⊙ ≤ ≥ ≦ ≧ ≨ ≩ ≺ ≻ ≼ ≽ ⊏ ⊐ ⊑ ⊒ ² ³ °
∠ ∟ ° ≅ ~ ‖ ⟂ ⫛
≡ ≜ ≈ ∝ ∞ ≪ ≫ ⌊⌋ ⌈⌉ ∘ ∏ ∐ ∑ ∧ ∨ ∩ ∪ ⨀ ⊕ ⊗ 𝖕 𝖖 𝖗 ⊲ ⊳
∅ ∖ ∁ ↦ ↣ ∩ ∪ ⊆ ⊂ ⊄ ⊊ ⊇ ⊃ ⊅ ⊋ ⊖ ∈ ∉ ∋ ∌ ℕ ℤ ℚ ℝ ℂ ℵ ℶ ℷ ℸ 𝓟
¬ ∨ ∧ ⊕ → ← ⇒ ⇐ ⇔ ∀ ∃ ∄ ∴ ∵ ⊤ ⊥ ⊢ ⊨ ⫤ ⊣ … ⋯ ⋮ ⋰ ⋱
∫ ∬ ∭ ∮ ∯ ∰ ∇ ∆ δ ∂ ℱ ℒ ℓ
𝛢𝛼 𝛣𝛽 𝛤𝛾 𝛥𝛿 𝛦𝜀𝜖 𝛧𝜁 𝛨𝜂 𝛩𝜃𝜗 𝛪𝜄 𝛫𝜅 𝛬𝜆 𝛭𝜇 𝛮𝜈 𝛯𝜉 𝛰𝜊 𝛱𝜋 𝛲𝜌 𝛴𝜎 𝛵𝜏 𝛶𝜐 𝛷𝜙𝜑 𝛸𝜒 𝛹𝜓 𝛺𝜔