mersenneforum.org  

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

Reply
 
Thread Tools
Old 2017-10-02, 16:24   #1
Siegmund
 
Siegmund's Avatar
 
Mar 2014

22×13 Posts
Default Unexpected exponent expiration

My first XXX-rated thread!

I observed this morning that one of my assignments had been expired. It was M75642647, cat 1 when I started it and cat 0 now.
But it was started promptly, reported on daily (87% yesterday, 13 days after assignment), and well within the 30 day allotment.

Looking at the exponent report, we see that someone else had the exponent in August, expired before I was assigned the exponent, and then turned in a result today.

Nothing wrong with somebody finishing a late result; I've had that happen before. But usually that turns my LL assignment into a D assignment, but doesn't change my expiration date.
This time, the result turned in was a "PRP" result rather than a "LL" result -- and apparently that does something different.

Is this a bug introduced by some recent experimenting with the PRP feature?

(By the time you read this, my LL test will have finished, and also appear in the exponent report.)
Siegmund is offline   Reply With Quote
Old 2017-10-02, 17:14   #2
GP2
 
GP2's Avatar
 
Sep 2003

5×11×47 Posts
Default

You seemed confident that your LL test would be finished, but it's still showing only the one, unverified result. There is also no expired assignment for you.

Did your LL test finish? If so, try copy-and-pasting the relevant lines from your results.txt file to https://www.mersenne.org/manual_result/

If the server really converted your assignment to a PRP test and then threw away your result when it was completed and reported, that would be a pretty big bug.

Last fiddled with by GP2 on 2017-10-02 at 17:19
GP2 is offline   Reply With Quote
Old 2017-10-02, 18:47   #3
Siegmund
 
Siegmund's Avatar
 
Mar 2014

22×13 Posts
Default

Yes, my LL test result was submitted a few minutes after I posted. (Mine is the Gordon Bower LL showing on the exponent report now.)

When my result was submitted, my name was removed from the list of expired assignments (where it appeared most of the morning.)

The bug, if there was one, afflicted the person who had the exponent before me, who had it from August 30th to September 19th and then submitted something -- I am not sure what -- early this morning.

My concern is that we know what odd thing the other tester did to the exponent, and why his result submission caused the server to become confused about my assignment. (And, of course, that I don't get charged with failing to complete an exponent on time and kicked back to Category 4...)
Siegmund is offline   Reply With Quote
Old 2017-10-02, 20:41   #4
preda
 
preda's Avatar
 
"Mihai Preda"
Apr 2015

2×691 Posts
Default

Quote:
Originally Posted by Siegmund View Post
The bug, if there was one, afflicted the person who had the exponent before me, who had it from August 30th to September 19th and then submitted something -- I am not sure what -- early this morning.
I submitted the PRP result. The story is like this: I got an LL manual assignment (there was no PRP assignment back then), did PRP on it on the GPU, then waited a while until the server was ready to accept the PRP result, and submitted the result today.
preda is offline   Reply With Quote
Old 2017-10-02, 22:59   #5
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

24×17×29 Posts
Default

This is somewhat a fallout of the upcoming PRP change.

preda is the author of gpuowl which does PRP tests on Mersenne numbers. PRP residues are incompatible with LL residues. When a PRP result was checked in, the server saw that it did not need your LL result to complete the double-check and just expired it as "poached".

The real problem is preda needs a good way to get PRP assignments that have never been assigned as LL (or at least unlikely to have an expired LL result returned). Also, we need to modify LL assignments to stay away from PRP tests that are recently expired and may have a PRP result unexpectedly returned. It will take time to get these rules right.

As to your exponent, it can now be DCed with either a PRP test or an LL test.
Prime95 is offline   Reply With Quote
Old 2017-10-02, 23:18   #6
Siegmund
 
Siegmund's Avatar
 
Mar 2014

22·13 Posts
Default

I appreciate the news from you both, and am sure that eventually all the dust will settle.

I just wanted to make sure that someone was aware there was some oddness going on.
Siegmund is offline   Reply With Quote
Old 2017-10-03, 03:21   #7
Madpoo
Serpentine Vermin Jar
 
Madpoo's Avatar
 
Jul 2014

52×7×19 Posts
Default

Quote:
Originally Posted by Prime95 View Post
This is somewhat a fallout of the upcoming PRP change.

preda is the author of gpuowl which does PRP tests on Mersenne numbers. PRP residues are incompatible with LL residues. When a PRP result was checked in, the server saw that it did not need your LL result to complete the double-check and just expired it as "poached".

The real problem is preda needs a good way to get PRP assignments that have never been assigned as LL (or at least unlikely to have an expired LL result returned). Also, we need to modify LL assignments to stay away from PRP tests that are recently expired and may have a PRP result unexpectedly returned. It will take time to get these rules right.

As to your exponent, it can now be DCed with either a PRP test or an LL test.
I'd suggest that PRP assignments shouldn't go out if there's an expired LL assignment but they've updated recently (even though expired). So, rather than just looking at "is it available" it also looks at the "last updated" for the previous assignment and skips it if it's within the past 30/60 days or whatever.

Vice versa too when PRP assignments are out there, making sure they're not assigned as LL in a similar situation.
Madpoo is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Strange expiration qingkong PrimeNet 3 2016-10-21 16:40
Expiration Dates Fred PrimeNet 3 2016-02-20 08:30
Expiration time NormanRKN GPU Computing 7 2013-06-28 23:53
LL and DC expiration times Chuck GPU to 72 8 2012-01-18 17:39
Exponent expiration drh PrimeNet 25 2011-10-24 04:51

All times are UTC. The time now is 06:47.


Fri May 27 06:47:34 UTC 2022 up 43 days, 4:48, 0 users, load averages: 1.38, 1.31, 1.14

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 𝛴𝜎𝜍 𝛵𝜏 𝛶𝜐 𝛷𝜙𝜑 𝛸𝜒 𝛹𝜓 𝛺𝜔