mersenneforum.org  

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

Reply
 
Thread Tools
Old 2013-08-31, 20:59   #1
cheesehead
 
cheesehead's Avatar
 
"Richard B. Woods"
Aug 2002
Wisconsin USA

22·3·641 Posts
Default Anomaly after ECM report; possible ECM data base integrity problem

Before I recently started some ECM on M412007, I checked its Exponent Status and ECM Progress, which showed that 246 curves had been done at B1=50000, and none at any higher bound.

Though my PrimeNet assignment of M412007 specified 150 curves, I decided to reduce that to 34 in my worktodo because that's all that remained to reach the customary 280 total for curves at that B1. (Yes, I knew that the other 116 assigned curves would be just as useful, but for various reasons I cut the count to 34 anyway.)

After the 34 curves were completed, I had prime95 report that result to PrimeNet. Everything appeared all right with that. Here are the relevant lines (with assignment ID omitted) from my prime.log file:
Code:
Sending result to server: UID: RichBWoods/Puttputt, M412007 completed 34 ECM curves, B1=50000, B2=5000000, Wd1: 0AE0016B, AID: xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

PrimeNet success code with additional info:
CPU credit is 0.3587 GHz-days.
Then I re-checked the ECM Progress for M412007. It showed that curves at B1=50000 were "Done", as I expected ... but also showed 56 curves at B1=250000 which is totally unexpected! So, I re-checked M412007 Exponent Status. Its last line was for my 34 curves at B1-50000. I examined all earlier report lines. No one else had reported any curves at B1=250000. I added all the numbers of curves reported. There were 246 before my report, and all at B1=50000, none at B1=250000.

Where did those 56 curves at B1=250000 in the ECM Progress report for M412007 come from???

IIRC, _if_ I had reported doing the entire 150 curves that PrimeNet assigned me, the 116 beyond the 34 needed to reach the B1=50000 standard of 280 would have been credited as a smaller number of curves at B1=250000. I understand that ... but I didn't perform those extra 116, and I didn't report those extra 116!!

Could it be that PrimeNet improperly treated my report of 34 curves as though it actually was for the 150 curves assigned by PrimeNet??? That it converted the potential "extra 116" to a credit of 56 curves at B1=250000, even though I neither performed nor reported those "extra 116"???

If so, this is a problem for the integrity of recorded totals of ECM curves in the GIMPS data base.

Last fiddled with by cheesehead on 2013-08-31 at 21:13 Reason: added prime.log lines
cheesehead is offline   Reply With Quote
Old 2013-09-01, 01:16   #2
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

2·11·349 Posts
Default

Quote:
Originally Posted by cheesehead View Post
Where did those 56 curves at B1=250000 in the ECM Progress report for M412007 come from???
This is ugly and confusing, but normal.

When you see at the top of the chart that 640 curves are required at B1=250K, that number came from GMP-ECM and assumes no ECM has been performed on the number. However, we have done a lot of ECM at B1=50K. In the primenet report, when a column competes, the next column is populated with the equivalent number of curves at the higher B1. In other words, the 280 curves done at B1=50K is equivalent to 56 curves at B1=250K.

When I wrote the PHP code for the web page, I felt I had three options:

1) Reduce the count of required curves at the top of the column (in your case to 640-56).
2) Implement the current system.
3) Start the 250K count at zero, but change it to "Done" once the 640-56th curve is reported.

All three options have their pitfalls. I arbitrarily chose the second option.
Prime95 is offline   Reply With Quote
Old 2013-09-01, 02:55   #3
cheesehead
 
cheesehead's Avatar
 
"Richard B. Woods"
Aug 2002
Wisconsin USA

1E0C16 Posts
Default

Quote:
Originally Posted by Prime95 View Post
This is ugly and confusing, but normal.

When you see at the top of the chart that 640 curves are required at B1=250K, that number came from GMP-ECM and assumes no ECM has been performed on the number. However, we have done a lot of ECM at B1=50K. In the primenet report, when a column competes, the next column is populated with the equivalent number of curves at the higher B1. In other words, the 280 curves done at B1=50K is equivalent to 56 curves at B1=250K.
O, I C now!

Thanks for the explanation. I must not have ever looked before-and-after at the ECM Progress report when this occurred.

Quote:
When I wrote the PHP code for the web page, I felt I had three options:

1) Reduce the count of required curves at the top of the column (in your case to 640-56).
2) Implement the current system.
3) Start the 250K count at zero, but change it to "Done" once the 640-56th curve is reported.

All three options have their pitfalls. I arbitrarily chose the second option.
Once again, not all is what it could be interpreted literally to be in GIMPSdom.

(Another note for my planned grand amalgamated FAQ at the wiki.)

:-)

- - -

Hmm...

Implication: The ECM Progress report will never show a "Done" followed by a blank space in the next column.

Last fiddled with by cheesehead on 2013-09-01 at 03:01
cheesehead is offline   Reply With Quote
Old 2013-09-01, 03:08   #4
cheesehead
 
cheesehead's Avatar
 
"Richard B. Woods"
Aug 2002
Wisconsin USA

1E0C16 Posts
Default

Quote:
Originally Posted by Prime95 View Post
1) Reduce the count of required curves at the top of the column (in your case to 640-56).
Wouldn't that require only a static one-time change to all the column headers (and the corresponding "Done" totals), not some dynamic calculation? I.e., the 250K column heading would simply show 584, and that column's entry be "Done" once that total was reached?

And each of the later columns would also show just the adjusted-for-completion-of-all-previous-columns totals?

A single-time change?

What's the pitfall here? Confusing the veteran ECMers who know the GMP-ECM curve numbers? Wouldn't a simple headnote or footnote of explanation be sufficient for them?

- - -

(Deleted something you'd already answered)

Last fiddled with by cheesehead on 2013-09-01 at 03:38 Reason: Deleted something you'd already answered
cheesehead is offline   Reply With Quote
Old 2013-09-01, 03:20   #5
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

2·11·349 Posts
Default

Quote:
Originally Posted by cheesehead View Post
What's the pitfall here? Confusing the veteran ECMers who know the GMP-ECM curve numbers? Wouldn't a simple headnote or footnote of explanation be sufficient for them?
Two pitfalls I see:

1) People ask why the GIMPS web page required curve counts don't agree with the GMP-ECM required curve counts.

2) Someone starts ECMing a number from scratch and does 584 curves at 250K. This does not result in a "Done" status since 280 curves at 50K were never done.

Yes, footnotes (or your grand FAQ) can address all these pitfalls.
Prime95 is offline   Reply With Quote
Old 2013-09-01, 03:41   #6
cheesehead
 
cheesehead's Avatar
 
"Richard B. Woods"
Aug 2002
Wisconsin USA

22·3·641 Posts
Default

Quote:
Originally Posted by Prime95 View Post
2) Someone starts ECMing a number from scratch and does 584 curves at 250K.
Oh, yeah. That would be all too easy.

Quote:
This does not result in a "Done" status since 280 curves at 50K were never done.
Instead of "Done", it could display one of those finger-pointing-to-the-left hands.

Well, okay ... we're at a local maximum that suffices.

Last fiddled with by cheesehead on 2013-09-01 at 03:53
cheesehead is offline   Reply With Quote
Old 2013-09-01, 03:51   #7
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

2·11·349 Posts
Default

BTW, there are more "problems". In the DB, ECM effort is recorded as one float value. If you do 10 1M curves on your number, that column will remain empty but the 250K column will go up by ~40 curves.
Prime95 is offline   Reply With Quote
Old 2013-09-01, 03:55   #8
cheesehead
 
cheesehead's Avatar
 
"Richard B. Woods"
Aug 2002
Wisconsin USA

22·3·641 Posts
Default

So, if I can get the rest of my life straightened out, it's still not too late for me to contribute programming?
cheesehead is offline   Reply With Quote
Old 2013-09-01, 04:27   #9
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

11101111111102 Posts
Default

Quote:
Originally Posted by cheesehead View Post
So, if I can get the rest of my life straightened out, it's still not too late for me to contribute programming?
Yes, but probably not here. I'm very reluctant to make any database changes.
Prime95 is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Integrity check of Prime95 archive? WaiCeeh Software 18 2017-06-23 14:26
Annoying Ambient Air Anomaly Analyzed and Answered! Xyzzy Miscellaneous Math 3 2015-09-06 06:47
ecm anomaly? swishzzz Factoring 14 2012-02-01 17:26
Intel Burn Test & LL integrity hj47 Hardware 12 2010-01-26 11:08
v4_computer vs. Machine Name anomaly ADBjester PrimeNet 5 2008-11-12 17:07

All times are UTC. The time now is 00:15.


Wed Dec 1 00:15:59 UTC 2021 up 130 days, 18:44, 0 users, load averages: 1.24, 1.53, 1.56

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, 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.