Wrong count at the ECM status page
 

today I finished and reportet >5000 curves stage 1 (stage 2 will be done with GMP-ECM) with B1=44e6 for each of the 4 exponents (2621, 2633, 2659, 2663).
[URL]http://www.mersenne.org/results/?exp_lo=2620&exp_hi=2700&limit=100[/URL]
At my accounts result-page everything is ok, but on the ECM status page were added only ~3000 curves @ 11e6:
[URL]http://www.mersenne.org/report_ecm/?txt=0&ecm_lo=1&ecm_hi=1&ecmnof_lo=2620&ecmnof_hi=2680[/URL]

It's pretty rare to find a factor in stage 1 ECM for numbers already trial-factored; until you run stage 2 on these curves, you haven't really contributed ECM.

hmmmm
as long as I did ECM on Mersenne numbers the GIMPS status page countet only the stages 1 of the curves. And as I said all stages 2 will be done with GMP-ECM (with B2~1000*B1 instead of only 100*B1, but only 1.5 times as long as with prime95). If I find a factor with those stages 2 you can be sure, you will hear from it :wink:. But if 20k+ curves say 'no factor' I don't post the GMP-ECM files...
The status page is only a help to decide which B1 is the best atm for a number. If the number shown there is to low, the next one who want to run ECM-curves on that number will choose a B1 to small.

The server did what it is supposed to do, you only uploaded the stage 1 results, so it calculates the probabilities based on that.

Why don't you report stage 2 results to the server?

[QUOTE=MatWur-S530113;382470]today I finished and reportet >5000 curves stage 1 (stage 2 will be done with GMP-ECM) with B1=44e6[/QUOTE]

The server calculated that 5000 curves at B1=44e6 is equivalent to 3000 curves at B1=44e6, B2=44e8. The ECM status page outputs the number of curves done assuming B2=100*B1.

When you finish B2 and report the 5000 curves B1=44e6, B2=lots*B1 then the server will again figure out the number of equivalent curves at B2=100*B1.

Note that this will "double count" the curves curves you just reported. This error in the table is not of great consequence. At worst someone might start doing curves at B1=110e6 at little earlier than optimal -- no big deal.

Because it is not possible.
Just for fun I tried to post my latest result file of one GMP-ECM-worker (~600 kB, ~600 results stage 2, to the manual results page of GIMPS, result:
[QUOTE][COLOR=Red]Did not understand 3,217 lines.[/COLOR][/QUOTE]I tried to post this file at James' page (mersenne.ca), too, same result...

Just an example of the file:

[CODE]Resuming ECM residue saved with Prime95
Input number is 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digits)
Using B1=1, B2=44000000-62879096296, polynomial Dickson(12), sigma=1536107296100755
Step 1 took 0ms
Step 2 took 92570ms
[/CODE]

So, tell me how an I will flood the server with some hundreds of MB in the near future (and will earn about more then 3 times the GHz-days for ECM)^^


addition: I read Prime95's answer in the preview of my post: The calculation is much clearer now, thank you. But how I have to post the stages 2 which I did with GMP-ECM ?

Does GMP-ECM post a summary at the end that totals the number of curves and the bounds? If so, we can get James to upgrade the manual results page.

If we cannot do that, then email the summary to me and your primenet userid and I can enter them into the manual results form under your userid. I make it look like the results come from prime95. The checksum won't be right, but the web form let's me bypass the checksum. I've done this for others. Fortunately, there aren't many doing the GMP-ECM thing.

[QUOTE=Prime95;382518]Does GMP-ECM post a summary at the end that totals the number of curves and the bounds? If so, we can get James to upgrade the manual results page.

If we cannot do that, then email the summary to me and your primenet userid and I can enter them into the manual results form under your userid. I make it look like the results come from prime95. The checksum won't be right, but the web form let's me bypass the checksum. I've done this for others. Fortunately, there aren't many doing the GMP-ECM thing.[/QUOTE]

'Of course' GMP-ECM does not post such a summary.:smile:
I remember that you already did this one time for me some years ago, but in this case it is not necessary.
I will pm James about this problem with some ideas for solution the next days. I'm sure we will find one.
Thanks for your time