mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Lounge (https://www.mersenneforum.org/forumdisplay.php?f=7)
-   -   What happened to 6977699? (https://www.mersenneforum.org/showthread.php?t=364)

dswanson 2003-02-13 06:06

What happened to 6977699?
 
What happened to 6977699? Yesterday it was the last remaining assigned exponent less than 7M. Now it appears in neither the assigned exponent report nor the cleared exponent report. Where did it go?

Prime95 2003-02-14 18:35

It was accidentally poached (see bug thread in server section) by someone that did the P-1 factoring only. I have no way to reassign the exponent back to the original user. The best I could do was put the exponent in an unassigned state. The next time the double-checker reports in the exponent will be reassigned to him.

trif 2003-02-15 07:12

It was very close to expiry. I suspect it expired and Primenet had the same problem of thinking it is both done and not done that it has when someone using the factoring limits connects to Primenet.

lycorn 2003-02-15 21:48

[b]trif wrote:[/b]

[color=brown]... thinking it is both done and not done that it has when [u]someone using the factoring limits [/u]connects to Primenet.[/color]

What does that mean?

trif 2003-02-15 21:55

It's in the undoc.txt. Those of us who are doing the prefactoring work before stuff goes to PrimeNet use it to limit how far we factor to, and then we report those results manually to George. However, if someone has used it, and then changes Prime95 to use Primenet, it will report the results, and then PrimeNet gets confused and thinks the exponent is done, but also not done.

lycorn 2003-02-15 22:04

OK, now I know what you´re talking about. I also use it when prefactoring, and it is stated in undoc.txt that it shouldn´t be used with Primenet.
Thanks, trif.

wfgarnett3 2003-02-17 21:29

did anyone finish this yet?
 
This exponent is not listed in either the assigned or cleared exponents list; is someone finishing this exponent?

trif 2003-02-17 23:24

It is in the cleared exponents now, shaneamy picked it up. As this exponent vanished due to expiring, I don't have a problem with that.

NickGlover 2003-02-18 01:19

[quote="trif"]It is in the cleared exponents now, shaneamy picked it up. As this exponent vanished due to expiring, I don't have a problem with that.[/quote]

shaneamy's result did not match, though (see this week's hrf3.txt), so the exponent needs a 5th LL test.

dswanson 2003-02-18 04:53

[quote="NickGlover"]shaneamy's result did not match, though (see this week's hrf3.txt), so the exponent needs a 5th LL test.[/quote]
Shane's aware of the mismatch. He's rerunning the exponent. If anyone else is running it too, we'll have duplicated effort.

trif 2003-02-18 08:46

4 LL tests, and none match??? This exponent isn't even near an FFT boundary.

I know Shane has had some problems with bad results in the past, but they've all looked good for awhile.

garo 2003-02-18 21:44

Well,
If both matching results are his, one of us should step up and do a sixth one just in case. Let's keep an eye out shall we?
Garo

dswanson 2003-02-18 22:56

Does George count a result as a match if both halves of the match come from the same user? If one half of the match comes from Shane, and the other half comes from one of the earlier tests, I don't think there's a concern. Basically this would say that Shane found his first result to be wrong. But if both halves come from the same place, I think the concern is legitimate.

Darak 2003-02-18 23:27

George doesn't count any two matching results from the same account. The reason being (apart from possibility of cheating & maintaining a diversity of hardware/software involved in the test) is that any time you toggle report results to Primenet all the results will be resubmitted (hence 2 of the 4 results are by user SW).

I'm 99.99% sure the result I just turned in was the one bad given it didn't match either of the other 2. I was in the process of building a new machine and swapped some ram around in my machines and as it turns out which bank the ram populated made a difference in its stability. I had just fixed the problem when dswanson let me know the result I turned in didn't work.

I've already rerun the exponent and indeed the 2 results didn't match. I’m just waiting for the server to come back to report the result.

Prime95 2003-02-19 05:33

[quote="dswanson"]Does George count a result as a match if both halves of the match come from the same user?[/quote]

Yes, but they have to be separate runs (the shift_count value on the results line is different). Brian Beesley runs triple-checks on these in his spare time.


All times are UTC. The time now is 06:23.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2023, Jelsoft Enterprises Ltd.