mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Data (https://www.mersenneforum.org/forumdisplay.php?f=21)
-   -   passing on partially completed exponent to another user (https://www.mersenneforum.org/showthread.php?t=1319)

PLeopard 2003-10-26 09:45

passing on partially completed exponent to another user
 
Is it possible for me to pass on a partially tested exponent to another user? I've tested 65% of M33512231.
19 days to complete on my Athlon XP 2400+

GP2 2003-10-26 15:54

You'll kick yourself if it turns out to be a prime worth $100K.

Also, your prior exponent 33413467 had an errorcode of 00000001, so it's possible that this one has also accumulated an error or two (that's more likely to happen for large exponents). So someone else might inherit an already bad intermediate result.

Why not let it finish? If you're switching to another distributed-computing project, you can always keep GIMPS running in the background as a backup project.

PLeopard 2003-10-26 19:03

I didn't realize my previous exponent had an error. I've tested my machine thoroughly for memory problems using memtest86 and I used the Prime95 torture test and all ran without fault. Should I be worried about this?

GP2 2003-10-26 19:43

[QUOTE][i]Originally posted by PLeopard [/i]
[B]Should I be worried about this? [/B][/QUOTE]

Probably not, not yet anyway. With an errorcode of 00000001, the result still has a 77% chance of being correct.

If you're worried you can do a few double-check exponents for quick verification.

PLeopard 2003-10-26 21:10

how did you know my last exponent had that error code? I certainly can't find such info.
Anyway, I'll finish the exponent I'm working on... won't pass it on.

GP2 2003-10-26 21:49

You can find it in the HRF3.TXT data file (see the bottom of the page at [url]http://www.mersenne.org/status.htm[/url]).

You can also take a look at the results.txt file on your computer.

antiroach 2003-10-27 00:05

hmm i see one of my double checks in that HRF3 file. its code is all zero's. Is this good or bad?

GP2 2003-10-27 01:18

00000000 is what you want. No errors detected.

Note, your result could still be bad due to undetected errors. A few machines have been known to return plenty of erroneous results with 00000000 error codes.

Error codes of the form 00XX0000 and AB00AB00 are also relatively harmless.

PLeopard 2003-10-28 02:23

GP2, I just wanted to mention that I will let it finish, but yes, I am looking to switch to another DC project. Sure, I could run two on this machine but that's not much fun, as progress on both would be slow.

I'm a bit surprised at that error code I got on the last exponent. I thoroughly test my computer for problems and did so before beginning that exponent. As I said earlier, I ran Memtest86 through, and found no errors, then I did the Prime95 torture test and there were no errrors... so the error code is quite dissapointing. That sure makes a few months of work (on my older Athlon) feel worthless.

markhl 2003-10-28 17:06

One way to do it: e-mail George your pxxx save file and results.txt. I've done that before.


All times are UTC. The time now is 12:23.

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