mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Software (https://www.mersenneforum.org/forumdisplay.php?f=10)
-   -   mprime finished factoring almost immediately, possible HW problem? (https://www.mersenneforum.org/showthread.php?t=2681)

optim 2004-06-21 04:57

mprime finished factoring almost immediately, possible HW problem?
 
Hi, I use Mprime23.5 under SuSE9.1Pro. After I installed it on a second PC (Athlon2600@2135MHz) I configured it to request a 10M number from the server, and it did so: It got 34011449 at 5:06AM UTC. However, after some minutes, at 5:18AM UTC, Mprime finished processing and sent back the number as factored composite with the factor 7615643711941069249. Then, it got 34011451 which is testing now.

Is is normal for mprime to finish factoring so quickly? I have never seen this before and I am somewhat worried whether it may involve some hardware error. The machine (which is OCed) is new (I just finished building it and installed the OS some hours ago) and it went only 10-20 minutes of memtest86 (no problems found). I dont know how to read CPU/MB temps under Linux, but the CPU is surely under 66 celsius (the MB makes some sound when it reaches 66). Before installing suse, I used a default heatsink and the cpu reached >66 and probably 70 celsius, so I shutted it down and changed the heatsink with a new one which seems to work ok.

How could I repeat the factoring of this particular number in my second suse box (Athlon2600@2103MHz) which I know it has no hardware problems (running prime95/mprime since October2003)? Note my second suse box also uses mprime 23.5 (running in the background). I would like to refactor the number in that machine to be sure the result generated by my new box is right.

sdbardwick 2004-06-21 05:08

Someone will respond with a more detailed reply (please!), but it looks like normal behavior. You just got lucky and found a factor in short order.:smile: It is unusual for it to happen so quickly after starting an assignment, but I've had it happen more than once.

The primenet server verifies each reported factor, so if the exponent shows up in the reports as factored, there was no error on your end.

Unregistered 2004-06-21 06:50

The 10M numbers you get from Primenet are tested for factors up to 62 bits at the moment (By LMH). The first step mprime have to do is TF up to 68 bits since this is not done by Primenet in that range. The factor you found has less than 63 bits, thus it was found after few minutes.

To repeat this, try in a separate directory a worktodo.ini with

factor=34011449,62

and add the following line to the prime.ini before starting :

FactorOverride=68

(63 would also be sufficient in this case)

garo 2004-06-21 09:30

Do not put the factoroverride line in your prime.ini That setting is only for use by Lone Mersenne Hunters and not for use with Primenet. Otherwise, it looks like you got lucky and found a factor fast. Nothing to be worried about.

JuanTutors 2004-06-21 22:19

I think the probability of finding a factor of M34011449 between 2^62 and 2^68 is about 5%. (Is that a good guess?) Assuming that the probability of finding a factor of M34011449 stays relatively constant throughout the trial factoring (though it probably goes down a small amount since the size of the potential factors increases, and maybe other speed reasons) and estimating that with the Athlon2600@2135MHz it would take 48 hours to complete the factoring in that range (I think that's in the ballpark), the probability of finding a factor in the first 12 minutes is

(5/100)*(12/60)/48 = 1 in 4800, or 0.021%

Don't bomb my house if I'm way off, but you owe me money if I'm close.


All times are UTC. The time now is 16:42.

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