Holy new Mersenne prime, Batman! (M47 related) - Page 45

biwema · 2008-09-08, 21:40

Quote:

Originally Posted by ET_

Amazingly, on September 6th, another computer claims finding a new Mersenne prime!!
Independent verification has begun and should complete on the 11th.

Last updated: September 8, 2008

It seems that the second one is much smaller... :)

Luigi

It is just guessing, but can it be, that during the verification process only FFT sizes are used that are a power of 2? For example because it is not so easy to run a 5*2^(n-3) on massive parallel computers?

That could be a reason, why a 4096k FFT was used for M45 and now a 2048k could be enough if the exponent is below 39 Million or so?

That way it is easily possible that the verification of such a candidate is more than twice as fast.

By the way...
In the improbable case that there is every 10 days or so a new Mersenne prime is dicovered until Christmas, do we have to wait until then before all Primes are published?...

Batalov · 2008-09-08, 21:58

Quote:

Originally Posted by ET_

On another thought, did you notice that 17 Mersenne numbers over 100 million digits have been manually reserved for LL test, 15 of which by "ANONYMOUS"?

Is someone testing his new software or cluster?

Luigi

Someone probably just didn't care to set "OutputIterations=1" and may not have a slightest idea that it will take them several CPU-years for each test. But hey, who knows! There's a 1 in 2,300,000 chance that the exponents someone is testing would be an M50. That's a chance worth taking.

Prime95 · 2008-09-08, 22:42

Quote:

Originally Posted by biwema

In the improbable case that there is every 10 days or so a new Mersenne prime is dicovered until Christmas, do we have to wait until then before all Primes are published?...

Yes.

ixfd64 · 2008-09-08, 22:53

In that case, people would probably find out what the earlier primes are and end up leaking the information. Someone with a quad-core machine or two could probably test almost a dozen exponents in four months.

jinydu · 2008-09-09, 01:13

Quote:

Originally Posted by patrik

I ran P-1 again on those potential prime exponents that had only stage 1 done. No factors were found. All results were reported via primenet, except the last one which primenet for some reason does not accept.

This could be a new line of attack. Is there a way you could send results about the other 10 potential primes to primenet?

ixfd64 · 2008-09-09, 01:42

The residue of M37763179 (0xB141C151483C99__) also sticks out. Does anyone notice the pattern of ones?

jinydu · 2008-09-09, 02:17

Quote:

Originally Posted by Prime95

Please no screaming, but Scott is looking at September 15th.

Why was the date pushed back? Maybe they want to wait for 2 verifications to finish. Or maybe they had to move the verification to a slower computer. Or maybe it's just to throw off the sleuthers.

Prime95 · 2008-09-09, 02:41

Quote:

Originally Posted by jinydu

Why was the date pushed back? Maybe they want to wait for 2 verifications to finish. Or maybe they had to move the verification to a slower computer. Or maybe it's just to throw off the sleuthers.

Just speculating, but maybe it is because Scott has a press contact that works Monday through Friday.

jinydu · 2008-09-09, 04:36

Quote:

Originally Posted by patrik

I ran P-1 again on those potential prime exponents that had only stage 1 done. No factors were found. All results were reported via primenet, except the last one which primenet for some reason does not accept.

On the other hand, it might have been nothing more than luck. How about you try reporting it a few more times?

If it still doesn't work after a few more attempts... What error message do you get?

Andi47 · 2008-09-09, 05:09

Quote:

Originally Posted by Prime95

Just speculating, but maybe it is because Scott has a press contact that works Monday through Friday.

According to mersenne.org the verification of "M46" is due on Sept. 11th --> that's a Thursday.

But... wait! It only says "...on the 11th" without giving a month. Does it mean October 11th, which is indeed a saturday? This would correspond to my wild guess...

Quote:

Originally Posted by Prime95

Bingo. The report won't change unless Scott modifies the program that creates the report. I doubt that is a high priority for him.

But why was it not able to keep saying "Prime, UNVERIFIED, 2"?

Uncwilly · 2008-09-09, 05:47

Quote:

Originally Posted by Andi47

According to mersenne.org the verification of "M46" is due on Sept. 11th --> that's a Thursday.

Think about it, what will get most coverage that day? The 7th anniversary of the attacks. Ok, lets wait till Friday. That is often a day to release a story that one wants to hide. Mondays are better.

Quote:

But... wait! It only says "...on the 11th" without giving a month.

In situations like this, the preceding month spoken of is to be inferred.

2008-09-08, 22:53	#488
ixfd64 Bemusing Prompter "Danny" Dec 2002 California 4536₈ Posts	In that case, people would probably find out what the earlier primes are and end up leaking the information. Someone with a quad-core machine or two could probably test almost a dozen exponents in four months. Last fiddled with by ixfd64 on 2008-09-08 at 22:57

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
(New ?) Wagstaff/Mersenne related property	T.Rex	Wagstaff PRP Search	6	2019-11-23 22:46
Holy Speedup, Batman!	R.D. Silverman	NFSNET Discussion	4	2008-10-02 01:28
Holy Beaverpotamus, Batman!	ewmayer	Science & Technology	4	2008-03-14 19:19
holy tethered cow! new Mersenne prime? (M43-related)	ixfd64	News	265	2006-01-04 09:47
Mersenne prime related shirts and other items	adpowers	Lounge	40	2004-08-12 22:05

2008-09-09, 01:42	#490
ixfd64 Bemusing Prompter "Danny" Dec 2002 California 2×11×109 Posts	The residue of M37763179 (0xB141C151483C99__) also sticks out. Does anyone notice the pattern of ones?