Should MM31 be added to the search?

ET_ · 2012-09-23, 10:43

I strongly need your opinion.

Aketilander just pointed out that E.W.Mayer worked on MM31 up to 10¹⁵, and showed his will to take it to 86 bits.

1 - I can't test mmff because of lacking of a new-generation GPU: is the program ready to work on MM31?
2 - Do you think that the new double Mersenne number shoud be added to the reservations list?

Thanks in advance for your opinion.

Luigi

aketilander · 2012-09-23, 11:44

Quote:

Originally Posted by ET_

I strongly need your opinion.

Aketilander just pointed out that E.W.Mayer worked on MM31 up to 10¹⁵, and showed his will to take it to 86 bits.

1 - I can't test mmff because of lacking of a new-generation GPU: is the program ready to work on MM31?
2 - Do you think that the new double Mersenne number shoud be added to the reservations list?

Thanks in advance for your opinion.

Luigi

Thank you Luigi. I am using mfaktc for the TF as I do in Operation Billion Digits so there is no need to modify mmff. If more persons would like to work on MM31 I would be more then willing to share the work. If I can reserve it up to ^86 I would be more then happy though since the work is already in progress.

ATH · 2012-09-23, 13:49

MM31 doesn't work at the moment in mmff, it says:
"WARNING: exponents < 1000000 are not supported!"

which is wrong since MM31 = M2147483647 > 1000000.

ET_ · 2012-09-23, 15:00

Quote:

Originally Posted by aketilander

Thank you Luigi. I am using mfaktc for the TF as I do in Operation Billion Digits so there is no need to modify mmff. If more persons would like to work on MM31 I would be more then willing to share the work. If I can reserve it up to ^86 I would be more then happy though since the work is already in progress.

IIRC, mfaktc has some limitations to go high on k (was it 93 bits?),, nd is bit-level driven instead of k-driven...

Has Oliver (TheJudger) worked on a version that permits to choose k ranges?

Luigi

aketilander · 2012-09-23, 19:17

Quote:

Originally Posted by ET_

IIRC, mfaktc has some limitations to go high on k (was it 93 bits?),, nd is bit-level driven instead of k-driven...

Has Oliver (TheJudger) worked on a version that permits to choose k ranges?

Luigi

I think 95 bits: mfaktc Limits

... but I never think we'll reach that high!?

ET_ · 2012-09-23, 19:31

Quote:

Originally Posted by aketilander

I think 95 bits: mfaktc Limits

... but I never think we'll reach that high!?

That was just an evaluation. I'm more concerned about giving k ranges to mfaktc.

Luigi

Prime95 · 2012-09-23, 19:48

Quote:

Originally Posted by ET_

IIRC, mfaktc has some limitations to go high on k (was it 93 bits?),, nd is bit-level driven instead of k-driven...

Has Oliver (TheJudger) worked on a version that permits to choose k ranges?

mfaktc's limit is 2^92 factors.

Oliver is not working on k ranges. It would probably be easier to yank the relevant kernels out of mfaktc and add them to mmff.

ET_ · 2012-09-23, 23:01

Quote:

Originally Posted by Prime95

mfaktc's limit is 2^92 factors.

Oliver is not working on k ranges. It would probably be easier to yank the relevant kernels out of mfaktc and add them to mmff.

I see...

Well, I will wait until mmff is mature for production use, then I will kindly ask if you (or someone around here) is going to take the challenge... or explain to me where and how to work

I still have little knowledge of CUDA programming, but I'm eager to learn.

Meanwhile, the website will be ready to delivery whatever double Mersenne with p prime.

Luigi

View Poll Results: Should MM31 be dded to the search?
Yes, of course!	11	73.33%
No, we must point our resources to the four "chosen" numbers.	0	0%
I don't care, as long as Luigi updates everything...	3	20.00%
42	1	6.67%
Voters: 15. You may not vote on this poll

2012-09-23, 10:43	#1
ET_ Banned "Luigi" Aug 2002 Team Italia 2²×7×173 Posts	Should MM31 be added to the search? I strongly need your opinion. Aketilander just pointed out that E.W.Mayer worked on MM31 up to 10¹⁵, and showed his will to take it to 86 bits. 1 - I can't test mmff because of lacking of a new-generation GPU: is the program ready to work on MM31? 2 - Do you think that the new double Mersenne number shoud be added to the reservations list? Thanks in advance for your opinion. Luigi Last fiddled with by ET_ on 2012-09-23 at 10:46

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Credit being added to my account	Gordon	PrimeNet	34	2014-06-18 08:21
Has SSE2 recently been added to the msieve code?	EdH	Msieve	10	2012-10-04 18:08
Recently added	T.Rex	mersennewiki	20	2011-04-09 02:23
MM31 factoring	biwema	Operazione Doppi Mersennes	6	2005-07-06 00:27
calculation time not added to account-stats	ulikilian	Software	1	2004-10-06 23:10

2012-09-23, 13:49	#3
ATH Einyen Dec 2003 Denmark 3313₁₀ Posts	MM31 doesn't work at the moment in mmff, it says: "WARNING: exponents < 1000000 are not supported!" which is wrong since MM31 = M2147483647 > 1000000.