Welcome to Lone Mersenne Hunters
 

Hi,

This forum is for all those who work on exponents that have not been assigned to them or others by the Primenet server. A number of people trialfactor large ranges of exponents to eliminate candidates as early as possible. Other people wish to do LL testing of exponents in ranges where they expect a higher chance to find a Mersenne Prime.
Check the postings over here to make sure you are not doing work already being worked on by someone else. Post here if you are trying something you want others to be aware of.

Happy hunting!

Tha.

Now I see that I`m not alone. We should change the name to orgranised Mersenne Hunters since we`re now in the open. I don`t have any computers running for trial factoring right now. I already turned in a few hundred thousands so far. My computer are working on some new ECM algorithms. I hope it turn out ok so I get back to Factor hunting.

This is a good idea!
We need to use some theory on this :rolleyes: though....
any ideas?

I still like "PrimeThinners", since that's more-or-less what you're doing.

OK, "PrimeCandidateThinners" to be picky, but it doesn't have the same ring....

9-)

[quote="QuintLeo"]I still like "PrimeThinners", since that's more-or-less what you're doing.[/quote]

There has been an assumption made that the first 10,000,000 digit Mersenne prime is very close to M41,000,000. Some people want to LL test in that region. They are welcome over here to post exponents they are working on too.

YotN,

Henk

[quote="tha"]

There has been an assumption made that the first 10,000,000 digit Mersenne prime is very close to M41,000,000.

YotN,

Henk[/quote]

Could you tell us more about that? Looking at the distribution of the first 38 or 39 Mersenne primes it would be difficult to predict the next one with a good accuracy (let's say less then 5,000,000). At least this is how I see it.

[quote="flava"]
Could you tell us more about that? Looking at the distribution of the first 38 or 39 Mersenne primes it would be difficult to predict the next one with a good accuracy (let's say less then 5,000,000). At least this is how I see it.[/quote]

From

http://www.mail-archive.com/mersenne@base.com/msg05046.html
M#40 was predicted well, M#39 apparently not.

M#39 - 53.7390% probability - range=10987349-11013853
M#39 - 64.0127% probability - range=10914203-11092621
M#39 - 81.6073% probability - range=10793527-11204183
M#39 - 97.3391% probability - range=10526447-11390453

M#40 - 61.4726% probability - range=13430227-13501387
M#40 - 77.3902% probability - range=13359163-13592549
M#40 - 86.0715% probability - range=13231913-13684399
M#40 - 96.5507% probability - range=13092361-13973117

M#43 - 58.3097% probability - range=41976841-42057331
M#43 - 71.6352% probability - range=41901683-42138559
M#43 - 79.7464% probability - range=41753977-42302809
M#43 - 93.4218% probability - range=41564021-42516373

THAnks 8)
I still wait for a Mersenne prime to pop out in the 10,000,000 - 12,000,000 range but it looks like the list is getting thin... it is very probable now that 13466917 is M#39

[quote="flava"]I still wait for a Mersenne prime to pop out in the 10,000,000 - 12,000,000 range but it looks like the list is getting thin... it is very probable now that 13466917 is M#39[/quote]
Alternatively, this may turn out to be an instance where a double-check turns up a prime that was missed the first time around. 10.9M - 11.1M might be an enticing range for someone to start doublechecking early...

[quote="tha"]
There has been an assumption made that the first 10,000,000 digit Mersenne prime is very close to M41,000,000. Some people want to LL test in that region. They are welcome over here to post exponents they are working on too.
[/quote]

I might note that I made the "assumption" based on some statistical analysis done... I might note that the prediction for 2^13,466,917 was also posted to the Mersenne mailing list a full 16 months before it was actually discovered...

That's why I have claimed the range 41,976,841 - 42,057,331 thru George... and have been doing full testing on it... I might mention that testing has been going on... since before 2^13,466,917 was discovered as well...

Anybody wishing to trial-factor, P-1 and/or L-L test the larger part of the predicted range (41,564,021 - 41,976,829 and 42,057,373 - 42,516373) is more than welcome to have at it... but as I have posted to the Mersenne mailing list... and the Yahoo list (now defunct)... there are NO guarantees or warranties offered up... Testing is at your own risk... of both success AND failure...

Eric

[quote="dswanson"]10.9M - 11.1M might be an enticing range for someone to start doublechecking early...[/quote]
I'll take myself up on my own offer. George, I'd like to reserve the range 10987349-11013853 for doublechecking. Is it sufficient to simply post that in this forum?

dswanson,
No posting on this forum is not sufficient. Also, one of the numbers in that range is still out for a first time test. You should probably email George to sort this out and/or look at HRF3.TXT from his status page to see which numbers you want to test.

Garo,

I've already pulled hrf3.txt and removed the one first-time test remaining in this range. I also pulled pminus1.txt and determined which ones still need P-1. I'll be doing that as well for the 61% of this range that hasn't yet been done.

I'll email George directly to officially reserve this range.

Dan

completely new topic - prolly a dumb question even

there's no way we can submit lmh results to get primenet credit, right?

also, is there a way to view our "gimps" credit as per what is read on the manual submission pages?

idk - it just seems kinda depressing at times to have that primenet number of p90 years stay at a total of about 22...

-Tom

I think George keeps an updated archive at

[url]http://mersenne.org/top.htm[/url]

Luigi

may i know what is this, so that i can also take part, as i am new sorry for such type of questions

[QUOTE=maheshexp]may i know what is this, so that i can also take part, as i am new sorry for such type of questions[/QUOTE]
Every kind of help is welcome. So are your questions. :smile:
In short form: We do [b]uncredited[/b] pre-factoring work for GIMPS. Some of us do it just for fun, some do it for the purpose of maximizing GIMPS throughput by putting machines (even old ones, like PII 200 etc.) to the kind of work they can do best.
But read the other sticky threads in this forum first...

i can't understand what this meant

34.0-34.2M 62 63 or 64 bit (next in Primenet)

[QUOTE=maheshexp]i can't understand what this meant

34.0-34.2M 62 63 or 64 bit (next in Primenet)[/QUOTE]

34.0-34.2M : a range of exponents, ie. 34,000,000 to 34,200,000

62 : factoring status of these exponents according the last nofactor.zip file, ie. most of the exponents are pre-factored to 62 bits

63 or 64 : wanted bits level for factoring (bold = a most wanted range)

next in Primenet : will be released to Primenet soon, ie. not much time left for factoring, frequent submitting of the results recommended

[QUOTE=hbock]In general P4 CPUs are more effective for factoring to 63 bit and higher (boost due to SSE2 optimization).[/QUOTE]Does this quote still strictly apply? With the recent optimizations related to the AMD 64, would an AMD 64 work as well as a P4 or is it still advantageous to do this work on a P4? Thank you.

I would say that with new optimizations etc. the Pentium-M and Athlon-64 are best for factoring.

I would say, it depends. (exception : AMD64 in 64 bit mode !)
I've collected some own data as well as data obtained with the AMD64 (look this [URL=http://www.mersenneforum.org/showthread.php?t=59&page=4&pp=50]thread[/URL] , obtained with version 24.11)
Please note that these are not the fastests version nowadays, you can get easily 10-20% (or even more) performance but the ranking will remain the same.
See picture added :

Hey that is a great graph. But if you stack the graph up next to the LL test times by the same machines you will find that AMD64 and Pentium-M are indeed more suitable. Particularly if we take into account the new factoring limits which mean that less time is spent on the bits above 64 which reduce the advantage of the AMD64 and P4.

Still, as the graph clearly shows, the AMD64 is teh 1ner.

Well, if we are talking about LMH ranges this will typically be, also in future, a factoring depth between 61 and 63 bits, ie. the following TF before the LL-test will take most of the time (>90%) for factoring from 64 to 67 bits or higher.
Nevertheless, everybody can do with his computer what he wants to do.
Maybe I'll add the graph of the LL timings for the same procs later, just as an additional for help for decision.