GIMPS or GIMFS?

[davieddy]

I realize that any change to the layout in the V5 presentation
(Mersenne.org) would undermine the effort "we", the cognoscenti, have
made over three years to understand it and find some useful stuff,
BUT:
Surely in a distributed project such as this, a participant shouldn't
be required to pass the "Silverman" swottishness test to feel at home.

As a start, may I suggest that "recent results/cleared" might be separated
into > or < 100M?
I doubt if even the most avid GIMFS enthusiast could peruse the
interminable "exponent 500M does or not have a 65 bit factor" spam
with much interest.
For GIMPS fans, this garbage seriously inconveniences our ability to
seek what we want to know.

Secondly, any chance of mfaktc/GPUs getting a mention or recognition?

Thirdly, making the important distinction between factoring prudently
in preparation for LL testing, and GIMFS, aka finding as many factors
as quickly as possible for the hell of it?

David

[Christenson]

Moi, GPUs will get their recognition...when mfaktc turns in the completed work automagically. I'm working on it, but slowly.

First factors found are always more important than factors not found, since they represent LL tests that won't need to be done anymore. If the recent results/cleared report is getting too long, I propose sorting it by type of work turned in.

And as for being baptised by RDS, well, he wants his math pure, and lots of homework. He doesn't seem to understand that other things besides math are being gotten out of this exercise. We don't see him much in the lounge or the "information and answers" fora. Maybe we should give him a private hotline to the mods to get non-mathematics out of his math areas, and tell him to keep it private. That I can only name a few annoying forum members here is truly remarkable for internet bulletin boards.

[chalsall]

Originally Posted by davieddy

As a start, may I suggest that "recent results/cleared" might be separated into > or < 100M? I doubt if even the most avid GIMFS enthusiast could peruse the interminable "exponent 500M does or not have a 65 bit factor" spam with much interest.

For GIMPS fans, this garbage seriously inconveniences our ability to seek what we want to know.

Three points... First, it has always been the accepted policy of GIMPS to allow participants to do whatever kind of work they enjoy / find useful, with the exception of poaching.

Second, by the end of this year, or early next, all high 64 -> 65 "bit" factoring work will be completed. At this time, the hourly "Recent Results" report should show at least an hours worth of results (currently it shows about 40 minutes worth). Thus, it may not be worth the effort of George and Scott to make the change(s) you have suggested.

Thirdly, I specifically wrote the Change report over on the GIMPS Visualisation Tool to allow people to see the work that has been done in the specific ranges different people are interested in, be it TF or LL.

[petrw1]

Originally Posted by chalsall

Second, by the end of this year, or early next, all high 64 -> 65 "bit" factoring work will be completed. At this time, the hourly "Recent Results" report should show at least an hours worth of results (currently it shows about 40 minutes worth). Thus, it may not be worth the effort of George and Scott to make the change(s) you have suggested.

Nothing personal ... I too am somewhat optomistic --- however I recall thinking the same when we finished 62 bits and then 63. It seems that every time a bit level is finished something (or someone --- more someones) changes in the factoring world that continues to keep the hourly last 1000 results below one complete hour's worth.

Fingers crossed

[davieddy]

Originally Posted by petrw1

Nothing personal ... I too am somewhat optomistic --- however I recall thinking the same when we finished 62 bits and then 63. It seems that every time a bit level is finished something (or someone --- more someones) changes in the factoring world that continues to keep the hourly last 1000 results below one complete hour's worth.

Fingers crossed

As Bob says, I am an old dog who won't be bullied into new tricks.
I also forget things faster than I learn them these days.

But each bit level requires a doubling in computing.
I think this is an example of "exponential" or "non-polynomial" behaviour.

Now how many levels should we do before LL testing a 100M digit number?
And how long does it take to TF to 65 bits?

David

[davieddy]

Originally Posted by Christenson

factors found are always more important than factors not found, since they represent LL tests that won't need to be done anymore

Originally Posted by chalsall

Three points... First, it has always been the accepted policy of GIMPS to allow participants to do whatever kind of work they enjoy / find useful, with the exception of poaching.
...
Thirdly, I specifically wrote the Change report over on the GIMPS Visualisation Tool to allow people to see the work that has been done in the specific ranges different people are interested in, be it TF or LL.

Sorry for the snips but:
@Eric: isn't this tantamount to saying "Mersenne Primes are more
interesting than composites"?

@Chalsall: I think I have said this at least three times, I have
nothing against floating your own boat. Just don't have a collision
at sea.
I was to Leeward and on Starboard, and will raise the protest flag!

BTW I love your clear presentation of where we are.
So where is GIMFS ATM?

@Eric (again) I was shocked to find that our 45M baby was not prime,
after all our efforts (and George's vacation).
Currently got a 52M expo which I am P-1 ing ATM.
I would ask your GPU for its services again, but I think I'll upgrade
to Prime95 V26... and grab a better lottery ticket.

David

PS looks like the TF assignments <55M have settled in for good

[Christenson]

What I left out about factors being important was, ASSUMING you can find them....that is, with reasonable effort. And since there are FAR fewer Mersenne primes than Mersenne composites, of course they are more interesting.

Too bad about the 45M exponent...
If you want, I can still run you an extra TF bit level or three...whichever version of P95 you run....I just need all the digits in your 54M exponent and the starting point...I can start in 30-48 hours. By the way, I have put my money with my mouth...I've got part of one GPU doing an LL test this week, though, since there is uncertainty about the programming, it's only doing a double-check.

[Brian-E]

Originally Posted by davieddy

@Chalsall: I think I have said this at least three times, I have
nothing against floating your own boat. Just don't have a collision
at sea.
I was to Leeward and on Starboard, and will raise the protest flag!

Make it a good flag, give your boat plenty of attention, and enjoy it. But don't forget that the sea is there for everyone. And because there are many boats there, your boat won't stand out from the others in the eyes of other people even though it will stand out in your own eyes. Just as the other boats will be the pride and joy of their owners.

[davieddy]

Kevin Ayers