Report Results Here

[Uncwilly]

You might want to extend the choices for pre done TF on it too.

[lavalamp]

Originally Posted by James Heinrich

Unless you have the actual original results.txt data sitting around from all users, please send me the aggregate data so I can make sure all the bits get updated where they should. If for nothing else, there is no way on the results upload to backdate results but I can do that manually.

Almost all reports of ranges have been done rather informally on the forum, so no results files I'm afraid. Even if there were many would be in a different format from mfaktc.

I'm still compiling a list so that appropriate credit can be applied, which I can either send in its current format for you to manually apply, or I could fake it into a results file format for easier upload and auto verification. Whichever you'd prefer.

As far as work that has been done, then yes, the graph is accurate.

There's also this table:
http://2721.hddkillers.com/billion/

It has some advanced controls to allow it to display various things, but perhaps the other most useful mode is to display the factors found:
http://2721.hddkillers.com/billion/?...&highsearcher=

Originally Posted by James Heinrich

When I get a little time (probably tomorrow) I'll see about making a simplified non-anonymous OBD reservation system (and possibly results return) to make things easier.

That would be useful.

Once we get past this phase where the data is all up in the air and your site is up to date with all the known credit and progress, I don't see why we couldn't open up the range and say go nuts with it. We have a section that has been searched quite deep, and still needs more work, but then there are literally millions more candidate exponents with little to no work done as well. People could then pick and choose which type of work they want to do.

I'll still need a little more time to collect it all together first, so I'd appreciate if the floodgates weren't opened just yet.

Originally Posted by James Heinrich

Assuming that lava's graph is accurate and reflects the most recent effort data

As far as I know, that's completely up to date as of right now.

Also, to be clear, all of the data pages here are always in sync, so it doesn't matter which one is used as a source:
http://2721.hddkillers.com/

[James Heinrich]

Originally Posted by Uncwilly

You might want to extend the choices for pre done TF on it too.

I've extended it from 85 to 95. My projections up to 4294M only show as high as 2⁹² for a target GPU TF level, so that gives a little headroom.

[James Heinrich]

Originally Posted by lavalamp

Almost all reports of ranges have been done rather informally on the forum, so no results files I'm afraid. Even if there were many would be in a different format from mfaktc. I'm still compiling a list so that appropriate credit can be applied, which I can either send in its current format for you to manually apply, or I could fake it into a results file format for easier upload and auto verification. Whichever you'd prefer.

I would prefer the raw data in whatever format you have an I can massage it into the format I need. I can input things like the date work was done that you wouldn't be able to by "faking" results.

Originally Posted by lavalamp

perhaps the other most useful mode is to display the factors found

That's useful in that it gives me usernames to associate with found factors (although presumably that will all be included in the data dump you send me when you have it compiled). However, it does illustrate a minor problem: usernames used in OBD may or may not match GIMPS usernames. Some are identical (lavalamp), some I happen to know (Luigi=ET_), some are ambiguous (William=?, Joshua=?), some may be typos (Carlo?=Carlos), etc.
I would need to know the translation table for usernames to match GIMPS usernames so I can credit results properly. For reference, the unique usernames I found, with my guesses to the matches I think I know:

Aillas          = 
aketilander     = aketilander
Andres Aitsen   = Andres Aitsen
ATH             = ATH
bloodIce        = bloodIce
byroz           = 
Carlo           = 
Carlos          = 
CedricVonck     = 
Christenson     = 
Curtis          = 
dforsi          = 
dsouza123       = 
firejuggler     = firejuggler
gangho          = 
gjmccrac        = 
hemster         = 
Joshua          = 
Joshua2         = 
lavalamp        = lavalamp
Luigi           = ET_
monst           = monst
MrHappy         = 
Oliver          = TheJudger
Patrick Schmeer = 
Phinne          = 
Rde             = 
Rodrigo         = Rodrigo
thomasn         = 
Uncwilly        = Uncwilly
William         =

[James Heinrich]

Originally Posted by lavalamp

That would be useful

The reservation system, while slightly crude, is functioning now:
https://www.mersenne.ca/obd.php?f=0
TF pledge level defaults to current+1
Username is freeform, and will be remembered after you enter it the first time.

Reservations are for 30 days, and should be cleared automatically when a result is submitted to the pledged bitlevel (or above), but if you pledged multiple bitlevels (e.g. 83-85) and return a partial result (e.g. 83-84) the reservation will not be cleared since you haven't finished yet.


Would you be interested in letting me replicate your snazzy graph on my OBD page, credited to you of course? If not that's fine, I can roll my own, but yours looks nice.

[WGJC3107]

My GIMPS username is WGJC3107123

[ET_]

Aillas          = 
aketilander     = aketilander
Andres Aitsen   = Andres Aitsen
ATH             = ATH
bloodIce        = bloodIce
byroz           = A friend of lavalamp
Carlo           = Carlo Falcinelli (a friend of mine)
Carlos          = em99010pepe
CedricVonck     = A forumite
Christenson     = A forumite
Curtis          = VBCurtis I think
dforsi          = Daniele Forsi (another friend of mine)
dsouza123       = A forumite
firejuggler     = firejuggler
gangho          = A forumite
gjmccrac        = A forumite
hemster         = A forumite
Joshua          = 
Joshua2         = A forumite
lavalamp        = lavalamp
Luigi           = ET_
monst           = monst
MrHappy         = A forumite
Oliver          = TheJudger
Patrick Schmeer = 
Phinne          = A forumite
Rde             = A forumite
Rodrigo         = Rodrigo
thomasn         = A forumite
Uncwilly        = Uncwilly
William         = Wblipp (William B. Lipp, the owner of the OBD concept)

Here is some update, based on what I remember.

[ET_]

Thanks to lavalamp and James for the work done!

UID: ET_/9800X, no factor for M3321928373 from 2^83 to 2^84 [mfaktc 0.21 barrett87_mul32_gs]

On another side, when I check the exponent on https://www.mersenne.ca/exponent/3321928373 I see a hole between 75 and 83 bits. I suppose it is part of the integration job of lavalamp.

[lavalamp]

Originally Posted by James Heinrich

I would prefer the raw data in whatever format you have an I can massage it into the format I need.

Alright, this is taking a bit longer than I had initially expected/hoped, so it could still be a few days yet. I'm trying to be as complete and accurate as can be.

Originally Posted by James Heinrich

However, it does illustrate a minor problem: usernames used in OBD may or may not match GIMPS usernames. Some are identical (lavalamp), some I happen to know (Luigi=ET_), some are ambiguous (William=?, Joshua=?), some may be typos (Carlo?=Carlos), etc.

I'm afraid all I know are forum names and sometimes an alt name, like Luigi=ET_. I do have a short list of alternate names for people, and Luigi sent me his list of pseudonyms in CSV format, but there's really no guarantee they match with any GIMPS username. For instance I would suspect that neither Carlo nor Byroz ever joined GIMPS. Carlo a friend of Luigi, and Byroz a friend of mine, that we likely roped into helping the project and posted updates for/from them.

Originally Posted by James Heinrich

Would you be interested in letting me replicate your snazzy graph on my OBD page, credited to you of course? If not that's fine, I can roll my own, but yours looks nice.

By all means, I can send you the PHP code it's generated with if you like (is that still a language people use? ). I might want to make a few tweaks to the code first though as it has been 10 years since I wrote it.

Originally Posted by ET_

Thanks to lavalamp and James for the work done!

Happy to do it.

Originally Posted by ET_

On another side, when I check the exponent on https://www.mersenne.ca/exponent/3321928373 I see a hole between 75 and 83 bits. I suppose it is part of the integration job of lavalamp.

Yep, many holes to plug.

No factors from 83 to 85 bits for 3321930613.

Instead of reserving another exponent now, I am running a verification job. I am taking EVERY exponent, regardless of status from nothing to 72 bits in the following range:
3,321,928,000 - 3,322,000,000

I'm a sucker for round(ish) numbers, so this is from just before the OBD range, and all the way up to the end of Uncwilly's broadest search.

I know this is a duplication of effort, (and in some cases triplication), but during this process of data mangling I want there to be ZERO doubt about the status of any of the exponents. This is 3310 candidates but should only take about 20 hours.

[ET_]

Lavalamp, as I was one of the maintainers of the search, I can take part of the work of retesting the holes, so please take me into consideration.

And I love round numbers as well...

[ET_]

Originally Posted by lavalamp

No factors from 83 to 85 bits for 3321930613.

As per https://www.mersenne.ca/exponent/3321930613 you did not use James' bulk factors submission page https://www.mersenne.ca/bulk-factors.php