mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   LMH > 100M (https://www.mersenneforum.org/forumdisplay.php?f=46)
-   -   332.2M - 333.9M (aka 100M digit range) (https://www.mersenneforum.org/showthread.php?t=10693)

Uncwilly 2008-09-25 06:41

332.2M - 333.9M (aka 100M digit range)
 
I am nominally taking the range starting at 332,192,831 (the first 100M digit number) up to 333,999,999.

Most is already at 2^61 already. I will take the early ones up to at least 66. I will also bring the earliest to 68.

I'll be doing it in fits and spurts, using the v5 server to report , gather [URL="http://v5www.mersenne.org/report_factoring_effort/?exp_lo=332192831&exp_hi=332292831&bits_lo=0&bits_hi=999&txt=1&B1=Get+Data"]work[/URL], and get it assigned.

[B]Fellow workers are welcome.[/B] If we get a few upto ^74, so that they will be ready for P-1, that would be great.

ET_ 2008-09-25 09:16

[QUOTE=Uncwilly;143721]I am nominally taking the range starting at 332,192,831 (the first 100M digit number) up to 333,999,999.

Most is already at 2^61 already. I will take the early ones up to at least 66. I will also bring the earliest to 68.

I'll be doing it in fits and spurts, using the v5 server to report , gather [URL="http://v5www.mersenne.org/report_factoring_effort/?exp_lo=332192831&exp_hi=332292831&bits_lo=0&bits_hi=999&txt=1&B1=Get+Data"]work[/URL], and get it assigned.

[B]Fellow workers are welcome.[/B] If we get a few upto ^74, so that they will be ready for P-1, that would be great.[/QUOTE]

I'd suggest a sticky thread with some indications on "how to" start, and a couple lines to describe the project as Uncwilly did.

I will "manually" join s soon as I complete my small GNFS work.

Luigi

mdettweiler 2008-09-25 14:13

Just curious...isn't the 100 Million Digits Prefactor Project already doing this range? Or is that project officially dead by now?

Edit: Never mind, I just saw the message about this in the 100 Million Digit Prefactor Project thread. Silly me for posting before I looked. :rolleyes:

Uncwilly 2008-09-25 18:14

[QUOTE=Anonymous;143745]Just curious...isn't the 100 Million Digits Prefactor Project already doing this range? Or is that project officially dead by now?

Edit: Never mind, I just saw the message about this in the 100 Million Digit Prefactor Project thread. Silly me for posting before I looked. :rolleyes:[/QUOTE] Old 100Mdpp: :deadhorse: I'll try to get into the old yahoo site and kill it or point to this thread.



Anonymous, if you are the one grabbing up the ones for LL that I took to 64 or better, are you doing any TF on them? George indicated that they should go to 74, then P-1, then to 76.

After doing some intial broad work to get at least the first 100 or so (that have no factors found) to 66, I'll set one work thread to bring the early ones up to 68, then to 69, then 70 (I'll stop at the level where it takes ~24 hrs to TF one expo one bit level). The other worker thread will be doing runs to 66 down the range.

mdettweiler 2008-09-25 18:55

[quote=Uncwilly;143776]Old 100Mdpp: :deadhorse: I'll try to get into the old yahoo site and kill it or point to this thread.



Anonymous, if you are the one grabbing up the ones for LL that I took to 64 or better, are you doing any TF on them? George indicated that they should go to 74, then P-1, then to 76.

After doing some intial broad work to get at least the first 100 or so (that have no factors found) to 66, I'll set one work thread to bring the early ones up to 68, then to 69, then 70 (I'll stop at the level where it takes ~24 hrs to TF one expo one bit level). The other worker thread will be doing runs to 66 down the range.[/quote]
No, I haven't done any exponents whatsoever in that range. I think the mysterious "Anonymous" user who's doing those is some kind of default no-name user on PrimeNet v5, or something like that.

For future reference, my PrimeNet ID (v4 and v5) is "mdettweiler"--not "Anonymous". In fact, I probably would have changed my forum name to something different by now if only it was possible. Hmm...I was just thinking, can Xyzzy change forum login names? If so, I might be interested...

Uncwilly 2008-09-25 21:35

[QUOTE=Anonymous;143778]For future reference, my PrimeNet ID (v4 and v5) is "mdettweiler"--not "Anonymous". In fact, I probably would have changed my forum name to something different by now if only it was possible. Hmm...I was just thinking, can Xyzzy change forum login names? If so, I might be interested...[/QUOTE]Mister uberpower can. He changed somebody's a while back. It breaks all the quotes, etc.

Becareful what you wish for...............

:drama:

mdettweiler 2008-09-25 23:39

[quote=Uncwilly;143797]Mister uberpower can. He changed somebody's a while back. It breaks all the quotes, etc.

Becareful what you wish for...............

:drama:[/quote]
Oh my. :lol: LOL--funny, indeed. I've sent a PM to Xyzzy requesting that it be changed slightly so as to avoid confusion at the two prime-search projects which I assistant-administer. :smile:

Uncwilly 2008-09-29 16:11

[QUOTE=Uncwilly;143776]After doing some intial broad work to get at least the first 100 or so (that have no factors found) to 66, I'll set one work thread to bring the early ones up to 68, then to 69, then 70 (I'll stop at the level where it takes ~24 hrs to TF one expo one bit level). The other worker thread will be doing runs to 66 down the range.[/QUOTE]All 100 are to 66. The first 20, (except those assigned to others are to 74). The first 55 are assigned or up to 70. (Hopefully data will come in on those soon.)

Before I reload my comp. I will check back in with the v5 server.

Uncwilly 2008-10-11 00:12

I will avoid being like starqwest .

Here is a progress status report
[code]Date 10/10/2008
Average bit depth for first 100 expos 69.88
Average bit depth for first 1000 expos 64.928
100th active expo (no factor found) 332197631
1000th active expo (no factor found) 332237557
Unitless total effort number 267040[/code]

Uncwilly 2008-10-13 19:32

[Oct 13 12:00] Trial factoring of M332194417 to 2^70 is 70.40% complete.
[Oct 13 12:10] M332194417 has a factor: 1166327369238839882369

This expo was assigned as LL to someone else. I just got a factor. Hopefully they can get called off.

Any of the expos assigned to LL, if no TF effort is being reported to the v5 server, I will take them to 2^70, then 2^71, and maybe once I get far enough up to 2^72. I am trying to get all at least 100 expos to 2^72 with no factors found. I will continue working this range and working the first 1000 to 2^68 (then higher), then start bringing the first 5000 up to 2^65 (then higher).

Uncwilly 2008-10-17 00:32

[QUOTE=Uncwilly;145307][Oct 13 12:00] Trial factoring of M332194417 to 2^70 is 70.40% complete.
[Oct 13 12:10] M332194417 has a factor: 1166327369238839882369

This expo was assigned as LL to someone else. I just got a factor. Hopefully they can get called off.[/QUOTE]Picked off another.
[Oct 16 11:12] Trial factoring of M332194901 to 2^70 is 49.78% complete. Time: 781.652 sec.
[Oct 16 11:23] M332194901 has a factor: 1048513725453002816831

Uncwilly 2008-10-21 04:05

Killed another one in the 2^70 range., assigned to Anonymous.

[Oct 20 20:01] Trial factoring of M332195359 to 2^70 is 22.12% complete. Time: 773.382 sec.
[Oct 20 20:11] M332195359 has a factor: 717794085796277137231


Here is a progress status report
[code]Date 10/20/2008
Average bit depth for first 100 expos 70.6
Average bit depth for first 1000 expos 65.28
100th active expo (no factor found) 332197793
1000th active expo (no factor found) 332238073
Unitless total effort number 298144
Number of first 100 exos to 2^70 88
Number of first 1000 exos to 2^70 95[/code]

Uncwilly 2008-10-25 18:18

Yet another kill of one assigned to LL (to Anonymous on 10-05)

[Oct 25 08:53] Trial factoring of M332197897 to 2^67 is 49.16% complete.
[Oct 25 08:54] M332197897 has a factor: 116889392736489117791

BTW all of the first 100 are to 2^70

mdettweiler 2008-10-25 18:48

[quote=Uncwilly;146490]Yet another kill of one assigned to LL (to Anonymous on 10-05)

[Oct 25 08:53] Trial factoring of M332197897 to 2^67 is 49.16% complete.
[Oct 25 08:54] M332197897 has a factor: 116889392736489117791

BTW all of the first 100 are to 2^70[/quote]
Does anyone know if whoever's reserved these is still actually wasting his time on LLing these eliminated ones, or is he dropping them as they're being factored? :smile: (If he's still continuing on them, then I'd say he [i]really[/i] didn't know what he was getting into when he started work on these huge exponents! :smiel:)

Uncwilly 2008-10-25 22:07

[QUOTE=mdettweiler;146500]Does anyone know if whoever's reserved these is still actually wasting his time on LLing these eliminated ones, or is he dropping them as they're being factored? :smile: (If he's still continuing on them, then I'd say he [i]really[/i] didn't know what he was getting into when he started work on these huge exponents! :smiel:)[/QUOTE]Hopefully, he/she has their machine set to report updates to the v5 server every 14-28 days. The v5 server should report back to Prime95 to stop work on that expo. Prime95 (25.x of course) would then abandon that number. Then Prime95 and v5 would then negociate a new assignment, likely a 100M digit.

Monday I plan to bring another core on board for this range. It will do all of the range above 2^70, while I have another doing all of the 67-70, and a third bringing things up from the basement to the 67 or 68 range.

[COLOR="Green"][B]Other workers are welcome.[/B][/COLOR] If you are letting one core do LL, you can TF with the other nicely.

jinydu 2008-10-26 05:40

[QUOTE=Uncwilly;146529]Hopefully, he/she has their machine set to report updates to the v5 server every 14-28 days. The v5 server should report back to Prime95 to stop work on that expo. Prime95 (25.x of course) would then abandon that number. Then Prime95 and v5 would then negociate a new assignment, likely a 100M digit.[/QUOTE]

Really? I would think v5 would try to assign a "normal-sized" number...

Anyway, that exponent is now listed as factored on v5. No mention of ANONYMOUS any more.

[url]http://v5www.mersenne.org/report_exponent/?exp_lo=332197897&exp_hi=&B1=Get+status[/url]

Uncwilly 2008-10-26 06:36

[QUOTE=jinydu;146584]Really? I would think v5 would try to assign a "normal-sized" number...[/quote]If they chose a 100M number, they likely didn't change their options, so between Prime95 and v5, the 100M option will reign.

[quote]Anyway, that exponent is now listed as factored on v5. No mention of ANONYMOUS any more.[/quote] yep that happens when the factor is submitted.

mdettweiler 2008-10-26 13:38

[quote=Uncwilly;146590]If they chose a 100M number, they likely didn't change their options, so between Prime95 and v5, the 100M option will reign.[/quote]
But how could PrimeNet remember a "100M option" when there is no 100M option? Wouldn't it just see the 100M tests as cherry-picked exponents that are simply deviations from the "standard routine" of whatever that user has left set as their default? For example, recently I "cherry-picked" a 17M doublecheck test on v5 by manually inserting it into worktodo.txt, and as soon as it was done my machine went write back to its normal work preference of trial factoring work.
[quote]yep that happens when the factor is submitted.[/quote]I've noticed that whoever's crunching these would at least appear to be updating their progress with GIMPS reasonably frequently (at least, when I checked it yesterday, the 25th, all the ones I saw said 10/24 and 10/25 on the Exponent Status page--presumably that means the last time Prime95 updated progress to PrimeNet). So, maybe there's still a reasonable chance that this guy is getting the bad news (good news? :smile:) about his exponents after all. :smile:

Uncwilly 2008-10-26 14:26

1 Attachment(s)
[QUOTE=mdettweiler;146628]But how could PrimeNet remember a "100M option" when there is no 100M option?[/QUOTE]There is an option, if you log in to the v5 server.

markr 2008-10-26 20:42

1 Attachment(s)
[QUOTE=mdettweiler;146628]But how could PrimeNet remember a "100M option" when there is no 100M option?[/QUOTE]It shows up in the v25 client when it talks to the v5 server, too:

henryzz 2008-10-26 20:53

[quote=markr;146658]It shows up in the v25 client when it talks to the v5 server, too:[/quote]
is that a 25.7 addition

mdettweiler 2008-10-26 21:30

[quote=Uncwilly;146631]There is an option, if you log in to the v5 server.[/quote]

[quote=markr;146658]It shows up in the v25 client when it talks to the v5 server, too:[/quote]
Ah, I see now. As henryzz said also, is this new in v25.7 (and presumably added to the server at around that time)? I personally haven't had any reason to look in that drop-down box on the v5 web pages since before 25.7 was released, and I also haven't yet gotten around to upgrading to 25.7 yet (still on 25.6), so that would explain my ignorance in regard to this. :smile:

jinydu 2008-10-27 03:16

1 Attachment(s)
There's also this:

(But of course "100 million digits test is a subset of World Record tests")

mdettweiler 2008-10-27 03:45

[quote=jinydu;146718]There's also this:

(But of course "World Record tests is a subset of 100 million digits test")[/quote]
Hmm...I thought the world record tests option was just the new equivalent of the old 10 million digit option, namely, tests big enough to be the biggest Mersenne prime yet (though not huge, huge tests like the 100 million digit ones)?

CADavis 2008-11-11 08:27

I want to help with this but I don't know what to do to get work assigned to me. I assume I have to manual reserve ranges but I am not sure how. Does Primenet assign it or is it just do it by yourself and report results to Primenet?

Freightyard 2008-11-11 09:14

Might be interesting in trying a couple out as well.

Uncwilly 2008-11-11 19:59

1 Attachment(s)
Attached to this post is a worktodo.add file (named as worktodo.add.txt to get around the forum's restrictions). It does not have any [worker #x] lines in it, add them if you wish. Make sure that when you finally put it in your prime95 folder/directory that you save it as "worktodo.add", no .txt on the end.

This is for a moderate size range taking the expos from 2^64 to 2^68. It should only take a few days for a single core. It is in the area that I am currently working/tracking. No one else is assigned in this range. If you desire, edit the file and go to 2^70 or higher, but not above 2^74.

If you find that you want more or large assignments, post back and we can work out more or deeper (to higher bit depth) assignments.

Currently I am trying to get all expos to at least 2^70 before anyone picks them up for LL. I am doing this in 2 or 3 passes, the first up to ^67 or ^68 then the next pass to ^70.

Speak up if you are going to take this batch. I am going to post another in a moment.

Uncwilly 2008-11-11 20:06

1 Attachment(s)
This is in another range.
Again it is clear of other folks (the LL crowd).



I currently tracking all numbers up to 332259937.

Speak up if you take it.

PS: Make sure that your prime.txt file has lines[COLOR="SeaGreen"] like[/COLOR] these in it.[code]MaxExponents=400
UnreserveDays=365[/code]

CADavis 2008-11-12 05:49

[QUOTE=Uncwilly;148884]Attached to this post is a worktodo.add file (named as worktodo.add.txt to get around the forum's restrictions). It does not have any [worker #x] lines in it, add them if you wish. Make sure that when you finally put it in your prime95 folder/directory that you save it as "worktodo.add", no .txt on the end.

This is for a moderate size range taking the expos from 2^64 to 2^68. It should only take a few days for a single core. It is in the area that I am currently working/tracking. No one else is assigned in this range. If you desire, edit the file and go to 2^70 or higher, but not above 2^74.

If you find that you want more or large assignments, post back and we can work out more or deeper (to higher bit depth) assignments.

Currently I am trying to get all expos to at least 2^70 before anyone picks them up for LL. I am doing this in 2 or 3 passes, the first up to ^67 or ^68 then the next pass to ^70.

Speak up if you are going to take this batch. I am going to post another in a moment.[/QUOTE]

I am taking it. Going to 2^70.

CADavis 2008-11-14 07:43

[QUOTE=Uncwilly;148885]This is in another range.
Again it is clear of other folks (the LL crowd).



I currently tracking all numbers up to 332259937.

Speak up if you take it.

PS: Make sure that your prime.txt file has lines[COLOR="SeaGreen"] like[/COLOR] these in it.[code]MaxExponents=400
UnreserveDays=365[/code][/QUOTE]

I am taking this as well. I like doing this kind of factoring.:smile:

so I'm currently working on 332216147 - 332217971 from 2^64 to 2^70, as well as 332230021 - 332232977 from 2^64 - 2^70, and I will do both ranges from 70 to 74 after I finish going to 70. I have my athlon 64 x2 both cores working on it now, going to check if vista 64-bit is quicker factoring than xp 32-bit (i seem to remember xp 64-bit was a long time ago when i tried it)

Mini-Geek 2008-11-14 12:41

[quote=CADavis;149259]I am taking this as well. I like doing this kind of factoring.:smile:

so I'm currently working on 332216147 - 332217971 from 2^64 to 2^70, as well as 332230021 - 332232977 from 2^64 - 2^70, and I will do both ranges from 70 to 74 after I finish going to 70. I have my athlon 64 x2 both cores working on it now, going to check if vista 64-bit is quicker factoring than xp 32-bit (i seem to remember xp 64-bit was a long time ago when i tried it)[/quote]
I've heard 64-bit OSs, I'd assume whether XP or Vista or Linux, can TF about twice as fast as 32-bit OSs (which equates to one more optimal bit level).

Uncwilly 2008-11-14 14:19

[QUOTE=CADavis;149259]I will do both ranges from 70 to 74 after I finish going to 70.[/QUOTE]Point of info: the 72 bit level takes 1.4 Ghz days per expo. Thus 74 will take a Ghz week per expo. I would sugget that, because others will be signing up for LL tests, get a few to 70, then if you are using multiple cores, have one do the range to 70 and have the second go to 72, then a thrid pass to 74. If anyone signs up to LL a number let them have it.

CADavis 2008-11-14 19:43

[QUOTE=Mini-Geek;149292]I've heard 64-bit OSs, I'd assume whether XP or Vista or Linux, can TF about twice as fast as 32-bit OSs (which equates to one more optimal bit level).[/QUOTE]

Yeah I switched to Vista 64-bit and it's going approx. twice as fast :-)
Is there a way to make the benchmark go to 74bit factors?

petrw1 2008-11-14 22:44

[QUOTE=CADavis;149351]Yeah I switched to Vista 64-bit and it's going approx. twice as fast :-)
Is there a way to make the benchmark go to 74bit factors?[/QUOTE]

Whoo Hoo I just ordered a new Q9550 with Vista64 ... due Tuesday.

Watch me factor now.:flex::flex::flex:

CADavis 2008-11-16 05:19

I think I'm changing my mind a bit. I figured out how to use the Factoring Effort page to make a worktodo file, and I think I want to do a wide range to lower levels.

So what I'm thinking now is to take my current range to 68 bits, then to 70 bits (maybe not though), then start on a new range from 61 - 64 bits

the new range will be starting at 332900021 and eventually go to 332999959, which is all only done to 61 bits.

what do you think, Uncwilly?

Uncwilly 2008-11-16 06:45

Fair enough. I haven't been tracking that area at the moment.


I am currently trying to clear areas ahead of the LL'ers. I have a few cores doing expos (even those assigned, but showing no signs of more TF) up to 73.


It looks like someone found a factor up in the 73 bit range. :)

jinydu 2008-11-19 06:28

I've found the usernames of two 100M digit LL testers:

simcon M332193833 and M332193859
jr007j M332194253 and M332194277

Uncwilly 2008-11-19 19:06

[QUOTE=jinydu;149830]I've found the usernames of two 100M digit LL testers:

simcon M332193833 and M332193859
jr007j M332194253 and M332194277[/QUOTE]There are several others that have expos assigned to them. Some are active, some seem not to be.

Uncwilly 2008-11-21 06:35

1 Attachment(s)
Here is a progress status report for the range from 332192831 to 332259937

[code]Date 11/20/2008
Average bit depth for first 100 expos 71.46
Average bit depth for first 1000 expos 67.96
100th active expo (no factor found) 332197793
1000th active expo (no factor found) 332239723
Unitless total effort number 813536
Number of first 100 exos to 2^71 45
Number of first 1000 exos to 2^70 275[/code][code]Bit # to bit level or higher
61 1527
62 1259
63 1258
64 1258
65 900
66 849
67 836
68 571
69 280
70 275
71 107
72 84
73 72
74 48
75 10
76 1
77 1[/code]

James Heinrich 2008-11-22 20:29

I grabbed [url=http://v5www.mersenne.org/report_exponent/?exp_lo=332203901]M332203901[/url] a couple weeks ago. So far I've completed TF from 2^71-2^75, as I write this P-1_stage1 is 60% done (P-1 both stages should be done around 10-Dec, then I'll (probably) let it continue TF up to 2^77, which will take me into Feb2009(?). I'll unreserve it once it gets to LL time; I don't have the patience to wait until May 2013 to confirm it's not prime :razz:

James Heinrich 2008-11-22 20:50

I noticed that M332252939 to M332292827 have currently only been TF'd to 2^61, so I figured it would only take me about 12 hours to bring those 1000 exponents up to 2^63, so probably by the time anyone reads this they'll all be at 2^63. Hopefully I'm not stepping on anyone's toes.

James Heinrich 2008-11-23 12:30

[QUOTE=James Heinrich;150276]I noticed that M332252939 to M332292827 have currently only been TF'd to 2^61[/QUOTE]1000 exponents TF'd to 2^63; 37 factors found.

Uncwilly 2008-11-24 00:42

[QUOTE=James Heinrich;150276]I noticed that M332252939 to M332292827 have currently only been TF'd to 2^61, so I figured it would only take me about 12 hours to bring those 1000 exponents up to 2^63, so probably by the time anyone reads this they'll all be at 2^63. Hopefully I'm not stepping on anyone's toes.[/QUOTE]You stepped on my toes a bit there. But no problem. I was taking the range to M332259937 up to 65. I saw your message and dropped the range. I put the core on to a different range to higher level.

I generally peek at the exponent status report to see if it has been assigned (and then if it has shown activity, a lot of the L-L'ers have shown no progress on the supporting TF or P-1).

CADavis 2008-11-26 05:47

[QUOTE=CADavis;149501]then start on a new range from 61 - 64 bits

the new range will be starting at 332900021 and eventually go to 332999959, [/QUOTE]

done, 332.9M - 333M to 64 bits, 116 factors found

CADavis 2008-11-26 05:56

starting again with 332216147 - 332217971, most are to 68 already, going to take the range to 72 first, then 72 - 74. the first few are assigned to LL testers already so should I bother with those at all? [URL="http://v5www.mersenne.org/report_exponent/?exp_lo=332216147&exp_hi=332217971&B1=Get+status"]http://v5www.mersenne.org/report_exponent/?exp_lo=332216147&exp_hi=332217971&B1=Get+status[/URL]

S485122 2008-11-26 06:43

[QUOTE=CADavis;150761]the first few are assigned to LL testers already so should I bother with those at all? [URL]http://v5www.mersenne.org/report_exponent/?exp_lo=332216147&exp_hi=332217971&B1=Get+status[/URL][/QUOTE]I think that once an exponent is assigned by the V5 server none other than the assignee should touch it. Of course in the 100 000 000 digit primes range some of the assignees will skip trialfactoring in the misunderstanding that they will go quicker. I still think it is better to just wait until they drop out.

Jacob

Uncwilly 2008-12-01 03:55

[QUOTE=S485122;150767]I think that once an exponent is assigned by the V5 server none other than the assignee should touch it. Of course in the 100 000 000 digit primes range some of the assignees will skip trialfactoring in the misunderstanding that they will go quicker. I still think it is better to just wait until they drop out.[/QUOTE]If they show no progress after the number has been assigned to them for 15-30, I think that it is likely that they have skipped the TF.

S485122 2008-12-01 06:17

[QUOTE=Uncwilly;151423]If they show no progress after the number has been assigned to them for 15-30, I think that it is likely that they have skipped the TF.[/QUOTE][size=+2][b]likely[/b][/size]

You are not sure and thus may be poaching. There is no urgency to treat those numbers. Even if it might be fun to search for the work ;-)

Jacob

cheesehead 2008-12-01 08:05

[quote=Uncwilly;151423]If they show no progress after the number has been assigned to them for 15-30, I think that it is likely that they have skipped the TF.[/quote]I'll echo Jacob (though more softly and verbosely).

When one tries to deduce what is going on in an assignee's system, one runs the risk of having one's invalid assumptions about what is and is not shown in the reports lead one to a false conclusion.

That was doubly true in my own case.

Several years ago, the nonstandard way in which I was processing my LL assignments may have made it appear to someone looking at PrimeNet reports that I would be "holding up" some milestones, but in fact I was careful not to do that.

Then, the way in which someone else reported LL results to PrimeNet for exponents that had been assigned to me made it appear to me that the person was deliberately poaching my assignments. However, after I recently established communication with that person, he emphatically assured me that he did nothing of the sort, and had tested only exponents that had been properly assigned to him. That made me realize that I had not previously considered the possibility that all the evidence I had could be explained by some flaw in the assignment system rather than by poaching.

So, please don't do anything on exponents that are assigned to someone else. There are an infinity of Mersenne numbers with prime exponents, enough for everyone to work on exclusively with plenty left over.

Uncwilly 2008-12-21 05:35

1 Attachment(s)
I have taken the advice of others to heart. I now grab the factor level report, make the worktodo.add, let Prime95 register assignments, then cull the N/A's.

Here is a progress status report for the range from 332192831 to 332259937

[code]Date 12/20/2008
Average bit depth for first 100 expos 72.17
Average bit depth for first 1000 expos 69.23
100th active expo (no factor found) 332197793
1000th active expo (no factor found) 332240201
Unitless total effort number 1556864
Number of first 100 exos to 2^71 72
Number of first 1000 exos to 2^70 370[/code][code]Bit # to bit level or higher
65 1482 <-all left in range
66 1234
67 1233
68 710
69 500
70 370
71 219
72 187
73 119
74 83
75 43
76 1
77 1[/code]

S485122 2008-12-23 14:23

[code]Exponent Range Composite Unproven Assigned Available
Start Count F NO-LL TF P-1 LL TF
332000000 51080 27401 23679 648 1 653 22398[/code]This shows that 21 assignments are overlapping (Unproven -Assigned-Available)

Jacob

Uncwilly 2008-12-23 23:32

[QUOTE=S485122;154726]This shows that 21 assignments are overlapping (Unproven -Assigned-Available)[/QUOTE]I think that there are some expos that the server shows as assigned to me to LL, that are actually left over TF's. I want to get my hands on the machine and look at the worktodo.txt. I want to make sure that the entries are right, or fix them, or.... I haven't figured out how it happened. They may be part of the 21.

That and maybe the ones that got assigned LL, but TF killed, as reported previously.

MercPrime 2009-01-09 05:16

Hey UncWilly, I'll join in when I'm done with my poking around below your range (321M 6-9 range).

Uncwilly 2009-01-21 03:14

1 Attachment(s)
Here is a progress status report for the range from 332192831 to 332259937

[code]Date 01/20/2008
Average bit depth for first 100 expos 72.69
Average bit depth for first 1000 expos 70.09
100th active expo (no factor found) 332197793
1000th active expo (no factor found) 332240677
Unitless total effort number 2309376 (accounts for only the 1st 1000 expos)
Number of first 100 exos to 2^71 96
Number of first 1000 exos to 2^70 394[/code][code]Bit # to bit level or higher
67 1461 <-all left in range
68 1300
69 766
70 394
71 310
72 249
73 170
74 120
75 81
76 2
77 1[/code]
Some of my cores have been sidelined from this due to lack of access. They are still working for GIMPS. I have put one new core to working the range above this. It is a slower one and I will keep it at the lower bit depths. Once I get all the exponents in this range to 2^68, I will put that core onto the next higher area, bringing those expos up to 67 then 68. I plan on reviving the other cores and have then work the 69-72 area for this range.

Uncwilly 2009-01-21 23:01

[QUOTE=Uncwilly;159660]Here is a progress status report for the range from 332192831 to 332259937[/QUOTE] I am now extending this particular range up to 332319979. The next stats (2/20) will reflect that. Anyone wanting to help can either work above me or take some of the higher bit levels.

shadowflicker 2009-02-01 08:57

I'm more than happy to devote the little bit of spare computing power I have to this.

Let me know if there is a specific [sub]range, exponent, and/or bit depth you would like me to work towards; otherwise I'll keep choosing semi-randomly.

Uncwilly 2009-02-02 01:26

If you are inclined toward lots of small work, I would suggest doing the range directly above 332319999. There are lots of low bit levels to work on. Depepnding on you processor, I would suggest that you move them up to 64 or 65 as a first step.

If you are inclined to slower, bigger chunks of work, then use the techniques in this thread: [url]http://www.mersenneforum.org/showthread.php?t=11308[/url]
especially step 8, to do work up to 70 bits and above within the current range of 332192831 to 332319999.

What ever you do decide to do, make sure that you follow the steps in the cited thread (so that you don't step on anyone's toes). And post what you are working on here.

shadowflicker 2009-02-02 04:34

[B]I'll plan on taking 332320000 - 332330000 to 64[/B] for starters.

I'm primarily using WinXP x32, for which I'd rather not use for TF work.
I also have Vista x64 (dual-boot), which I set Prime95 up on last night mostly to make sure I understood the config/techniques to do this correctly. Once it finishes TF on a small number of assignments in the 333 range, I'll start working on the range I mentioned above.

My relevant hardware:
Intel Core 2 Duo E8500
4GB DDR2 800

I'm inclined to set one core to TF, and let the other utilize a large amount of RAM for P-1. Any opinion on that?

Uncwilly 2009-02-02 05:20

[QUOTE=shadowflicker;161349][B]I'll plan on taking 332320000 - 332330000 to 64[/B] for starters.[/quote]
Fair enough. I will set up to track that. It is about 250 expos. Going from 61 to 64 on that machine shouldn't take too long.

[quote]I'm inclined to set one core to TF, and let the other utilize a large amount of RAM for P-1. Any opinion on that?[/QUOTE]Sounds ok. Check out this thread: [url]http://www.mersenneforum.org/showthread.php?t=11387[/url]

If you are going to work in the 100m digit range, check for expos that are at the 75 bit level.

shadowflicker 2009-02-03 22:50

Given I get no complaints over the electric bill, I'll increase the range I'm working on to [B]332320000 - 332399999[/B].

As for the expos already at 75 bits (within your range), it seems most have LL testers assigned to them already. I'm not sure if I have the patience for a P-1 test with expos this large, but it seems I would have to go for an expo already being LL tested should I want to find out.

Uncwilly 2009-02-04 00:54

[QUOTE=shadowflicker;161481]Given I get no complaints over the electric bill, I'll increase the range I'm working on to [B]332320000 - 332399999[/B].[/quote]Getting 'kills' is fun at this level, huh? Between the lot of us there were 10 in the last day or less. I have extended my tracking upto the new limit. Currently 4815 active (no factor found) exponents.

[quote]As for the expos already at 75 bits (within your range), it seems most have LL testers assigned to them already. I'm not sure if I have the patience for a P-1 test with expos this large, but it seems I would have to go for an expo already being LL tested should I want to find out.[/QUOTE]You can hold onto one that you have taken to 64 and with a single core get it assigned to TF to 75, then P-1, then TF to 77. Just, put them in successive lines in your worktodo.add.

On the next monthly report, there should show some good progress.
The "Unitless total effort number" line in the report reflects the amount of effort, based on 2^61 level = 1 unit, 2^62 = 2, 2^63 = 4, etc. I have increased the range of this to all the active area.

Uncwilly 2009-02-21 20:06

Here is a progress status report for the range from 332192831 to 332399999

[code]Date 02/21/2009
Average bit depth for first 100 expos 72.74
Average bit depth for first 1000 expos 70.47
100th active expo (no factor found) 332197793
1000th active expo (no factor found) 332241307
Unitless total effort number 2982222 (accounts for only the 1st 4800 expos)
Number of first 100 expos to 2^71 97
Number of first 1000 expos to 2^70 470[/code][code]Bit # to bit level or higher
61 4717 <-all left in range
62 3893
63 3892
64 3891
65 2523
66 2160
67 2159
68 1537
69 1073
70 472
71 360
72 278
73 199
74 149
75 108
76 2
77 2[/code]

My personal core availablity continues to flucuate. I plan to remove one from DC's to TF in the 100m range.

Uncwilly 2009-03-21 05:27

1 Attachment(s)
Here is a progress status report for the range from 332192831 to 332399999

[code]Date 03/20/2009
Average bit depth for first 100 expos 72.81
Average bit depth for first 1000 expos 70.66
100th active expo (no factor found) 332197793
1000th active expo (no factor found) 332241479
Unitless total effort number 3392664
Number of first 100 expos to 2^71 97
Number of first 1000 expos to 2^70 547[/code][code]Bit # to bit level or higher
64 4623 <-all left in range
65 3948
66 2686
67 2543
68 1696
69 1292
70 561
71 410
72 298
73 215
74 164
75 125
76 4
77 2[/code]

Uncwilly 2009-04-22 02:18

Here is the monthly status report for the range from 332192831 to 332399999

[code]Date 04/21/2009
Average bit depth for first 100 expos 72.94
Average bit depth for first 1000 expos 70.85
100th active expo (no factor found) 332197793 (no change since 10/20/08)
1000th active expo (no factor found) 332241691
Unitless total effort number 3994208
Number of first 100 expos to 2^71 97 (static for 2 months)
Number of first 1000 expos to 2^70 610[/code][code]Bit # to bit level or higher
65 4581 <-all left in range
66 4349
67 2994
68 1808
69 1286 <- yes this is lower
70 679
71 476
72 344
73 247
74 181
75 120
76 9
77 5[/code]
Great strides have been made this month by those signed up to LL.

Would a chart with x at y bit depth chart be better?

James Heinrich 2009-04-22 10:37

[QUOTE=Uncwilly;170431]Would a chart with x at y bit depth chart be better?[/QUOTE]I think it would be (make one, then we can judge better :smile:)

BTW: I've grabbed 47 exponents (332290237-332292677) and upgrading them from 2^67 to 2^68; they should be done within a week or so.

Uncwilly 2009-04-22 23:32

1 Attachment(s)
[QUOTE=James Heinrich;170499]I think it would be (make one, then we can judge better :smile:)[/quote]
Attached I have an image of what it might look like. This is a pretty excel sheet. The raw stats will be uglier

[quote]BTW: I've grabbed 47 exponents (332290237-332292677) and upgrading them from 2^67 to 2^68; they should be done within a week or so.[/QUOTE]As long as you played by the rules and didn't take any that are assigned (thanks George for that switch)... that is fine. V5 serve makes it easy.

James Heinrich 2009-04-23 00:39

1 Attachment(s)
You can make pretty graphs with Google, like this (attached).
I could make a quick tool for you to generate the chart if you don't feel like editing cryptic strings, just lemme know, would take 5 mins to put together.

(generated, for my own reference, with this:)[code]http://chart.apis.google.com/chart?chxt=y,r&chxl=0:|P-1|77|76|75|74|73|72|71|70|69|68|67|66|65|64|63|62|61|1:|7|5|4|131|42|70|92|132|207|603|522|1205|1333|232|0|0|0|0&cht=bhs&chd=t:0,0,0,0,232,1333,1208,522,603,207,132,92,70,42,131,4,5,7&chco=76A4FB&chs=400x200&chbh=r,0.3&chds=0,1333[/code]

Uncwilly 2009-04-23 06:01

1 Attachment(s)
[[B][COLOR="Red"][SIZE="4"][FONT="System"]rant[/FONT][/SIZE][/COLOR][/B]]I have spent about 2.5 full days this month fighting with excel at work to get graphs.:rant:
Specifically because of excel's poor graphing skills. :furious: I know of a program/package that would do in less than an hour (that includes print time) what I and a colleague take over 12 man days a year to do.
For example, one can't specify that all charts on a sheet or workbook have a specifed single range, nor use a cell reference as a value for a range.:rant:

[quote]"You graph with the programs that you have not the programs that you wish you had.":raman:[/quote]
[[B][COLOR="Red"][SIZE="4"][FONT="System"]/rant[/FONT][/SIZE][/COLOR][/B]]

Here is the pig all tarted up for you.

James Heinrich 2009-04-23 12:32

[QUOTE=James Heinrich;170627]I could make a quick tool for you to generate the chart if you don't feel like editing cryptic strings[/QUOTE]Whether you wanted it or not, I made it anyways :cool:
[url]http://mersenne-aries.sili.net/uncwilly.php[/url]

ET_ 2009-04-23 13:24

[QUOTE=James Heinrich;170705]Whether you wanted it or not, I made it anyways :cool:
[url]http://mersenne-aries.sili.net/uncwilly.php[/url][/QUOTE]

BUG! :smile:

Maximum graph size Google allows is 300k pixels (554x554 = 306,916 pixels)

but 300k = 300*1024 = 307,200

I'm just kidding...

Luigi

James Heinrich 2009-04-23 23:32

[QUOTE=ET_;170709]but 300k = 300*1024 = 307,200[/QUOTE]Sorry, no -- 300k = 300000 (whereas 300[url=http://physics.nist.gov/cuu/Units/binary.html]Ki[/url] == 307200)
Nevertheless I have changed the wording in my error message to be more clear :smile:

unconnected 2009-05-02 23:45

I'll advance 332380000-332399999 from 66 to 67 bit.

James Heinrich 2009-05-03 23:25

[QUOTE=Uncwilly;148884]If you desire, edit the file and go to 2^70 or higher, but not above 2^74.[/QUOTE][QUOTE=Uncwilly;143776]George indicated that they should go to 74, then P-1, then to 76.[/QUOTE]In [i]commonc.h[/i] the trial-factoring limits are defined:[code]#define FAC80 516000000L
#define FAC79 420400000L
#define FAC78 337400000L
#define FAC77 264600000L
#define FAC76 227300000L
...[/code]For this range of exponents, TF should go up to 2^77.

I believe George recently (between [url=http://mersenneforum.org/showpost.php?p=168654&postcount=155]2009-Apr-09[/url] and [url=http://mersenneforum.org/showpost.php?p=170219&postcount=180]2009-Apr-20[/url]) also changed from having P-1 done at TF-2 to TF-1, so pre-P-1 TF should be to at least 2^75, but probably 2^76.

Uncwilly 2009-05-04 01:05

[QUOTE=James Heinrich;172162]In [i]commonc.h[/i] the trial-factoring limits are defined:
For this range of exponents, TF should go up to 2^77.

I believe George recently also changed from having P-1 done at TF-2 to TF-1, so pre-P-1 TF should be to at least 2^75, but probably 2^76.[/QUOTE]
Yes, yes. I think that the range was increased to 77 after I saw a posting. Also, the change to 76 for P-1 has been recent.

James Heinrich 2009-05-04 11:13

I've grabbed 10 exponents (332195837-332249327) to extend from 2^72/73 up to 2^76. Should take about a month each on a single core of my Phenom. That'll keep that core busy for the next year. :cool:

Uncwilly 2009-05-04 14:45

I've just notice an almost overnight jump in ~400 expos from 2^66 to ^67. This has me stumped.

Was that you shadowflicker dumping a bunch of results?

Uncwilly 2009-05-13 01:20

100mdpp's most wanted
 
Here is a list of our most wanted expos. These are the ones that are the stragglers. Each of the listed numbers are the first number in this range that are not already at the indicated bit depth. I would appreciate anyone that already has these numbers to report some work on them. There are some that are unassigned, anyone that wants to grab one, that would be cool.

332192897 77
332192897 76
332192957 75
332193457 74
332194069 73
332194939 72
332194939 71
332204699 70
332208827 69
332282557 68
332309833 67

This list was generated automagically. I had troubles getting the exsmell formulas right, then I found something that I had forgotten about and viola!

~8days until the swine make-up fest begins again.

henryzz 2009-05-13 05:18

what does the 6809 > 6502 under your name mean?

Uncwilly 2009-05-13 05:36

1 Attachment(s)
[QUOTE=henryzz;173378]what does the 6809 > 6502 under your name mean?[/QUOTE]It means.....:google:
I didn't put it there, the gerbils (r.i.p.) did.

10metreh 2009-05-13 06:17

You know, I'd never spotted the "Mr Burns" on that smiley.

Uncwilly 2009-05-21 06:02

1 Attachment(s)
Here is the monthly status report for the range from 332192831 to 332399999

[code]Date 05/20/2009
Average bit depth for first 100 expos 72.99
Average bit depth for first 1000 expos 71.24
1000th active expo (no factor found) 332241941
Unitless total effort number 4728192
Number of first 1000 expos to 2^70 758[/code][code]Bit # to bit level or higher
65 4556 <-all left in range
66 4324
67 3894
68 2299
69 1412
70 887
71 520
72 393
73 297
74 220
75 159
76 14
77 6
P-1 11[/code]
The number of exponents assigned to LL continues to bounce up and down.
All of the expos at 65 bits are currently assigened to others.

Attached is a bar chart with the progress of the exponents. The large spikes at 61 bits are from when the range was expanded. P-1 status was not tracked before 4/20, thus it is not included.

Uncwilly 2009-05-27 01:23

Who dunnit?
 
Who just picked up ~7000 expo in 332M and 333M range??????????

Is this serious?

And what sort of power are you throwing at this?

If it is real the help is welcome.

Uncwilly 2009-05-27 04:28

1 Attachment(s)
I think I found out who is the perp....
[attach]3710[/attach]
Monst.

monst 2009-05-27 16:17

That was me... I'm factoring them up to 2^63 and then I'll release them. So it will be productive work. Is that OK with everyone on this thread?

-- Monst

monst 2009-05-27 16:36

They are on my fastest factoring machine - a two-headed Opteron. I've moved the 333.5 to 333.9M range up to the top of the worktodo file. They are being released as they are factored to 2^63.

Uncwilly 2009-05-27 20:57

There is no problems with your help. I had pulled a whole bunch ~4000 to do the same thing earlier. I was checking the status of things (I am tracking the total expo/factored expo fraction) and noticed a huge jump in the number of assigned expos to factoring. It shocked me. To know that it is real work being done is great.

Carry on and thanks.
:bow wave:

Sunfish 2009-05-31 10:19

Just wanted to let you know that I picked up the 2430 expos in the 334.0M - 334.1M range (i.e. 334000013 - 334099999).

Initially I will factor them up to 2^64 which will take a couple of days only using a single core on my Xeon server, then moving up to 2^66. By then some of the other cores should have completed their current LL-work so that I can start to throw in more crunching power to get them up to 2^68.

Already 108 expos factored up to 2^64 within the first 12 hours -> 6 factors found so far :smile:

Sunfish 2009-06-03 20:43

Seems as if the first post trap has hit one more time because my very first post suddenly popped up about 3 days later. :smile:

Just picked up another 2360 expos in the 333.2 - 333.3M range (333200041 - 333299951) to advance them from 63 to 64 bits. This will be done on both cores of my Athlon X2 which just completed some lower range P-1 work. Should be completed in less than a week.

Uncwilly 2009-06-03 22:24

[QUOTE=Sunfish;175779](333200041 - 333299951) to advance them from 63 to 64 bits.[/QUOTE]Cool! that should knock out a bunch of expos that have factors. I took a bunch that are greater than 332399999 up 2 bits from 61 or 62.

While I won't track the range that you are working on in the same level of detail that I am in the lowest range (PrimeNet is handling all the grunt work), but I am tracking the total/factored ratio. I am not recording it, but just watching it.

monst 2009-06-05 12:46

All the exponents from 332.2M to 333.9M have been factored to 2^63. I'm still working on 332.0M to 332.2M, but they are outside the range of this thread.

Sunfish 2009-06-06 20:34

All 2360 expos in the 333.2 - 333.3M range (333200041 - 333299951) have been factored to 2^64.

---> 32 factors found

Now I picked up 1808 expos in the 332.3 - 332.4M range (332307641 - 332399999) that are currently at the lowest level (at least out of those that are available).

I will take them from 66,67 bit level up to 2^68 which should take about a month.

Uncwilly 2009-06-07 00:10

Looking good.

I temporarily have one of my machines doing TF on the M47 candidate. I'll be back running in a week or less. My other slaves are working on expos in the 70-73 bit range.

Uncwilly 2009-06-09 05:30

We have just passed 6000 exponents removed via factoring, 6001 is the current number.
Here are some stat on the factors:[CODE]
smallest: 664385759 (29.3 bits)
largest: 28032821251527749272817 (74.6 bits)
total factors: 7444
average bit depth: 43.7
est. total effort applied to these factors: 68100
(using same baseless value as the other stats)[/CODE]

Sunfish 2009-06-14 11:50

How to get P-1 work in this range
 
I'm thinking to do some P-1 work in this range which raises a couple of questions. I read through quite a bunch of threads and posts but some points still remain unclear so that hopefully someone can shade some light on them.

I understand that in this range TF should be done up to 2^76, then P-1, then the final TF up to 2^77, before the exponent is assigned to LL-work. Looking at some of the expos currently assigned to LL-testing it seems however, that most of them have never been P-1 tested and TF stopped at 2^76 or lower. This means that LL-testers usually grab the expos at lower levels so that there are no expos available for P-1 in general.

So it's necessary to bring an exponent up to the 2^76 level before one can do some P-1 testing. The sequence in the worktodo file would look something like this for a given exponent n currently at 71 bit level:

factor=n,71,76,
PFactor=1,2,n,-1,76,2 (to let the program choose the optimal boundaries)
factor=n,76,77,

However, you can only register for one step at a time. Hence I assume that the exponent will be given to some LL-tester as soon as it is released at 2^76 level before I have a chance to re-register it for P-1 work.

So is there a way to do the whole procedure (TF to 76 bit, then P-1, then TF to 77 bit) with a single line in the worktodo file, either explicitly or implicitly?

I also understand that stage 2 of P-1 needs LOTS of memory for expos in this range but I can easily allow around 5.000MB for this stage which should be more than sufficient based on the discussions in some other threads.

If there is no convinient way to get P-1 work in this range, I might have to look at the lower ranges but any idea is welcome.

Mini-Geek 2009-06-14 12:27

[quote=Sunfish;177524]So is there a way to do the whole procedure (TF to 76 bit, then P-1, then TF to 77 bit) with a single line in the worktodo file, either explicitly or implicitly?[/quote]
Someone else may be able to give you a better answer, but the best option I know of would be Test=n,currentbitlevel and then manually cancelling the LL once it starts.

James Heinrich 2009-06-14 12:41

[QUOTE=Mini-Geek;177526]...then manually cancelling the LL once it starts.[/QUOTE]If you add [b]SequentialWorkToDo=0[/b] to prime.txt you're unlikely to waste much (if any) time doing unwanted LL tests:[quote=undoc.txt]The program used to do factoring and P-1 testing on exponents even if they were not the first entry in worktodo.txt. The rationale was that if the number had a factor, the server could be contacted and another exponent reserved. This avoids the possible scenario where the LL test for the first worktodo.txt line completes, a factor is quickly found for the second line in worktodo.txt and the computer now sits idle until the server can be contacted. This behavior was confusing, especially to newcomers. To restore this old behavior add this line to prime.txt:
SequentialWorkToDo=0
One added benefit is time estimates in Test/Status will be more accurate because we'll know no factors will be found and the LL test must be run.[/quote]

Uncwilly 2009-06-14 12:49

[QUOTE=James Heinrich;177527]If you add [b]SequentialWorkToDo=0[/b] to prime.txt you're unlikely to waste much (if any) time doing unwanted LL tests:[/QUOTE]Thanks James!
I think that currently, the new bit level is 75, then P-1, then 76-77. Here are the current expo count vs bit level in this range[code]51 74
172 75
14 76
6 77
13 P-1[/code]


All times are UTC. The time now is 00:25.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2022, Jelsoft Enterprises Ltd.