Call me a cynic, but let me know if it goes up:smile:

Monitoring the countdown to DCing M42 is an easy way
to see DC progress.

If they gave a countdown to testing below 60M (say)
once, we could monitor first time tests just as easily.

Note that "countdown to proving..." counts DC only.

David

Ahh....I was wondering this. So with the DC countdown let's say the statement is this...


Countdown to proving M(_________) is the XX Mersenne prime = 1,000 but let's say that exponent yyyyyyyyyyy has yet to even receive a first-time LL. Is it only counted once and not twice in this scenario?

[QUOTE=Primeinator;223989]Countdown to proving M(_________) is the XX Mersenne prime = 1,000 but let's say that exponent yyyyyyyyyyy has yet to even receive a first-time LL. Is it only counted once and not twice in this scenario?[/QUOTE]

Yes. It is a countdown of exponents - not a countdown of LL tests needed.

[QUOTE=davieddy;223971]If they gave a countdown to testing below 60M (say) once, we could monitor first time tests just as easily.
David[/QUOTE]

How about 50M? or 100M

I have a report in this thread that with a bit of ciphering can give you those counts....and shows progress since 7 days, 30 days, 2010-01-01, 2009-01-01.

[url]http://www.mersenneforum.org/showpost.php?p=223961&postcount=94[/url]

[QUOTE=petrw1;222263]George: I probably missed the updates when I was off but what ever happened to the thread discussion this Spring about getting a new PrimeNet Server? Is it already in place?[/QUOTE]

I'm happy to be corrected if I'm wrong, but I don't believe this question has been answered.

[quote=petrw1;224029]How about 50M? or 100M

I have a report in this thread that with a bit of ciphering can give you those counts....and shows progress since 7 days, 30 days, 2010-01-01, 2009-01-01.

[URL]http://www.mersenneforum.org/showpost.php?p=223961&postcount=94[/URL][/quote]
Yes, thanks: I look at it every month.

But a 60M countdown would have the following advantages:

Hourly.
Exact number.
Includes 50M+ LLs.
Doesn't include factoring between 60M and 100M.

BTW countdown to proving M42 decreased by 454 in 100 hours
recently. This is equivalent to ~40,000 DCs per year, nearly
enough for the DC wavefront to advance by 2M per year.
(Mind you, FFT boundary approaching (24.9M IIRC)).

OTOH, first time testers seem to be getting indigestion*
between 40M and 50M. Rate of first time LLs is below 80,000/year
and will continue to fall unless the "doling out" wavefront starts
advancing instead of going backwards sometime soon!

David

*Whereas below 40M it is more a case of constipation:-)

[QUOTE=Prime95;223995]Yes. It is a countdown of exponents - not a countdown of LL tests needed.[/QUOTE]

Thank you.

[QUOTE=chalsall;224069]I'm happy to be corrected if I'm wrong, but I don't believe this question has been answered.[/QUOTE]

Scott and I were arguing over minor aspects of the system specs and it never got ordered. The money is still there, but it awaits a renewed effort on our part (more Scott than me since once it arrives it will take a big chunk of Scott's time to get it up and running).

[QUOTE=Prime95;224098]Scott and I were arguing over minor aspects of the system specs and it never got ordered. The money is still there, but it awaits a renewed effort on our part (more Scott than me since once it arrives it will take a big chunk of Scott's time to get it up and running).[/QUOTE]

Let us know if you need to beta-test it when ready :smile:

Luigi

[quote=henryzz;222521]I have attached the current Active assignments webpage so that we can compare in a few days.[/quote]
In the 13 days since I posted this the count upto 22M has decreased from 69-66.
We are slowly getting there.

[quote=henryzz;224141]In the 13 days since I posted this the count upto 22M has decreased from 69-66.
We are slowly getting there.[/quote]

But I'm wondering why so slowly. My last eleven assignments have been in the 22m to 24m range. If my machine is to do work in this range why are the smallest numbers that are still outstanding not assigned rather than ones that are a few places higher in the sequence?

G

[QUOTE=GARYP166;226518]But I'm wondering why so slowly. My last eleven assignments have been in the 22m to 24m range. If my machine is to do work in this range why are the smallest numbers that are still outstanding not assigned rather than ones that are a few places higher in the sequence?
[/QUOTE]
Because the report shows exponents that are assigned to someone. One of them can become available to be assigned to someone else (eg, you) if the present assignee releases it, or if it expires because the computer is 60 days past due in updating the server.

The tail is mostly made up of machines making slower progress, that's just how it works.

After another 17 days we are down to 56.

Very close to two more milestones:

-Countdown to testing all exponents below M(32582657) once: 3

-Countdown to testing all exponents below M(37156667) once: 13

Both prime! :smile:

[quote=Primeinator;226781]Very close to two more milestones:

-Countdown to testing all exponents below M(32582657) once: 3

-Countdown to testing all exponents below M(37156667) once: 13

Both prime! :smile:[/quote]
I was just looking at the "Active Assignments" page to see the status of the remaining tests below them, and noticed that there are a handful of tests in the vicinity of 30M-32582657 that are all assigned as LLs (not DCs)--despite the fact that there should only be 3 untested assignments in this range.

An example of this is 30133123: it is assigned as an LL to an ANONYMOUS v4 client, yet the [url=http://mersenne.org/report_exponent/?exp_lo=30133123&exp_hi=&B1=Get+status]exponent status[/url] query reveals it to have an unverified LL sitting in the database. At first I thought that it was a manually assigned DC that was registered as an LL because the user typed that into his worktodo.ini file, but it's unlikely that someone running as ANONYMOUS with a v4 client is going to care to select a particular DC exponent by hand. My second theory was that it was poached, but I see that the first pass residual was submitted by "C. Cooper / S. Boone"--surely those guys wouldn't have poached it.

Anyone know how something like this happens?

[QUOTE=chalsall;221625]This statement confuses me a bit.

I have two "borged" machines which are still running V4 (doing DC work), and yet I can unassign their assignments through the V5 web interface just like I can for any other machine running V5.[/QUOTE]

So... I tried to do so; unassign all but the currently active assignment from my V4 PC....and as of today all these very same assignments are back again; mind you still incorrectly reported as LL even though they are doing TF. And because of that others are getting assigned the TF and when this PC eventually finished the actual TF it is rejected as NOT needed.

I know it is assigned TF (and NOT LL) for three reasons:
1. That is what I set it to when I set it up and still had access
2. The assignents are in the 53-63M range. PrimeNet is NOT assigning anyone LL in that range.
3. Even though it a 2.0 Ghz P4 and runs very sproadically the estimated completing times are less than a week.

[QUOTE=petrw1;226965]So... I tried to do so...[/QUOTE]

I'm sure George and Scott will be appropriately interested.

After all, it has been said that advancement in science is not generally the "eureka moment", but instead is when one is presented with things which makes one think "hmmm... that's strange".

[quote=petrw1;226965]So... I tried to do so; unassign all but the currently active assignment from my V4 PC....and as of today all these very same assignments are back again[/quote]

[URL]http://www.mersenneforum.org/showthread.php?t=13761[/URL]

Maybe a similar issue to mine - I had an exponent against my name that I did not reserve and which appeared when I was in the middle of another test rather than near the end.

G

[quote=henryzz;226538]After another 17 days we are down to 56.[/quote]
There are now 51 exponents remaining less than 22M,
A speadsheet regression line seems to suggest mid 2012 but need more data to be accurate.

[LIST][*][SIZE=2]All exponents below 31,494,937 have been tested at least once.[/SIZE][*][SIZE=2]Countdown to testing all exponents below M(32582657) once: 1[/SIZE][*][SIZE=2]Countdown to testing all exponents below M(37156667) once: 3[/SIZE][/LIST][SIZE=2]:spot:
[/SIZE]

[QUOTE=ckdo;237913][LIST][*][SIZE=2]All exponents below 31,494,937 have been tested at least once.[/SIZE][*][SIZE=2]Countdown to testing all exponents below M(32582657) once: 1[/SIZE][*][SIZE=2]Countdown to testing all exponents below M(37156667) once: 3[/SIZE][/LIST][SIZE=2]:spot:
[/SIZE][/QUOTE]I was checking earlier today and it was 2 and 4

1 and 3 now.

TF has broken beyond the classical GIMPS space
 

Received two TF assignments above 79.3M on Saturday the 20th. These are the first "mainstream" assignments that I have received beyond the classical 79.3M GIMPS search space.

I can't get enthusastic about the effort to factor up to
the penultimate bit for exponents so far ahead of the LL wavefront.
The chance of finding a factor is small, and leaves well over
half the desired work (P-1 and the last bit) undone.

Meanwhile, the assignment of first time LL tests is
faltering. A few months ago, nearly all the available 50M
exponents had been assigned. The number available then
climbed to 7000 before dropping again.
Currently DCs are being assigned faster than first time tests.

David

[QUOTE=ckdo;237913][LIST][*][SIZE=2]All exponents below 31,494,937 have been tested at least once.[/SIZE][*][SIZE=2]Countdown to testing all exponents below M(32582657) once: 1[/SIZE][/LIST][/QUOTE]
That last exponent is out in assignment for over 2 years :exclaim: (LL testing to "alaendle" on 2008-11-15). Time to steal that back and finish off in somewhat 3-5 days?

I wouldn´t do that.
The exponent is being worked on and is nearly finished.
Moreover, no need to start another flame war about poaching...
Your CPU cycles will be much more well spent doing exponents that are not assigned.

[QUOTE=lycorn;238161]I wouldn´t do that.
The exponent is being worked on and is nearly finished.
Moreover, no need to start another flame war about poaching...
Your CPU cycles will be much more well spent doing exponents that are not assigned.[/QUOTE]
:goodposting:

(As usual):smile:

David

Or to stay true to type:[url=http://www.youtube.com/watch?v=3GPEIetf-4c]As usual[/url]

[QUOTE=lycorn;238161]The exponent is being worked on and is nearly finished.[/QUOTE]How much is it "nearly" finished? several months to go? I'm just worried about such donators slowing GIMPS...

[QUOTE=Commaster;238231]How much is it "nearly" finished? several months to go? I'm just worried about such donators slowing GIMPS...[/QUOTE]

Do you have an advantage, if the test is finished earlier? Just wait for the exponent to finish. Where is absolutely no need for poaching. I can´t understand these stressing people...

[QUOTE=Commaster;238231]How much is it "nearly" finished? several months to go? I'm just worried about such donators slowing GIMPS...[/QUOTE]

Post #5: [URL="http://www.mersenneforum.org/showthread.php?t=14052"]http://www.mersenneforum.org/showthread.php?t=14052[/URL]

and [URL="http://www.mersenne.org/assignments/?exp_lo=31494937&exp_hi=31494937&B1=Get+Assignments"]http://www.mersenne.org/assignments/?exp_lo=31494937&exp_hi=31494937&B1=Get+Assignments[/URL]

[QUOTE=Commaster;238231]I'm just worried about such donators slowing GIMPS...[/QUOTE]

I understand your willingness to see the milestones cleared (I feel just the same). But consider the following:

1. The current assignee has already donated a lot of time to process this exponent and [U]he is still working on it.[/U] Therefore, it is unethical to take it away from him.
2. If you are concerned about GIMPS being slowed, testing an exponent that is already 92%+ tested is definitely not the right thing to do. It is a duplication of efforts, to absolutely no benefit. If you (or someone else) do it now, the original assignee will either notice that and possibly drop the exponent or even quit his GIMPS participation altogether in protest, or just finish the exponent and the result will be a useless triple check. That´s why I wrote your CPU cycles would be better used on not-yet-assigned exponents.

Ok, ok, I'm done. But i'll say one last thing: "Nice CPU there, [B]alaendle[/B]. Stands the test of time!" ^_^
[B]ATH[/B], thanks for the assignments page, new to me...

[QUOTE=lycorn;238248]If you (or someone else) do it now, the original assignee will either notice that and possibly drop the exponent or even quit his GIMPS participation altogether in protest, [/QUOTE]

I concur with Lycorn et al.

My first PC in GIMPS was a 400 Mhz PII - and it spent 15 months on 1 LL. If I had been notified that it was NOT REQUIRED ("poached") I may have looked to donate my cycles to another cause ... and certainly would NOT have signed up a dozen family members and friends.

However, 7 years later I have at any one time between 10 and 15 PC's (mostly Duals and Quads) and have completed hundreds of LLs and DCs and made it to top 40 lifetime.

Please don't take this as bragging because I am still by no means a MAJOR player in GIMPS but at the same time the 7500 or so participants below me in the standings have contributed MORE in total than the big players at the top. And that list includes the one working on this assignment. And by the way, his ponts show that he has completed LL assignments before.

small correction ...
 

[QUOTE=petrw1;238252]However, 7 years later I have ... made it to top 40 [U]lifetime[/U].[/QUOTE]

Should say top 40[U] in the last 365 days[/U]

[QUOTE=lycorn;238248]...or just finish the exponent and the result will be a useless triple check.[/QUOTE]

Of course, I'm not saying I condone poaching at all, just pointing out a minor technical inaccuracy here: this number has not been completed before. If someone were to poach it and complete it before the proper owner, the proper owner's test would be the double check (always useful), not the triple check (usually useless). And not all triple checks are completely useless, so even in the case where the proper owner's assignment was a double (or triple, etc.) check, there's a small chance (very roughly 5%, not accounting for the fact that the owner's computer is likely to be slower and so have more time to make an error) that their result will still be useful.

[QUOTE=Mini-Geek;238272]Of course, I'm not saying I condone poaching at all, just pointing out a minor technical inaccuracy here: this number has not been completed before. If someone were to poach it and complete it before the proper owner, the proper owner's test would be the double check (always useful), not the triple check (usually useless). And not all triple checks are completely useless, so even in the case where the proper owner's assignment was a double (or triple, etc.) check, there's a small chance (very roughly 5%, not accounting for the fact that the owner's computer is likely to be slower and so have more time to make an error) that their result will still be useful.[/QUOTE]

Although you make a valid point, it is not a strong one. Allowing the server to allocate tests prevents multiple verifications. As soon as this test is done, the double check will be handed out to a single computer, rather than multiple users jumping in and poaching and thereby wasting resources. Especially now given that there have been recent changes in the allocation of tests to favour double checks to those systems that are better adept at completing them in a timely manner.

For a proper doublecheck we need some sort of different shift done on one of the tests. Two first time tests aren't enough. I don't know what happens on the server if the doublecheck(shifted test) is returned first. If someone wants to poach it without causing any extra work they should run it as a doublecheck not a firsttime ll.

[QUOTE=henryzz;238280]For a proper doublecheck we need some sort of different shift done on one of the tests. Two first time tests aren't enough.[/QUOTE]All test are randomly shifted. For instance the following first time test :
[Mon Nov 03 18:46:04 2008]
UID: S485122/Q67-W, M43954151 is not prime. Res64: 735245335F4DCB15. Wd1: EDD5BF6B,40843060,00000000, AID: 9938BE64F3076E05033F8AC703520BF4
had a shift of 40843060.

Jacob

[QUOTE=S485122;238284]All test are randomly shifted. For instance the following first time test :
[Mon Nov 03 18:46:04 2008]
UID: S485122/Q67-W, M43954151 is not prime. Res64: 735245335F4DCB15. Wd1: EDD5BF6B,40843060,00000000, AID: 9938BE64F3076E05033F8AC703520BF4
had a shift of 40843060.

Jacob[/QUOTE]
I thought that was only done for doubechecks.:blush:

[QUOTE=Commaster;238231]I'm just worried about such donators slowing GIMPS...[/QUOTE]GIMPS runs fastest when everyone sticks to their own assignments.

By poaching someone else's assignment, you would be SLOWING progress because the work you could otherwise be doing goes undone.

As long as the assignee is making regular progress -- even at a pace that seems "slow" to you -- then that assignee is making progress on a part of GIMPS work [I]that no one else has to do instead[/I], so that everyone else can make progress on [I]their own assignments[/I].

Unassigned duplication of effort does not speed up GIMPS. Please don't do that.

Suppose there were only two contributors to GIMPS, one with a system that completes 100 assignments per month and another with a system that completes only 1 assignment per month. (Also, assume that every assignment involves identical work.) As long as each sticks to his own assignments, GIMPS progresses at a rate of 101 assignments per month.

If the user with the "fast" system poaches the assignment given to the "slow" system, then the "fast" system will still complete only 100 assignments per month, one of which is an unnecessary duplication of the assignment given to the "slow" system. One of the assignments to the "fast" system goes undone! Result: GIMPS gets only 100, not 101, assignments completed that month (one of which is needlessly duplicated) -- the poacher has SLOWED, not SPEEDED progress.

[QUOTE=Mini-Geek;238272]Of course, I'm not saying I condone poaching at all,[/QUOTE]... but that's the effect of your words that try to justify poaching, whether you want to admit it or not. (You've been around here long enough -- over 3,000 posts --- that you should know that by now.)

[quote]just pointing out a minor technical inaccuracy here: this number has not been completed before. If someone were to poach it and complete it before the proper owner, the proper owner's test would be the double check (always useful),[/quote]Why is it that you and other poaching-apologizers omit to take into account that poaching could have a discouraging effect on the poachee, which could lead to GIMPS's losing that contributor's future efforts? (Don't say it can't happen. It happened to me, and GIMPS lost some of my efforts for a while as a result.) Are you empathy-impaired?

Don't you consider the feelings of folks who follow the rules to be worth anything? If you do, then please don't step on them by going outside the rules to do someone else's assignment.

Suppose that other fellow was interested in doing only first-time LL tests, to have the slim chance of being a Mersnne prime discoverer. The poacher would ruin that fellow's enjoyment when he found out that his first-time test had turned out to be _only_ a doublecheck. (Don't say it can't happen. It happened to me, multiple times.)

Some folks out there are dreaming that their "slow" system could discover a Mersenne prime, by following all GIMPS rules and making steady progress. Please don't squash such dreams by poaching.

[quote]not the triple check (usually useless). And not all triple checks are completely useless, so even in the case where the proper owner's assignment was a double (or triple, etc.) check, there's a small chance (very roughly 5%, not accounting for the fact that the owner's computer is likely to be slower and so have more time to make an error) that their result will still be useful.[/quote]... which is just more of an extended attempt to justify poaching without mentioning the side effects that hurt GIMPS.

- - -

A double-check will still be useful [I]when it's done as a proper GIMPS assignment[/I], and in that case has the ADVANTAGE of not causing the deleterious effects of poaching.

A triple-check will still be useful to GIMPS when it's done as a proper GIMPS assignment, and in that case has the advantage of really being necessary. Poaching a DC will [I]usually[/I] [I](roughly 95% of the time, [U]according to your own measure[/U])[/I] just SLOW DOWN our progress by needlessly diverting resources to doing something that doesn't turn out to need doing -- in addition to the previously-mentioned deleterious effects.

Please, for the good of GIMPS, stop trying to excuse poaching.

[QUOTE=Mini-Geek;238272](...) this number has not been completed before.[/QUOTE]

You´re right. I was thinking of DCs. But the main point remains.

@cheesehead: cool down, man! Mini-Geek is not advocating nor excusing poaching, just pointing out some (minor) details.
(In line with his nickname... :smile: )