If you use(d) the DB too much for ECMing, it found lots of largish factors that you would have liked to find yourself. I admit running ECM can be a bit boring, but it isn't when

[code][Jul 30 2009, 13:26:37] c89: running 214 ecm curves at B1=5e4...
Using B1=[COLOR="Red"]50000[/COLOR], B2=12746592, polynomial x^2, sigma=1385519609
Step 1 took 719ms
Step 2 took 578ms
********** Factor found in step 2: 348637699946385147700920287923
[Jul 30 2009, 13:27:36] *** prp30 = 348637699946385147700920287923[/code]

Admittedly, that is not my ECM personal best, and it isn't my B1=5e4 personal best (p32). But I deleted the logs that contain them, and this is the biggest ECM factor I've got in my current logs. However, here are a couple that might be more interesting:

[code][Jul 29 2009, 18:18:51] c70: running 74 ecm curves at B1=11e3...
Using B1=[COLOR="Red"]11000[/COLOR], B2=1873422, polynomial x^1, sigma=3482709144
Step 1 took 156ms
Step 2 took 156ms
********** Factor found in step 2: 129724528443865898423660249
[Jul 29 2009, 18:18:58] *** prp27 = 129724528443865898423660249[/code]

[code][Aug 08 2009, 19:29:46] c70: running [COLOR="Red"]P+1[/COLOR] x3 at B1=11e4...
Using B1=[COLOR="Red"]110000[/COLOR], B2=341719732, polynomial x^1, x0=1184906562
Step 1 took 266ms
Step 2 took 890ms
********** Factor found in step 2: 175577667793217083523297099
[Aug 08 2009, 19:29:49] *** prp27 = 175577667793217083523297099[/code]

The last one was found with one of the very first P+1 runs aliqueit did, and both of these saved what would probably have been a QS run. Anyway, P+1 = 2^2 * 3 * 5^2 * 12619 * 54617 * 66191 * 79309 * 161761 - so B2 could have been much lower.

[quote=gd_barnes;186656]Is anyone experiencing the same problems with the new version of the DB that I am?

My ID (gd_barnes) is showing that I have 1.00 points available and I'm only trying to submit small-pointed jobs but am not getting the highly satisfactory response that I used to get.


Gary[/quote]

Yes - with the new version of the DB I can not submit aliquot sequences with the Report Button. It seems the .elf file was send but the sequence is not up to date after hours.
I think you can only get points if you run a worker yourself-the text on the worker site say this:

Run your own worker and collect points? Get the [URL="http://factorization.ath.cx/worker.pl"]perl-script![/URL] Note: The worker script is buggy! Try it but dont expect it to work.

Regards Andi_HB

early on(beta actually) i got points by submitting ecm curves done

run worker to get points? NO, it's BUGGY!
 

[QUOTE=Andi_HB;186700]Yes - with the new version of the DB I can not submit aliquot sequences with the Report Button. It seems the .elf file was send but the sequence is not up to date after hours.
I think you can only get points if you run a worker yourself-the text on the worker site say this:

Run your own worker and collect points? Get the [URL="http://factorization.ath.cx/worker.pl"]perl-script![/URL] Note: The worker script is buggy! Try it but dont expect it to work.

Regards Andi_HB[/QUOTE]

I would go as far and say: DON'T use the script under windows - it seems to have some sort of "tasks finish immedediately but haven't really be done" bug. :nuke: (thousands of curves on a c15x(?) at B1=11M within minutes on a single CPU at 2 GHz? NEVER.)

[quote=Andi47;186755]I would go as far and say: DON'T use the script under windows - it seems to have some sort of "tasks finish immedediately but haven't really be done" bug. :nuke: (thousands of curves on a c15x(?) at B1=11M within minutes on a single CPU at 2 GHz? NEVER.)[/quote]
Or to be more accurate: never run the script under Windows without going through cygwin's perl.exe (until the script is fixed to work in Windows properly).
The worst part is that anybody might not notice the problem because they don't know they're being assigned curves that should take a long time - only that they're being assigned *some* work and that the work is going extremely fast. The only real clue is that most things require very few points, but the points stack up very quickly.
Anyone know how to write the script for Windows so that it actually works right? (wouldn't have to work on Linux as well, there could be separate versions)

I think you guys missed my point.

All I want to do is "ECM low limits" and "ECM medium limits" and have them come back in 5 to 15 seconds with accurate factorization. Now it takes literally hours and it's still hard to tell if it has completed the test or not. Sometimes it doesn't find factors that it should.

The low limits and medium limits require < .01 of a point but I have a full 1 point available. Also, I seem to ALWAYS have 1 point available. Don't I ever "spend" points? This doesn't make sense. I understand that I would need to contribute workers if I wanted something that "cost" more than 1 point such as ECM curves where B1=3M. That's no problem. But the rest of it doesn't make sense to me.

It also doesn't automatically factor #'s to 2000 like it did before, which now requires extra keystrokes.

What's the deal? I used to get great turnaround. Now, virtually nothing. Are there virtually no available workers now?

It used to be so clear before. Now it's very unclear on what it is doing. Is there a set of instructions somewhere?

What it's come down to is practically this: It's only a DB with little actual computing capability. If that is the direction it is heading, then that is fine. I'd just like to know. I'm only looking to do moderate amounts of work on it.

The DB is still a tremendous tool for storing vast amounts of information but its capacity for computing such information seems to have gone south with the latest release. Does anyone agree with this?


Gary

@Gary
A lot of work was assigned a few days ago(70k small factorizations) and at the same time about 10 t35s were started by anonymous. The workers have been struggling to catch up. It is now down to 11336. Once the queue is empty of loads of small assignments the database should speed up. It was extremely quick before they were assigned.

[QUOTE=henryzz;186825]@Gary
A lot of work was assigned a few days ago(70k small factorizations) and at the same time about 10 t35s were started by anonymous. The workers have been struggling to catch up. It is now down to 11336. Once the queue is empty of loads of small assignments the database should speed up. It was extremely quick before they were assigned.[/QUOTE]

And I think I know why those small factorizations suddenly entered the queue. The database finished checking all the aliquot sequences (which had taken a few days) and moved on to checking home prime sequences in bases from 2 to 11, and most of those hadn't been taken very far.

Feature request: When reporting ECM-work, people often seem to left the line for B2 blank. Please auto-add the the default B2 when the line has been left blank.

[URL="http://factordb.com/search.php?se=1&aq=517392"]517392[/URL] is broken (partial squared line bug on line 754, it repeated c102=p6*p96).

Markus,

Would it be possible to hard-code just [URL="http://www.mersenneforum.org/showthread.php?t=10439"]these Aurifeuillian forms[/URL] for starters? (triggerred when a number is entered in the form [B]a^n+-1[/B] and a and n match these forms)
This would be in addition to a set of rules that you may already have; if you don't, that's an omission. (E.g. [B]a[sup]2n[/sup]-1[/B] = a[sup]n[/sup]-1 . a[sup]n[/sup]+1, trivially; then [B]a[sup]2n+1[/sup]+-1[/B] = ...) [COLOR=green]EDIT: Yes, you do (tested: 17^105+1 , 23^108-1 ...)[/COLOR]

At the same time, the "Additional information" for the number corresponding to the 2nd parenthesized cofactor would benefit by being set to ...[B]L[/B] (e.g. 7,707[B]L[/B]), and the 3rd cofactor to ...[B]M[/B]

-S