OFFICIAL "SERVER PROBLEMS" THREAD

[GP2]

In the Account Assignment Details page, in the bottom part where the worktodo.txt lines are displayed, the ECM2= lines don't show the quoted list of comma-separated factors anymore.

Instead it just displays "1410" for ECM, and "1000" for Fermat exponents.

[Madpoo]

Originally Posted by GP2

In the Account Assignment Details page, in the bottom part where the worktodo.txt lines are displayed, the ECM2= lines don't show the quoted list of comma-separated factors anymore.

Instead it just displays "1410" for ECM, and "1000" for Fermat exponents.

Weird. I did adjust that page the other day to try and better list the PRP assignments. Not sure I got those quite right though, and you're really better off using the line from when you got the assignment originally.

I didn't do anything with any other assignment types though, so that's strange. I'll have to look back at previous versions of that page to see what else might have changed recently.

Actually, I just looked back at the past couple iterations of that page, going back quite a bit, and I don't see that it would have ever included known factors for ECM2. If it ever did, that would have been nice because I had to work a comma separated list of factors into a SQL query as a text string (a little FOR XML magic there).

[thyw]

On ecm_report page the number of curves are "already done", (it's going up as i observed). It would suggest, that the number of curves (280, 640, 1580...) are the maximum(optimal) for the current ECM "level". But e.g. at 818339, it is still not flipped over to Done, and when requesting an assigment it still assigns one with lower bounds (250k).
So it's either the minimum requiment for stepping on to next level, or it should be count as completed (on that bound).

[Prime95]

Originally Posted by thyw

On ecm_report page the number of curves are "already done", (it's going up as i observed). It would suggest, that the number of curves (280, 640, 1580...) are the maximum(optimal) for the current ECM "level". But e.g. at 818339, it is still not flipped over to Done, and when requesting an assigment it still assigns one with lower bounds (250k).
So it's either the minimum requiment for stepping on to next level, or it should be count as completed (on that bound).

My guess is a floating point roundoff issue. ECM curves done is stored as a float and converted to an integer for display. So, 639.7 might round up to display as 640 yet not satisfy the "if >= 640" to display "Done". Please watch as the next ECM curve completes.

[GP2]

Just today, LL residues were submitted for M11, M23 and various other tiny exponents.

These should have been rejected by the server, because these exponents are already factored, not to mention they're under 1 million.

To make matters worse, the residues submitted aren't even correct. For M11 it should be 00000000000006C8 and for M23 it should be 00000000005D32F7.

Independently of that, some other user also submitted PRP cofactor test results for these exponents, which are at least sort of useful, although kind of obvious.

[GP2]

Originally Posted by GP2

Just today, LL residues were submitted for M11, M23 and various other tiny exponents.

Obviously, those were PRP residues rather than LL residues. Not sure how I failed to see that. So it's the same problem that was reported earlier.

[0PolarBearsHere]

I know it's slightly off topic, but how come the matching factors aren't shown on the results pages (i.e. 89 for M11 and178481for M23)?

[retina]

Originally Posted by 0PolarBearsHere

I know it's slightly off topic, but how come the matching factors aren't shown on the results pages (i.e. 89 for M11 and178481for M23)?

Because the mathematical convention to not show the largest factor outweighs all other practical and sensible alternatives? Because people are scared to go against conventions out of fear? Because we can't expect to be too lazy and should expect to compute the final factor ourselves?

[GP2]

Originally Posted by 0PolarBearsHere

I know it's slightly off topic, but how come the matching factors aren't shown on the results pages (i.e. 89 for M11 and178481for M23)?

Here's the list of all known Mersenne PRPs (i.e., Mersenne numbers with prime exponent and either fully-factored or with a PRP cofactor).

There are 320 of them, and once you get past the small exponents, the cofactors rapidly get too large to display. For instance M7080247 has a 2,131,318-digit PRP cofactor.

So one way or another, there would have to be some cutoff point after which you don't display that cofactor. So the convention is simply to not display any of them, even the trivially small ones, because they are after all easily calculable.

[James Heinrich]

Originally Posted by GP2

Here's the list of all known Mersenne PRPs ...
So the convention is simply to not display any of them, even the trivially small ones, because they are after all easily calculable.

And on the above page, if you click the last-factor placeholder (e.g. "P-105") it will display the decimal representation of the actual factor, should you be so inclined.

[retina]

Originally Posted by James Heinrich

And on the above page, if you click the last-factor placeholder (e.g. "P-105") it will display the decimal representation of the actual factor, should you be so inclined.

Only if JS is enabled.

I also don't understand that if the data is downloaded with the page anyway why not just display it without the extra steps of navigating the mouse cursor and pressing a button?