[QUOTE=10metreh;167092]When you can fit a t35 into 1/4 of the sieving time (or less), that doesn't really apply![/QUOTE]

For this, you have to count CPU-time, not wallclock-time!

No offence intended!
 

[quote=Andi47;167094]For this, you have to count CPU-time, not wallclock-time![/quote]

Andi, Markus and Sander, wherever you are, I'm sorry if I've offended you. :sad:

[QUOTE=10metreh;167082]That's because someone has tried to extend mklasson's tables for k*2^n-1 to k=15. They have put a lot of composites from that into the very high limits queue. (It could be mklasson himself.)[/quote]

It most certainly is not. I've never once used Syd's computers to do my factoring. It's a superb thing he's put together and it seems a great shame to me to abuse it.

And btw, whoever it is, you're regrettably probably wasting effort. I've been doing quite a lot of work on k=15 already "in silence". Everything below n=556 is completely factored for example. I just haven't put it up on the page because I'd like to do something first that'll better show the ecm efforts that's been done. That, plus I feel a little silly expending all that effort on an almost completely useless project. :lol:

And, aah, before I manage to step on too many egos... :lol: There's of course a thin line between using and abusing Syd's workers, but I think it's fairly clear which side my queuing 3000 numbers would fall on.

Furthermore, if the person who was doing 15*2^n+-1 on there feel like continuing to crack those numbers I'll hurry up and make my 15-tables "official" so you can enjoy unspoiled composites instead. Well, I'll do it anyway to prevent further unintended redundancy. There, it's all up at [url]http://mklasson.com/factors/[/url].

[quote=mklasson;167116]And, aah, before I manage to step on too many egos... :lol: There's of course a thin line between using and abusing Syd's workers, but I think it's fairly clear which side my queuing 3000 numbers would fall on.

Furthermore, if the person who was doing 15*2^n+-1 on there feel like continuing to crack those numbers I'll hurry up and make my 15-tables "official" so you can enjoy unspoiled composites instead. Well, I'll do it anyway to prevent further unintended redundancy. There, it's all up at [URL]http://mklasson.com/factors/[/URL].[/quote]

Just spotted an unfactored C102. Looks to be in your "easy" range. I won't reserve it, though.

[QUOTE=10metreh;167117]Just spotted an unfactored C102. Looks to be in your "easy" range.[/quote]

It hasn't had much ecm done on it either so there could be a p30-35 or something lurking inside it.

[QUOTE=10metreh;167117]I won't reserve it, though.[/QUOTE]

I made the reservation system because people got a little miffed when numbers they had spent weeks running gnfs on was factored by someone else before they had time to submit the factor. It was intended to ease tensions and reduce waste, but I personally don't really mind if you don't use it now. It isn't even an issue when there's hardly anyone factoring the numbers in the first place nowadays. :smile:

[quote=Andi47;166919]Workers 13 through 16 seem to be offline.[/quote]

I shutdown the workers from time to time if the CPU's are needed for something else.

[quote]Markus, could you add a time zone setting? The "Reported" times are always in GMT+1. (Of course, tomorrow will be really confusing, because that's when BST starts.)[/quote]Once I've figured out how the timezone functions work in php i'll add it-

[quote] I have the impression that for very high ECM, newly queued composites seem to get the first place in the queue.

I noticed this as there is a c296 in the queue since Friday - and it's at the rear end of the queue.

What about doing very high ECM in the order of when they come in instead of in order of their size (or whatever)?

BTW: I just clicked the "run now" button for the oldest composite (which did not get any curve since March 27th, ~3 p.m.), it got ~50 curves at B1=1M before it was crowded out again.[/quote]The intention of the workers was to quickly do some ecm, sieving and so on without having the users start lots of instances on their own, not to do huge jobs in background. The queue is just there to keep a job if another job comes in, and of cause to keep all computers busy during night.
New jobs are placed in front of the queue, and if you want another job to move up in queue click the "run now" button, it moves the job up.

[quote]Edit: And someone (probably that person) has clicked "stop" on the C95 I was working on.[/quote]I hope there is no need to add "codes" to stop jobs...

[quote]Edit: Sometimes I see numbers which have more than one p-1 run with the same bounds. The P-1 method finds a factor p [I]in the first run[/I] [B]iff[/B] the penultimate factor of p-1 is smaller than B1 and the largest factor of p-1 is smaller than B2. So a second run at the same (or even lower) bounds is useless. [/quote]Glad you mentioned it - I'll change it.

edit:
just an idea - the "worker" is a little C-program, with a few changes it may connect via TCP. Is anyone willing to run it? Of cause I'll give out the source so you can see what its doing.

Is it possible to implement different queues for each ecm level and process the queue with the smallest B1 first? New numbers can than be put last in the queue, but still have their 250K curves done fast before they wait in line for their 1M and 3M curves?

[quote=smh;167135]Is it possible to implement different queues for each ecm level and process the queue with the smallest B1 first? New numbers can than be put last in the queue, but still have their 250K curves done fast before they wait in line for their 1M and 3M curves?[/quote]
i was thinking exactly the same thing this morning

What happened to the workers? Currently "only" workers 1 through 4 and 10 are visible, and ~half an hour before, a "worker 100" was the only visible worker on the worker status page?

[QUOTE] Is it possible to implement different queues for each ecm level and process the queue with the smallest B1 first? New numbers can than be put last in the queue, but still have their 250K curves done fast before they wait in line for their 1M and 3M curves?
[/QUOTE]

That would require a lot of changes, so maybe later.

[QUOTE]What happened to the workers? Currently "only" workers 1 through 4 and 10 are visible, and ~half an hour before, a "worker 100" was the only visible worker on the worker status page?[/QUOTE]

I currently modify them to connect via TCP, all workers >100 connected from remote, the others are local workers.