Reserving F1157.

Please reserve F1629 for us.

:smile:

Reserving a cup of tea...lol

Two questions:

Does doing the post-processing work scale equally as you add cores? (For example, one core does the work in a month, two cores do the work in two weeks and four cores do the work in a week.)

How many boxes are involved in the sieving effort?

[QUOTE=Xyzzy;311070]Two questions:

Does doing the post-processing work scale equally as you add cores? (For example, one core does the work in a month, two cores do the work in two weeks and four cores do the work in a week.)

How many boxes are involved in the sieving effort?[/QUOTE]

1. No. It does scale relatively well but not linearly. The code runs into memory bandwidth issues.
2. Currently there are about 1000 users and 2000 computers in the total sieving effort, with about 850 users reporting in the last 24 hours. A bit over 200 of those users contribute to the lasieved sieving.

Why are there two pages?

[url]http://escatter11.fullerton.edu/nfs/crunching.php[/url]
[url]http://escatter11.fullerton.edu/nfs/crunching_e.php[/url]

Does S3m637 need a volunteer for post-processing when it is done? (If so, we volunteer!)

:smile:

The second page contains larger numbers using the lasievee application. I typically do the postprocessing for those on the cluster here, but S3m637 shouldn't be particularly hard. It'll probably require 10-12 GB of memory and ~2 weeks on your computer. Are you up for that?

[QUOTE]It'll probably require 10-12 GB of memory and ~2 weeks on your computer. Are you up for that?[/QUOTE]Sure!

We just set up the (overly complex) boinc-manager on our primary workstation. It appears to work.

There is an option (Advanced -> Select Computer) that implies (?) that we can connect to other computers on our network and manage them graphically even if they are headless?

Also, we were unable to find a way to limit the work queued by the client. We, at times, may need to run a job that takes a few days, so we would like to avoid having a lot of work left in limbo.

Finally, is it possible to choose which number you want to work on?

:help:

[QUOTE=Xyzzy;311097]Also, we were unable to find a way to limit the work queued by the client. We, at times, may need to run a job that takes a few days, so we would like to avoid having a lot of work left in limbo.[/QUOTE]

Here ([URL]http://escatter11.fullerton.edu/nfs/prefs.php?subset=global[/URL]) there's an option to "Maintain enough tasks to keep busy for at least X days". You can choose X days. You can always abort the wu's when you are tired.

[QUOTE=Xyzzy;311097]Finally, is it possible to choose which number you want to work on?

:help:[/QUOTE]

It is only possible to choose what application to run, here: [URL]http://escatter11.fullerton.edu/nfs/prefs.php?subset=project[/URL]

lasieved - app for RSALS subproject, uses less than 0.5 GB memory: yes
lasievee - work nearly always available, uses up to 0.5 GB memory: no
lasievef - used for huge factorizations, uses up to 1 GB memory: no
lasieve5f - used for huge factorizations, uses up to 1 GB memory: no

[QUOTE=Xyzzy;311097]We just set up the (overly complex) boinc-manager on our primary workstation. It appears to work.

There is an option (Advanced -> Select Computer) that implies (?) that we can connect to other computers on our network and manage them graphically even if they are headless?

Also, we were unable to find a way to limit the work queued by the client. We, at times, may need to run a job that takes a few days, so we would like to avoid having a lot of work left in limbo.

Finally, is it possible to choose which number you want to work on?

:help:[/QUOTE]

Yes, you can manage multiple headless boinc installations. On the headless computer, install boinc, create gui_rpc_auth.cfg and remote_hosts.cfg in the boinc directory, then run boinc. See the page [url]http://boinc.berkeley.edu/wiki/Controlling_BOINC_remotely[/url] for details, and let me know if you have problems getting it to work.

To limit the work queued, go to Tools, Computing preferences. On the network usage tab, reduce Minimum work buffer and Max additional work buffer. I often set them both to 0.1 days.

You can't choose a particular number to work on, but you can choose which apps you want to use. Go to [url]http://escatter11.fullerton.edu/nfs/prefs.php?subset=project[/url] and select Edit NFS@Home preferences. lasieved is used for the smallest numbers previously done by RSALS, lasievee is used for medium-sized numbers, and lasievef and lasieve5f are used for the largest numbers like, currently, M1019.