Posts by ChertseyAl
log in
1) Message boards : Number crunching : POGSTEST (Message 1876)
Posted 2 Dec 2013 by Profile ChertseyAl

Pogstest is a server that we use for internal testing only it does no real data work.


Why not just run it on your own LAN then? Seems daft to open it up to the great unwashed of the BOINC community if you don't want their help.

Cheers,

Al.
2) Message boards : Number crunching : POGSTEST (Message 1864)
Posted 29 Nov 2013 by Profile ChertseyAl
This is also messing up the stats sites :(

Cheers,

Al.
3) Message boards : Number crunching : POGSTEST (Message 1861)
Posted 29 Nov 2013 by Profile ChertseyAl
I don't know if you guys know, but the POGSTEST project at http://54.208.29.24/pogstest/ is now a copy of this project?

It's going to make life difficult returning completed WUs that came from the real test project :)

Cheers,

Al.
4) Message boards : Number crunching : connection impossible (Message 777)
Posted 28 Oct 2012 by Profile ChertseyAl
Is it this problem? :

http://ec2-23-23-126-96.compute-1.amazonaws.com/pogs/forum_thread.php?id=57#326

Cheers,

Al.
5) Message boards : Number crunching : TSN Java vs TSN BOINC (Message 669)
Posted 4 Oct 2012 by Profile ChertseyAl

I expected that they will share the CPU resources evenly. Looking at the Task Manager and the Resource Monitor reveals though that the Java clients get the lion share in a better than 3:1 ratio


It's been ages since I ran the Java version, but ISTR that it did play fair alongside BOINC (obviously not running this project, as it didn't exist at the time). Perhaps Java is actually running at a higher priority than it would appear.

no_priority_change option in the cc_config.xml I expected to see it running the POGS BOINC clients at Below Normal as well


Hmmm. As I understand it that just makes the application run at the same priority as the BOINC client, which I thought was NORMAL, and would seriously slow a host down.

I do remember that there was some tweak you could do when running BOINC and Folding@Home to make one take precedence. It think it just tweaked Folding to be just above IDLE so that LOW applications would always run first.

Can't remember to be honest :)

Cheers,

Al.
6) Message boards : Number crunching : Always get too much work (Message 598)
Posted 19 Sep 2012 by Profile ChertseyAl
I think the problem is that the old client reports a near zero runtime for each WU. As a result the server sends me the maximum number of WUs per core thinking that the whole lot will complete in less than a second (I wish!).

I can probably work around it by only running this project so that there will be time to complete them all. I'll clear my current backlog, grab a few milestones elsewhere and then come back for my 100k here :)

Cheers,

Al.
7) Message boards : Number crunching : Always get too much work (Message 588)
Posted 18 Sep 2012 by Profile ChertseyAl
Probably due to me running 5.10.45 clients, but is anyone else seeing this problem?

I have a 1 day cache, 0.1 days extra buffer, but I always get about a weeks worth of work that I can't complete on time, so end up aborting about 10% of the WUs I receive.

I guess it doesn't matter as the WUs get reissued, but it means that this project gets stuck in panic mode very quickly and locks other projects out until they hit panic mode :(

Cheers,

Al.
8) Message boards : Number crunching : Work fetch (Message 516)
Posted 3 Sep 2012 by Profile ChertseyAl
Don't know if this is new, or if things have always been like this, but I'm not always getting work when requesting it:

03/09/2012 19:28:40|pogs|Sending scheduler request: To fetch work. Requesting 672 seconds of work, reporting 0 completed tasks
03/09/2012 19:28:45|pogs|Scheduler request succeeded: got 0 new tasks
03/09/2012 19:28:45|pogs|Message from server: No tasks sent
03/09/2012 19:29:45|pogs|Sending scheduler request: To fetch work. Requesting 638 seconds of work, reporting 0 completed tasks
03/09/2012 19:29:50|pogs|Scheduler request succeeded: got 0 new tasks
03/09/2012 19:29:50|pogs|Message from server: No tasks sent
03/09/2012 19:30:51|pogs|Sending scheduler request: To fetch work. Requesting 595 seconds of work, reporting 0 completed tasks
03/09/2012 19:30:56|pogs|Scheduler request succeeded: got 0 new tasks
03/09/2012 19:30:56|pogs|Message from server: No tasks sent
03/09/2012 19:31:57|pogs|Sending scheduler request: To fetch work. Requesting 568 seconds of work, reporting 0 completed tasks
03/09/2012 19:32:02|pogs|Scheduler request succeeded: got 1 new tasks
03/09/2012 19:32:04|pogs|Started download of NGC0428_area35675
03/09/2012 19:32:04|pogs|Started download of NGC0428_area35675.job.xml
03/09/2012 19:32:06|pogs|Finished download of NGC0428_area35675
03/09/2012 19:32:06|pogs|Finished download of NGC0428_area35675.job.xml

Possibly related to the change to HR? Or maybe it's always been like this and I'd never noticed :) Did get work after 3 tries though, so not a disaster.

Cheers,

Al.
9) Questions and Answers : Web site : Twitter (Message 514)
Posted 3 Sep 2012 by Profile ChertseyAl
Probably a dirty word with a lot of people, but I'm finding I'm using Twitter in preference to RSS for news updates from many sites.

Might be useful/interesting to set up a Twitter account to post news items (or just links to news items). I follow @BOINCprojects which syndicates a lot of news, but not all of it.

I guess Facebook might be useful too, but I don't have much experience with it these days.

Just a thought.

Cheers,

Al.
10) Message boards : Number crunching : WU runs, but elapsed time is very wrong (Message 513)
Posted 3 Sep 2012 by Profile ChertseyAl
@kevin

Can't get any work anyway:

Mon 03 Sep 2012 19:05:23 BST|pogs|Sending scheduler request: To fetch work. Requesting 37060 seconds of work, reporting 0 completed tasks
Mon 03 Sep 2012 19:05:28 BST|pogs|Scheduler request succeeded: got 0 new tasks
Mon 03 Sep 2012 19:05:28 BST|pogs|Message from server: No tasks sent

Maybe switching on HR has changed things. Not actually sure that I'm getting work on Windows machines, best go check!

Cheers,

Al.
11) Message boards : Number crunching : RAM usage - Not a complaint, just a question! (Message 512)
Posted 3 Sep 2012 by Profile ChertseyAl

I dropped in by a factor of 10. Hasn't it propagated properly?


Doesn't look like it:

03/09/2012 19:04:06 pogs Requesting new tasks
03/09/2012 19:04:08 pogs Scheduler request completed: got 0 new tasks
03/09/2012 19:04:08 pogs Message from server: No tasks sent
03/09/2012 19:04:08 pogs Message from server: fitsedwrapper needs 476.84 MB RAM but only 470.00 MB is available for use.

Not a big deal.

Cheers,

Al.
12) Message boards : News : David's presentation on POGS (Message 511)
Posted 3 Sep 2012 by Profile ChertseyAl
Thanks for making this available, always nice to see what we're actually doing :)

Cheers,

Al.
13) Message boards : Number crunching : Cross validation (Message 500)
Posted 2 Sep 2012 by Profile ChertseyAl

So for me all problems are with Windows 64 Bit computers.


There was a problem at riojascience@home with cross-validation of W7 64 machines I believe, so not a unique problem. Don't know what the fix was though.

Cheers,

Al.
14) Message boards : Number crunching : WU runs, but elapsed time is very wrong (Message 498)
Posted 1 Sep 2012 by Profile ChertseyAl
Just wondering if the linux-32 application is working before I risk crunching for a fictional 2 billion hours for the WU to die with 'time exceeded' :)

Cheers,

Al.
15) Message boards : Number crunching : RAM usage - Not a complaint, just a question! (Message 497)
Posted 1 Sep 2012 by Profile ChertseyAl
I can drop that right down now I know real numbers on many machines


Still unchanged ;)

Not a big deal, but it gives me a chance to add another old wreck to the project :)

Cheers,

Al.
16) Message boards : News : Display all Galaxies (Message 496)
Posted 1 Sep 2012 by Profile ChertseyAl
@Rob

Just wondering if there's a way of displaying the galaxies that are most relevant to the user. Perhaps have an option to display a list/thumbs in order of the most 'real estate' crunched by the user?

Actually, maybe that should be an option in the 'view your own galaxies' page.

Cheers,

Al.
17) Message boards : Number crunching : Had to detach (Message 495)
Posted 1 Sep 2012 by Profile ChertseyAl
@ZoSo

As others have already said, stick with it.

The percentage is accurate - Updates for every pixel, and once you know how long a pixel takes (you can calculate it easily from the result output from a WU) you can easily work out what's happening to a WU.

I run old BOINC clients, and my runtimes and CPU times are nonsense. But the work is valid and credited. Be nice when the wrapper-less app is available though.

Actually, this project runs very nicely :)

Cheers,

Al.
18) Message boards : News : Load Test Status (Message 466)
Posted 30 Aug 2012 by Profile ChertseyAl
The server status page has:

Users #
with recent credit 352
with credit 361
registered in past 24 hours 15

Computers #
with recent credit 736
with credit 750
registered in past 24 hours 35
current GigaFLOPs 1,034

users in last 24h 137


I suspect the 2121 is a typo for what was 121 at the time.

Not that it matters anyway :)

Cheers,

Al.
19) Questions and Answers : Web site : Had to reset password (Message 442)
Posted 27 Aug 2012 by Profile ChertseyAl

It happened to me too, tried adding another PC yesterday, password /login error, was still logged in at home so managed to change password there.


Same problem here. Tried to attach another machine, login error via BOINC client. I was (and still am) logged in to the website though.

Clicked the 'forgot password' link in the client, got the email, followed the link and changed password to the existing password, and have now been able to attach a new machine.

FWIW, I'm not a member of any team, I was an early joiner (before the team import stuff), and don't recall seeing this problem with any other project :(

Cheers,

Al.
20) Message boards : Number crunching : Ideas to improve this project (Message 380)
Posted 21 Aug 2012 by Profile ChertseyAl

Para 2. Ownership of a Galaxy. Is it at all possible for a single member (plus wingman/men) to completely crunch a single galaxy?


I really like this idea. But if it's going to take a month for one user + wingnut to finish, that galaxy is stuck, whereas distributing it gets a galaxy done quicker. Also, what happens when the wingnut times out? Another month (or whatever) delay. What if the two replications fail to validate? But then again, with trickles I guess work is still getting done.

Perhaps it might be possible to 'reserve' a galaxy for a length of time, so that you can crunch as much as you can before your reservation runs out. That would focus people to get stuff done! Also, with trickles, other users would be validating bits of the galaxy so useful data would keep flowing in. Obviously one one user can reserve a galaxy or we're back to the original idea.

Thinking of projects like Einstein and Primegrid - You get your name in print. If that matters to you :) (Mine already is, in astronomical journals bizarrely enough!).

Cheers,

Al.


Next 20

Main page · Your account · Message boards


Copyright © 2017 The International Centre for Radio Astronomy Research