1) Forums : General Topics : CPU Time Less Than Elapsed Time (Message 8838)
Posted 21 Jan 2010 by sygopet
Post:
In BOINC Manager>Tasks, for (as far as I know) ALL projects, and for any task that has started, the properties button will show "elapsed time" and "CPU time" and (as you would expect) elapsed is always slightly greater than CPU time.

For a task that is running there is also the CPU time at the last checkpoint.
It is this latter time which, for Cosmology, is often an hour or more adrift - leading to potential waste of processing time if something goes wrong - and we really do need more built-in checkpointing to overcome that.

All other BOINC projects, in my experience, checkpoint at much more frequent intervals so the "CPU time at the last checkpoint" is never more than a minute or two behind "CPU time".

Elapsed time is another matter and the difference between this and CPU time depends how busy your computer is with other things; I find typically less than 5% for my projects, for completed tasks. But I can't give you the figure for Cosmology because that information is not given.
2) Forums : Announcements : We're back! (Message 8722)
Posted 19 Nov 2009 by sygopet
Post:
. . . and we are back in operation!


Not quite, I'm afraid!
See posting in "Urgent problems" thread in Technical Support.
Another little job for Scott.
3) Forums : General Topics : No Work Available? Is this project active? (Message 8710)
Posted 17 Nov 2009 by sygopet
Post:
Thank you for the information. Unfortunately, BOINC polls the project for work
every 3 minutes or so. I will just detach and check back in another week.


No need to detach: just go to the "projects" page in the BOINC manager, select Cosmology and click on "no new tasks". You can "allow new tasks" when units become available again.
In the mean time, your other projects will take up the slack.
4) Forums : General Topics : Credits (Message 8499)
Posted 26 Jul 2009 by sygopet
Post:
. . . I find my single 3.0ghz P4 Prescott (hyperthreaded) with 2 gigs of ram takes 40 n to 50 hours per work unit for 200 credits. I used to get 140 credits for 6-8 hourd per work unit on the same machine. . .


I've just checked, since you have a similar setup to myself. Your unit processed actually received (the normal) 420 credits.
That's not a lot per hour for that one (although comparable with some other Boinc projects), but to make decisions on the basis of a single result may be regarded as premature - I suggest you try a few more units since the time taken per unit is somewhat variable and 40 to 50 hours looks to be on the high end of the distribution.
5) Forums : Technical Support : URGENT PROBLEMS THREAD (2009 and after) (Message 8480)
Posted 16 Jul 2009 by sygopet
Post:
I have a silelar (?) problem with wu_070109_063008_3_0_0. It gets to about 67% then resets to 60% again.

Looks like there is some kind of problem here ...

Allan


This problem is well known and references to it have appeared in this thread and elsewhere for several months.
Your 60% will, I believe, have returned to where it was before, after running the unit for a few minutes.
It doesn't seem to cause any lasting problem and, I suspect, the people running the project take the view that there are plenty of more important things to spend their time fixing at present. Like the excessive memory requirements.
6) Forums : General Topics : Download error. (Message 8477)
Posted 13 Jul 2009 by sygopet
Post:
I lately got the impression, that those "download failed" errors are connected to a missed deadline of the first one who received a result of that workunit.


That is certainly my observation also, over the past several months.
The process seems to be that an initial unit is sent (apparently) to one computer, but doesn't arrive - it is a "ghost" unit.
15 days later, the unit is "timed out - no response" and, over the next hour or two, further units are prepared and sent to other computers - it is these which are seen as giving the "error which downloading" or exit status -186 (0xffffffffffffff46) message for the task.
So eventually this process causes a "Too many error results. Too many total results" message to be attached to the WU.
The source of all this mayhem seems to lie in the initial (ghost) unit.
I suspect the problem hasn't been "fixed" because it causes no harm to anyone while other, more urgent, problems have had to be dealt with.
7) Forums : General Topics : Credits (Message 8460)
Posted 1 Jul 2009 by sygopet
Post:
I have being looking at my total stats on my 11 projects for the last 7 days on the boinc stats and see my daily credits have been 2477 2420 1725 2514 1598 1753 and 1894 so I have had 14381 credits in a week averaging at 2010 every 24 hours
I have one task for Cosmology running at the moment and has been running for 23 hours and 36 minutes is 93.685% and I get 420 credits for this.

So I have received 840 from Cosmology for 48 hours processing on one CPU against the 48 hour average of the other projects which is 3180 on the 2nd cpu

yesterdays 2477 did not include any cosmolgy figures

Paul

I think I get what you are saying . .
You are concerned that the credit return for your work with Cosmology is not up to the rest of your projects?
A few comments and questions:
Does your Cosmology get 50% of your processing resources? I don't think it does.
Judging from the stats I see, you have recently (about 2 weeks ago) substantially increased the proportion of your computer's efforts to processing Milky Way units such that most of your credits are now derived from this project and Cosmology is now a minor part of your "portfolio".
Now you may have discovered that MW provides a better rate of return but it is not fair to compare Cosmology with a collection of projects which is biased towards one which gives a high credit per hour figure. Your observations may be less a criticism of Cosmology (for giving a low return) and more an observation that another project is giving an inordinately high return.
Rather, compare your returns from Cosmology with, say, Climate prediction or Seti or one of the other more established projects and see how they match up.
I don't think 840 credits (420 x 2 cores) a day is too bad a return for your machine - even if you were running Cosmology full time.
Let's have some information based on individual projects.
8) Forums : Technical Support : Download problems (errors?) (Message 8436)
Posted 17 Jun 2009 by sygopet
Post:
What can I do?
Change my parameter maybe?

If your computer board will allow it, increase your RAM to 1GB minimum.
If not, find another BOINC project to work on.
We are told that the problem is being thought about - but don't expect any solution in the near future.
9) Forums : Technical Support : Download problems (errors?) (Message 8433)
Posted 16 Jun 2009 by sygopet
Post:
I'm afraid the answer to your question is in the message
CAMB needs 476.84 MB RAM but only 147.85 MB is available for use.

The current units are very hungry for memory and need, as far as I can see, at least 1GB of RAM to function.
Your computers are hidden but, judging from your RAC, its a good while since you have worked on a Cosmology unit. You have probably been getting this message for the last 2 months or more.
10) Forums : Technical Support : Memory workload (Message 8409)
Posted 4 Jun 2009 by sygopet
Post:
Oops, sorry everyone - I'm not concentrating!
Any reference in message 8408 to any project other than Cosmology is unintended and irrelevant.
I'm going back to catch up on my sleep.
11) Forums : Technical Support : Memory workload (Message 8408)
Posted 4 Jun 2009 by sygopet
Post:
I am starting work on the CAMB application today to reduce its memory and processor usage as much as possible.


Thanks for that update - it would be good to get my old machine back into action on Milky Way.

It occurs to me that if your (in-the-future) project has a need for lots more memory it could still be run as an optional "Double Creamy" Way, in the same way that Climate Prediction, Seti and possibly other projects have (or had) and allow the client to decide if they want to risk the potential thrombosis.

Or stick with the safer "semi-skimmed" (= half and half) for those of us with older machines under doctor's orders.
12) Forums : Technical Support : Memory workload (Message 8394)
Posted 1 Jun 2009 by sygopet
Post:
Do you feel that using a large amount of memory is a problem and if so, why?

No problem at all, if you want to exclude any computer without very large amounts of RAM from participating!
My P4 with 512MB RAM can only get:
Message from server: CAMB needs 476.84 MB RAM but only 459.07 MB is available for use.
It makes no difference how I juggle the memory usage settings. I would be interested to hear from anyone who has managed to get Cosmology to function on 512MB and how they did it.
Since the whole idea of BOINC is to allow work to take place in the background without (unreasonably) affecting the computers primary tasks I would say that Cosmology needs to go back to the drawing board and design work units that need less than 100MB RAM to function. (I had no such problem with app version 2.15 - can we return to that?!)
I support the originator of this thread and many others who have tried to get this message through.
There is no warning for anyone attempting to join the project that RAM is critical.
There is little evidence that Cosmology staff are taking note.
13) Forums : General Topics : Credits (Message 8382)
Posted 31 May 2009 by sygopet
Post:
And in Jan. at Cosmology@home I did 1 w.u. for 5 or 6 hours and got 5,120 credits

To be correct, you would have received 140 credits for your 5 or 6 hours in January for any completed w.u., plus 5000 "bonus" points simply for having an account in April. So the 420 credits for 18.5 hours last week is very much in line. (Current units take 3 to 4 times longer than before April)
Add a further 140 credits, for another unit presumably completed between January and April, and the grand total is the 5700 stated in your information panel.
You have nothing to worry about.
14) Forums : General Topics : Credits (Message 8376)
Posted 29 May 2009 by sygopet
Post:
I just completed a 18.5 hr. w.u. for a measly 420 credits.


Doesn't sound too bad to me - your computers are hidden so we can't compare your results but bear in mind that different units take different times to complete and (in my experience) there can be around a 3:1 ratio in time between the longest and the shortest.
It would be interesting to know what your average "credits per hour" comes to (including the shorter ones), what you consider to be a fair return for your computer's labour, and how that compares with what you are getting on other projects.
And what you are doing with all these credits when you've got them!

At least you are getting units to crunch - my P4 is effectively excluded from participating because of the inordinately high memory requirements.
15) Forums : Technical Support : More Credit Needed (Message 8074)
Posted 10 Apr 2009 by sygopet
Post:
On the contrary, bring the size of the units back to where they were a month ago (or unchange whatever you have changed) so they can be crunched by computers with 0.5GB RAM! Keep the credit allowance at 140 - that worked fine and gave a reasonable return, comparable with other projects.
Mostly, I would like to see some explanation given for making the change. I can understand and sympathise that hardware problems are behind the recent down time but stuffing up the units on return seems a poor move from this side of the line.
p.s. please put your clock right!
16) Forums : Technical Support : Problem getting work` (failed downloads) (Message 8018)
Posted 7 Apr 2009 by sygopet
Post:
Here's another puzzle for you all:

07/04/2009 15:24:26|Cosmology@Home|Message from server: No work sent
07/04/2009 15:24:26|Cosmology@Home|Message from server: CAMB needs 476.84 MB RAM but only 459.07 MB is available for use.

I don't see anything on the site to suggest a minimum RAM level and this certainly hasn't been a problem before.
17) Forums : Technical Support : URGENT Problems Thread (Message 7904)
Posted 6 Mar 2009 by sygopet
Post:
The newest WUs I got were very strange:
12787402, 12787400, 12787399, 12787372, 12787357

They finished after about 500 sec, got validated and received 360 credits each.
Something went incredibly wrong here. Cosmo was always granting too many credits, but this is outrageous.

These are \"test\" units (see front page latest news) which are rewarded with 3x the normal.
Since they take (from my one experience) less than a tenth of the usual time to complete I am inclined to share your view that too many credits are being awarded!
18) Forums : Technical Support : download of params_101908_003016_0.ini failed since 2 days ? (Message 7631)
Posted 13 Nov 2008 by sygopet
Post:


Hi everyone,
. . . A certain trend seems to show up on these failing WU\'s though and I wanted to your help to confirm. It looks to me like this happens on WUs that have been lying around in the db after creation for a few days and not picked up. So you will typically not see this on newer ones, ie. No WUs created within the past week should fail download.
Anshul


Sorry, Anshul - I see no correlation between \"age\" of workunit and the problems with downloading. Newly created units are, in my experience, just as prone to cause trouble. There will, indeed, be a greater proportion of failing units which are a week (or more) old but that is simply because the same unit has probably been around the system several times and, unless the user is in a position to spot, and abort, such units quickly there could be up to 10 days before it is timed out and resent to the next \"victim\". A temporary reduction to one failure (causes the unit to be removed from circulation) would cut down on the problems experienced considerably.
It seems to me that you have access to the database that should show if failing units are generated on particular days, or are from particular batches, or are truly random.
19) Forums : Technical Support : Long running WU aborted by project - no credit (Message 7616)
Posted 10 Nov 2008 by sygopet
Post:
.....look at the low claimed credit: only 560.80 for 816,696.90 CPU seconds.


Something is wrong! With your setup you should definitely be processing units within a few hours. I wouldn\'t have thought continuing with the others would have any worth either.
Could be worth trying a project reset.
It may be significant that two of the units (and possibly the third) you mention are ones where others have had downloading problems. so you\'ve probably just got rubbish units.
You might have a claimed credit of 560 but the standard awarded is just 140 at present.
20) Forums : Technical Support : Starved of work (Message 7604)
Posted 8 Nov 2008 by sygopet
Post:
[quote
You will . . . . get credit if you report first but late.
Phoneman1[/quote]

With a quorum of one, as soon as your time is up, the work unit gets errored out and a new one created. You don\'t get any credit if you report late, even by seconds (I know - I have experience!).
One of the (other) problems with Cosmology is the very non-linear \"progress\" with time indicated in Boinc Manager, so that it is difficult to impossible to predict whether a unit will finish processing before the deadline.


Next 20