I've completed a few task from version 1.02 and all seems fine so far, but as there doesn't seem to be any validator running I can't tell if they will validate against results from other hosts.
Edit: Changed the thread title because the original problem is solved and the thread has developed to a general feedback thread.
One thing we need to test most is the workunit generator, which in a sense is very different from the ones we used before. When we find it doesn't produce the workunits as we want it, we will need to cancel the workunits and start over again, even if there is nothing technically wrong with the workunits it already produced. We try our best to detect errors as early as possible, in this case before too many tasks are sent out, but I suspect canceling S6CasA workunits will happen rather frequently in the next few days. After all, this is our test project.
Thanks for the information about the cancelled work units. I noticed that even running ones have been cancelled, but if you don't need those work units, then you don't need those work units.
I have at least 13 work units now that "can't validate" due to the work unit being cancelled, so I will get no points for them I suppose, even though they were successful and appeared error free.
I know it is a test project and understand about fine tuning and cancelling work to get the results you are after.
I have run a few other test projects so understand your reasoning.
But it would be nice if some credit is given for the successful testing of these work units as they take many hours to run.
Just happened to open the Windows task manager and noticed that each of the three running S6 Directed Search tasks was using ~600 MB of RAM each, compared to the S6 LineVeto search over at Einstein where each task use ~110-115 MB of RAM.
I'm now wondering if this is to be expected of the new run or is this temporary during testing?
I don't mind as I have 16 GB of RAM but other volunteers will probably have less RAM available.
Just happened to open the Windows task manager and noticed that each of the three running S6 Directed Search tasks was using ~600 MB of RAM each, compared to the S6 LineVeto search over at Einstein where each task use ~110-115 MB of RAM.
I'm now wondering if this is to be expected of the new run or is this temporary during testing?
I don't mind as I have 16 GB of RAM but other volunteers will probably have less RAM available.
One of the (many) differences between this and the previous run is that work units will vary a lot more, both in memory consumption and in the data volume that is required for a specific task. Among the currently generated test work units will be some of the most demanding in this respect, kind of a worst-case test.
BOINC allows to configure limits both for RAM and disk usage by BOINC, so it's possible with some care to make sure that BOINC will not take more ressources that the user is prepared to contribute.
I attached my i7-2600K/GTX460/HD7770 host to Albert@home to generate a RAC so i could post my Android Bug report,
i initially got a CPU, Nvidia and AMD Gamma-ray pulsar search Wu each, then i got some SSE2 Gravitational Wave S6 Directed Search (CasA) Wu's, but the scheduler timed out with an Internal Error,
some of these Wu's got resent on the two scheduler requests, a little later these tasks got cancelled by the project, the stderr.txt of Wu's started says so too:
Quote:
7.0.64
aborted by project - no longer usable
[/b]
2013-05-26 11:32:25.0107 (6716) [normal]: This program is published under the GNU General Public License, version 2
2013-05-26 11:32:25.0107 (6716) [normal]: For details see http://einstein.phys.uwm.edu/license.php
2013-05-26 11:32:25.0107 (6716) [normal]: This Einstein@home App was built at: May 24 2013 11:16:55
But the task page is wrong, it says the user aborted them:
The combination of the timed out Wu's and the Cancelled by Project Wu's have lowered my Max Tasks perday for this app_version to 1,
while i accept the timed out Wu's probably should lower this, if the project cancels them then it shouldn't.
At a first glance on my machine the new run takes about 11000s or 3 hours longer per task to run than the LineVeto search over at Einstein but got the same amount paid. I suspect that this is going to be adjusted before we go live on Einstein, right?
Now if only a few more would like to run some task then this new GW-search should be ready for Einstein before the current one runs dry.
Apropos, why are the auto-format buttons like [url] so far off-screen to the right here? I'm using Chrome on Windows XP.
I didn't want to spam the boards with my stats - just milestone theads - but apparently signatures are no longer optional. Follow the link if you're interested.
Apropos, why are the auto-format buttons like [url] so far off-screen to the right here? I'm using Chrome on Windows XP.
It's the same on IE 10 on Win 7
Claggy
Hi!
There will be a mojor overhaul of the Web presence in the near future, so forgive us for not investing any significant time in fixing stuff in the legacy web front-end now.
I've completed a few task
)
I've completed a few task from version 1.02 and all seems fine so far, but as there doesn't seem to be any validator running I can't tell if they will validate against results from other hosts.
Edit: Changed the thread title because the original problem is solved and the thread has developed to a general feedback thread.
One thing we need to test
)
One thing we need to test most is the workunit generator, which in a sense is very different from the ones we used before. When we find it doesn't produce the workunits as we want it, we will need to cancel the workunits and start over again, even if there is nothing technically wrong with the workunits it already produced. We try our best to detect errors as early as possible, in this case before too many tasks are sent out, but I suspect canceling S6CasA workunits will happen rather frequently in the next few days. After all, this is our test project.
BM
BM
Thanks for the information
)
Thanks for the information about the cancelled work units. I noticed that even running ones have been cancelled, but if you don't need those work units, then you don't need those work units.
I have at least 13 work units now that "can't validate" due to the work unit being cancelled, so I will get no points for them I suppose, even though they were successful and appeared error free.
I know it is a test project and understand about fine tuning and cancelling work to get the results you are after.
I have run a few other test projects so understand your reasoning.
But it would be nice if some credit is given for the successful testing of these work units as they take many hours to run.
Thanks
Conan
Just happened to open the
)
Just happened to open the Windows task manager and noticed that each of the three running S6 Directed Search tasks was using ~600 MB of RAM each, compared to the S6 LineVeto search over at Einstein where each task use ~110-115 MB of RAM.
I'm now wondering if this is to be expected of the new run or is this temporary during testing?
I don't mind as I have 16 GB of RAM but other volunteers will probably have less RAM available.
RE: Just happened to open
)
One of the (many) differences between this and the previous run is that work units will vary a lot more, both in memory consumption and in the data volume that is required for a specific task. Among the currently generated test work units will be some of the most demanding in this respect, kind of a worst-case test.
BOINC allows to configure limits both for RAM and disk usage by BOINC, so it's possible with some care to make sure that BOINC will not take more ressources that the user is prepared to contribute.
Cheers
HB
I attached my
)
I attached my i7-2600K/GTX460/HD7770 host to Albert@home to generate a RAC so i could post my Android Bug report,
i initially got a CPU, Nvidia and AMD Gamma-ray pulsar search Wu each, then i got some SSE2 Gravitational Wave S6 Directed Search (CasA) Wu's, but the scheduler timed out with an Internal Error,
some of these Wu's got resent on the two scheduler requests, a little later these tasks got cancelled by the project, the stderr.txt of Wu's started says so too:
But the task page is wrong, it says the user aborted them:
All Gravitational Wave S6 Directed Search (CasA) tasks for computer 7514
At Seti Beta, the cancellation of Wu's causes a Cancelled by server message:
Error tasks for computer 45274
The combination of the timed out Wu's and the Cancelled by Project Wu's have lowered my Max Tasks perday for this app_version to 1,
while i accept the timed out Wu's probably should lower this, if the project cancels them then it shouldn't.
Claggy
First result validated, nice!
)
First result validated, nice!
Link to workunit #371351
At a first glance on my machine the new run takes about 11000s or 3 hours longer per task to run than the LineVeto search over at Einstein but got the same amount paid. I suspect that this is going to be adjusted before we go live on Einstein, right?
Now if only a few more would like to run some task then this new GW-search should be ready for Einstein before the current one runs dry.
RE: But the task page is
)
Sorry, missed that one - I was busy elsewhere.
That's due to a missing update to result.inc on the web server.
Full details at Status 'Cancelled by server' changed
Apropos, why are the auto-format buttons like [url] so far off-screen to the right here? I'm using Chrome on Windows XP.
I didn't want to spam the boards with my stats - just milestone theads - but apparently signatures are no longer optional. Follow the link if you're interested.
http://www.boincsynergy.com/images/stats/comb-3475.jpg
RE: Apropos, why are the
)
It's the same on IE 10 on Win 7
Claggy
RE: RE: Apropos, why are
)
Hi!
There will be a mojor overhaul of the Web presence in the near future, so forgive us for not investing any significant time in fixing stuff in the legacy web front-end now.
Cheers
HB