'Aborted by user' instead of 'Error while computing'

[Sutaru Tsureku]

SETI@home have now a new strange WU quota system.
Yes, maybe it need some time for to work well.

I get for example ~ 100 WUs/day, although my PC need at least ~ 191 normal WUs/day.
Some time and he will idle.


I don't know how this quota will be disturbed because of VLARkill.
Maybe this user will never again get WUs..


If I had a wingman with Linux and Crunch3r's CUDA app, I saw his app do 'Aborted by user' instead of 'Error while computing'.
Maybe it's possible that the Windows CUDA app do this also like this?
Then I guess the quota isn't disturbed.


Yes, I know, ReSchedule tool.. ..but not all members know this program, or don't want to use it..

[Sutaru Tsureku]

This would protect the science also for the worst case, that a WU with the WOW signal will be deleted from the DB because of 6 errored results (VLARkilled and other possible errors).

[Richard Haselgrove]

If they know enough to install the VLARkill app, they can be educated to use ReSchedule instead.

Sutaru is right to say that VLARkill will block any attempt to download more than 100 tasks a day, however many validations occur - while there are any VLARs in the work stream, of course.

@ Jason,

How easy would it be to put together an interim version of the unified installer, with stock applications only for CUDA? And including cuda_fermi - we don't want to be seen to be adding to the project's troubles.

And the 64-bit <platform> tag troubles seem to be back - I'm trying to get a handle on just what the problem is.

[Josef W. Segur]

...
If I had a wingman with Linux and Crunch3r's CUDA app, I saw his app do 'Aborted by user' instead of 'Error while computing'.
Maybe it's possible that the Windows CUDA app do this also like this?
Then I guess the quota isn't disturbed.
...

Bad guess, anything other than a "Success" with a valid result file will reduce consecutive_valid to zero according to the source code available from the BOINC repository. I can't think of any reason they would have changed that when patching the new methods into the S@H Scheduler code.

Exactly how the S@H Scheduler is reacting to zeroed consecutive_valid isn't clear yet, but it's apparently one factor in limiting a host to 100 tasks per day. Obviously a host which is producing bad results needs to be throttled more than that.
                                                                                      Joe