The Binary Radio Pulsar Search in Parkes data will not be extended.
Can you say why not? I might be reading too much into it, but to me this implies something more than "we caught up on the available beams, and everyone is too busy with LIGO data to generate a new wider set of BRP templates to test against".
This is indeed the end of BRP6. The SSP shows when we expect the search to end, i.e. when we expect all tasks to be returned, not when we run out of work to send out.
........but what happens to the couple hundred I still have in progress?
Well, I believe your machines will keep crunching and uploading the results, just like before. Then those tasks will be validated. Nothing has changed in that. If a machine fails to finish a task before deadline, then the project server will cancel that task and a new task will be regenerated/resent to someone else. That's why there will be a few tasks still available now and then.
MAGIC Quantum Mechanic wrote:
Edit: funny thing but just now that host I mentioned just got a new task from Binary Radio Pulsar Search (Parkes PMPS XT) v1.57 (BRP6-Beta-cuda55)
Edit: funny thing but just now that host I mentioned just got a new task from Binary Radio Pulsar Search (Parkes PMPS XT) v1.57 (BRP6-Beta-cuda55)
Probably not new, but reissued. If you look at the last two characters of the task ID of BRP6 tasks coming to you now, you will likely find after the underscore a digit greater than 1. A review of the actual WU will likely show you that the task sent to you was an extra one created because one of the two original tasks was not returned by deadline, was aborted by the recipient, or that the initial quorum, while returned, failed to validate.
There will be lots of these for a while, but hungry hosts seeking only BRP6 will mop of the available supply rapidly after each appearance. Then the supply will dwindle. It may get a boost if the project issues extras in an attempt to finish off the stragglers.
Any reason why not release
)
Any reason why not release BRP4G v1.57 cuda55 windows_intelx86 for all windows system including x64 ?
-----
Bernd Machenschalk wrote:The
)
Can you say why not? I might be reading too much into it, but to me this implies something more than "we caught up on the available beams, and everyone is too busy with LIGO data to generate a new wider set of BRP templates to test against".
BRP6 Tasks to send: 0 Is
)
BRP6 Tasks to send: 0
Is this already the end? Seems sooner than expected.
-----
The SSP says we should be
)
The SSP says we should be good for a while.
BRP6 (PMPS XT) progress
This is indeed the end of
)
This is indeed the end of BRP6. The SSP shows when we expect the search to end, i.e. when we expect all tasks to be returned, not when we run out of work to send out.
BM
Yes it is official.......I
)
Yes it is official.......I have one machine with just one task left so I tried to get new work and getting nothing reminded me about this.
I guess I should read the rest of the thread again........but what happens to the couple hundred I still have in progress?
I hope we are switching over to a new GPU task version here asap.
Edit: funny thing but just now that host I mentioned just got a new task from Binary Radio Pulsar Search (Parkes PMPS XT) v1.57 (BRP6-Beta-cuda55)
MAGIC Quantum Mechanic
)
Well, I believe your machines will keep crunching and uploading the results, just like before. Then those tasks will be validated. Nothing has changed in that. If a machine fails to finish a task before deadline, then the project server will cancel that task and a new task will be regenerated/resent to someone else. That's why there will be a few tasks still available now and then.
Just like that.
MAGIC QUANTUM MECHANIC
)
Probably not new, but reissued. If you look at the last two characters of the task ID of BRP6 tasks coming to you now, you will likely find after the underscore a digit greater than 1. A review of the actual WU will likely show you that the task sent to you was an extra one created because one of the two original tasks was not returned by deadline, was aborted by the recipient, or that the initial quorum, while returned, failed to validate.
There will be lots of these for a while, but hungry hosts seeking only BRP6 will mop of the available supply rapidly after each appearance. Then the supply will dwindle. It may get a boost if the project issues extras in an attempt to finish off the stragglers.
RIP BRP6 Bump: Mumak
)
RIP BRP6
Bump:
-----
Any word if you will be
)
Any word if you will be reissuing the cuda 55 version?
Would like to see how it performs against the cuda 32 version.