Rampant resending
15 Jul 2006 23:46:32 UTC
Topic 191574
(moderation:
Hi, is it normal to have a work unit sent out to a raft of computers? It seems a real waste of resource when only 2 or 3 are required for a quorum.
The following work unit (at the time of writing) was sent out again on 6, 11, 12, 13, 14 and 15 July, yet the only timed-out reply was on 26 June.
http://einsteinathome.org/workunit/9651496
So, up to 4 of the computers seem to be wasting their time?
Rampant resending
)
That is a S4, which went to 3 systems at a time. Why it has gone nuts, I am unsure.
RE: That is a S4, which
)
That was the most extreme example but now you mention it, it seems to be normal for S4 runs from my recent and limited experience as a newbie! Maybe this is a deliberate policy to throw them at a large number of computers to complete the runs ASAP:
"June 16, 2006
We are beginning large-scale distribution of workunits from LIGO and GEO's fifth science run S5. We have finished generating the remaining S4 workunits, and expect that the processing of these S4 workunits will be completed during the next two to three weeks."
This wasn't normal behavior
)
This wasn't normal behavior for S4, but I think Graham's right about the idea being to finish off the last of the work asap so that they can start analizing it.
As of right now (from the
)
As of right now (from the status page):
S4 WU (no canonical result) 93
There are now only a handful
)
There are now only a handful of remaining S4 workunits (27 at the last check). I am gradually increasing the number of target results for these in order to get them done as fast as possible, so that the postprocessing of the S4 data can get underway in earnest.
Bruce
Director, Einstein@Home
RE: I am gradually
)
Hi Bruce, thanks for clearing up the puzzle!