Searching for pulsars in PALFA data from Arecibo
13 Dec 2008 21:00:04 UTC
Topic 13729
(moderation:
We are starting some limited public testing of a new pulsar search on Einstein@Home. This search uses data from the PALFA collaboration, taken at the Arecibo radio observatory. Science information will be available in this thread in the Science Message Board area.
Please use this thread (which is in the Problems and Bug Reports Messsage Board area) to report bugs and problems with this new search. This is especially useful during our initial public testing!
Bruce Allen
Searching for pulsars in PALFA data from Arecibo
)
There is no need for you to do anything, the tasks will be issued randomly between the usual "HierarchicalSearch" tasks. The number of tasks will be very limited during the testing phase (628 workunits at a time), it is rather unlikely you get one. The name of the application will be "einsteinbinary_ABP1" instead of the current "einstein_S5R4". You will be able to opt-out from the "Arecibo binary pulsar search" in your Einstein@home preferences (once the App is in the database, which will be some time tomorrow afternoon CET). Users currently running the Windows Beta App will not automatically get ABP1 tasks, I'll publish instructions and a new app_info.xml in the Beta App thread in the next days. For various technical reasons the ABP1 App will not be available for Mac OS X on PowerPC.
BM
BM
The first 628 "ABP1"
)
The first 628 "ABP1" workunits are being produced right now and should be delivered in about 1h.
For this to happen our scheduler runs in a new "mixed" mode, which could break something for the normal operation (S5R4 workunits), too. We did our best to test this, but haven't done this before on the scale of the whole project.
On a machine that needs about 9h10m for a S5R4 task a ABP1 task ran 6h12m.
BM
BM
RE: It would be good idea
)
That's a good idea. We'll modify the server status page to show this.
Two issues are (now) known
)
Two issues are (now) known for this application:
1. On Windows 98/ME the communication between the worker and the screensaver isn't working, i.e. you'll see the starsphere, but not the data that's usually displayed. That's a common problem to all MinGW-built Windows Applications (i.e. S5R4 Beta Apps, too). We're working on this.
2. The App had a bug that occasionally lead to "general access violations" / "segfaults", preferably on Windows. The bug was in the part of the application that communicated with the screensaver, it didn't affect the scientific computation. This bug has been fixed and a new application version was published, Tasks that are resent because they errored out previously should be run with the new version of the application (minor version 02).
BM
BM
The server that serves the
)
The server that serves the ABP1 workunit data files shut itself down for an emergency at about 19:30 CET. We are investigating, it will probably not be back up before tomorow. ABP1 workunits downloaded now will probably end up as download errors (nope, they are not served by the mirror network).
BM
BM
The ABP1 validator stumbled
)
The ABP1 validator stumbled over a result file that apparently forced it into an infinite loop. The program has been diabled until the problem has been fixed. No credit for ABP1 workunits in the next days.
BM
BM
The ABP1 validator has been
)
The ABP1 validator has been fixed and is running again.
BM
BM
RE: Now that ABP1 has been
)
The handling of ABP1 results will be moved to a different machine, for some time the server status page won't be able to show the daemon status correctly.
Yes.
BM
BM
RE: Windows XP , BOINC
)
How stupid is that? templates_400Hz.bank is a plain ASCII file!
BM
BM
RE: RE: I just noticed
)
Are these Intel or PowerPC iMacs?
BM
BM