Binary Radio Pulsar Search (Parkes PMPS XT) "BRP6"
Bernd Machenschalk
Moderator
Administrator
Joined: 15 Oct 04
Posts: 4312
Credit: 250459996
RAC: 35134
9 Feb 2015 13:52:09 UTC
Topic 197965
(moderation:
The first few hundred tasks of the BRP6 search have been generated and sent out.
It was announced a few days ago in the science forum and is meant to supersede the BRP5 search as the main GPU-based search on Einstein@Home.
BM
BM
Binary Radio Pulsar Search (Parkes PMPS XT) "BRP6"
)
Hi all,
in case you're wondering what BRP6 (PMPS XT) will be doing in detail: I have just added a short scientific explainer in the thread in the science forum.
Cheers,
Benjamin
Einstein@Home Project
I have some of those running
)
I have some of those running on my BRP5 machine and I'm surprised how well they blend in. You wouldn't notice them except for the different name and the longer run time, and that's no issue since the flops value has been adjusted appropriately. I'd say anyone running BRP5 can just activate BRP6 and forget about it.
RE: The first few hundred
)
How are BRP6 workunits packaged wrt BRP5? Memory? GFlops?
Gord
Is there a limitation
)
Is there a limitation currently on how many of these units are being issued? I have not yet received any, and I see there are none on the server to send.
Thanks.
Yes I just was looking at the
)
Yes I just was looking at the status of all of my hosts and noticed these new tasks so I did a quick search to find this thread.
Tried to get some to test but I guess I was too late so far and then checked the server after stopping by here.
http://einstein.phys.uwm.edu/server_status.html
Not running at this time.
Hi! That's how we usually
)
Hi!
That's how we usually start a new search or "run": release just a few hundred or so workunits (typically with a short deadline), wait for the results to come in, look at the results to check everything is OK (e.g. the validator is adjusted correctly etc). If everyone involved gives a "GO", then the big valve will be opened and there will be enough work for everyone ;-).
Cheers
HB
Did get one on my Mac. Runs
)
Did get one on my Mac. Runs significantly slower than Perseus. The latter normally 9 to 10 hours on this machine. Parkes now runs for 25.5 hours at 87% now. So about factor 3 longer in duration when completed. No idea why.
Allright ! I got one Parkes
)
Allright ! I got one Parkes PMPS XT and it crunched just fine.
Waiting for that there Big Valve to be opened )
Bill
The run time of Bill's jobs
)
The run time of Bill's jobs match the 30% increase in run tine that was intended. The runtime from Xandro BA is clearly excessive, I wonder whether the GPU throttled down perhaps.
HB
For the record. It was the
)
For the record. It was the only task running for most of the time. In beginning ran 1 or 2 grp4 tasks one after another. Turned grp4 off after that hoping the runtime would shorten. No other projects running. Did not use the machine at all for anything else.