In last days ATLAS have a "poor" recent credit... :(
Actually we shut down the BOINC backfill on ATLAS for a while, as it seemed it was giving the project database a hard time. We are still investigating what could be done in the configuration to avoid this in the future.
Well I'm late to the question but happy with the decisions made to have atlas crunch on it's own, when it crunches. Of course I have recently brought the awesome power of my new iMac 3.06 Core 2 duo and my older 1.66 Core duo mac mini (a whopping 4 cores total) back online and will be for the foreseeable future on Einstein.
Actually we shut down the BOINC backfill on ATLAS for a while, as it seemed it was giving the project database a hard time. We are still investigating what could be done in the configuration to avoid this in the future.
BM
If ATLAS is stressing E@H too hard, couldn't this be solved by some reconfiguring? e.g.:
1) Add another BOINC project to ATLAS so it isn't always working on E@H
2) Reduce the number of cores per node, or the number of nodes, that run BOINC (perhaps limit BOINC to 1 core per node?)
3) Move E@H to ATLAS, and use the current E@H servers as a mirror (like the LIGO labs), and get them to run E@H as backfill
Actually we shut down the BOINC backfill on ATLAS for a while, as it seemed it was giving the project database a hard time. We are still investigating what could be done in the configuration to avoid this in the future.
BM
If ATLAS is stressing E@H too hard, couldn't this be solved by some reconfiguring? e.g.:
1) Add another BOINC project to ATLAS so it isn't always working on E@H
2) Reduce the number of cores per node, or the number of nodes, that run BOINC (perhaps limit BOINC to 1 core per node?)
3) Move E@H to ATLAS, and use the current E@H servers as a mirror (like the LIGO labs), and get them to run E@H as backfill
Or, may be(?) start E@H tasks not on all ATLAS nodes simultaneously, but gradually? For example each day start tasks on next 10 nodes. 300 nodes per months and "ATLAS per quarter"? :)
There's been a recent increase in ATLAS activity again
Yep. Increasing the work cache of the account (i.e. all clients) apparently did help.
However at the very moment ATLAS hasn't much spare time to contribute, only two nodes are running BOINC.
BTW: As our scheduler fills ordinary jobs from the "front", I'd expect nodes with higher numbers to get the highest RAC. But I'm afraid ordinary user accounts can't see the node names / IP addresses.
There's been a recent increase in ATLAS activity again
Yep. Increasing the work cache of the account (i.e. all clients) apparently did help.
However at the very moment ATLAS hasn't much spare time to contribute, only two nodes are running BOINC.
BTW: As our scheduler fills ordinary jobs from the "front", I'd expect nodes with higher numbers to get the highest RAC. But I'm afraid ordinary user accounts can't see the node names / IP addresses.
BM
My AMD Opteron 1210 at 1.8 GHz running Linux completed a WU in 46k s against 26k s of one of the ATLAS nodes, which is a Xeon quad at 2.4 GHz. Not bad.
Tullio
Does anyone recognise the user name 'Britta Daudert'?
If he's bringing a massive cluster online, then that's fine: otherwise, someone should check out hosts for user 282952 - he seems to be bringing a new one online every few seconds - 4978 and counting.
Does anyone recognise the user name 'Britta Daudert'?
If he's bringing a massive cluster online, then that's fine: otherwise, someone should check out hosts for user 282952 - he seems to be bringing a new one online every few seconds - 4978 and counting.
Well, based on the name I think that 'he' is in this case 'she' :-)
Thats quite a lot of muscle fro Einstein, hopefully we can utilize that kind of 'monster-clusters'
Hello! In last days ATLAS
)
Hello!
In last days ATLAS have a "poor" recent credit... :(
RE: In last days ATLAS have
)
Actually we shut down the BOINC backfill on ATLAS for a while, as it seemed it was giving the project database a hard time. We are still investigating what could be done in the configuration to avoid this in the future.
BM
BM
Well I'm late to the question
)
Well I'm late to the question but happy with the decisions made to have atlas crunch on it's own, when it crunches. Of course I have recently brought the awesome power of my new iMac 3.06 Core 2 duo and my older 1.66 Core duo mac mini (a whopping 4 cores total) back online and will be for the foreseeable future on Einstein.
Team MacNN - The best Macintosh team ever.
RE: Actually we shut down
)
If ATLAS is stressing E@H too hard, couldn't this be solved by some reconfiguring? e.g.:
1) Add another BOINC project to ATLAS so it isn't always working on E@H
2) Reduce the number of cores per node, or the number of nodes, that run BOINC (perhaps limit BOINC to 1 core per node?)
3) Move E@H to ATLAS, and use the current E@H servers as a mirror (like the LIGO labs), and get them to run E@H as backfill
RE: RE: Actually we shut
)
Or, may be(?) start E@H tasks not on all ATLAS nodes simultaneously, but gradually? For example each day start tasks on next 10 nodes. 300 nodes per months and "ATLAS per quarter"? :)
There's been a recent
)
There's been a recent increase in ATLAS activity again:
CU
Bikeman
RE: There's been a recent
)
Yep. Increasing the work cache of the account (i.e. all clients) apparently did help.
However at the very moment ATLAS hasn't much spare time to contribute, only two nodes are running BOINC.
BTW: As our scheduler fills ordinary jobs from the "front", I'd expect nodes with higher numbers to get the highest RAC. But I'm afraid ordinary user accounts can't see the node names / IP addresses.
BM
BM
RE: RE: There's been a
)
My AMD Opteron 1210 at 1.8 GHz running Linux completed a WU in 46k s against 26k s of one of the ATLAS nodes, which is a Xeon quad at 2.4 GHz. Not bad.
Tullio
Does anyone recognise the
)
Does anyone recognise the user name 'Britta Daudert'?
If he's bringing a massive cluster online, then that's fine: otherwise, someone should check out hosts for user 282952 - he seems to be bringing a new one online every few seconds - 4978 and counting.
RE: Does anyone recognise
)
Well, based on the name I think that 'he' is in this case 'she' :-)
Thats quite a lot of muscle fro Einstein, hopefully we can utilize that kind of 'monster-clusters'
-- Lundi --