How about a third option: Not crunch Einstein at all, but attach it to all the other projects out there that need their work done? (As part of a team on its own)
If that's a totally laughable solution, then add me to the no team option.
Is Bruce still switching between random teams with his cluster, or did he stop doing that?
How about a third option: Not crunch Einstein at all, but attach it to all the other projects out there that need their work done? (As part of a team on its own)
If that's a totally laughable solution, then add me to the no team option.
Oh, I like the idea. But ATALS was built and funded for data analysis @AEI, which includes Einstein@home as an AEI project, but no other (BOINC) projects.
Quote:
Is Bruce still switching between random teams with his cluster, or did he stop doing that?
As I recently wrote in some other thread the contribution of Bruce/"Nemo" to Einstein@home is negligible by now. I think he (i.e. his account) is still stuck with team "Ireland". He once wanted to write a script that does the team change automatically, but never got around to that. He asked us to do this for ATLAS, but I thought I'd better ask first what people want.
From what it currently looks like ATLAS will run as 1342 quad-core machines (and possibly a few more management machines) on a single user account that will join no team at all. I don't think that that "user" will keep the machines hidden.
BM
Berndt,
Doesn't this mean that the current (11,165,000) wu will be processed significantly sooner than the 1 year estimate and the project will all be over in a matter of months?
Are there more runs of data yet to be processed (eg S5R5, S5R6 or S6R1 perhaps)?
From what it currently looks like ATLAS will run as 1342 quad-core machines (and possibly a few more management machines) on a single user account that will join no team at all. I don't think that that "user" will keep the machines hidden.
BM
Berndt,
Doesn't this mean that the current (11,165,000) wu will be processed significantly sooner than the 1 year estimate and the project will all be over in a matter of months?
Are there more runs of data yet to be processed (eg S5R5, S5R6 or S6R1 perhaps)?
While ATLAS is a hell of a machine, teh E@H volunteer network itself is also huge. According to a rough estimate, I'd say that if ATLAS was running 24/7 exclusively for E@H, it would increase the output E@H by a factor of 1.5, which would cut the S5R4 runtime to ca. 8 - 9 month.
But.... "only" some idle time of ATLAS will be used, so the speed-up will be smaller. But I would not be surprised it ATLAS could cut the run short by maybe 2 or 3 month.
There's also some room for further optimization in the Windows app (already implemented in the Linux and OSX apps) that might also shorten the run by a few weeks.
OTOH, it's possible that the current projection of S5R4 length is, however, still too optimistic. At the start of the run (because of the DCF problem), some clients received more work than their fair share and went into EDF / high priority mode. Those hosts will afterwards pause to request E@H work.
So the total duration of S5R4 is still everybody's guess, but if I had to make a bet I'd put my money on something around 12 months, including ATLAS and some more app optimizations.
So the total duration of S5R4 is still everybody's guess, but if I had to make a bet I'd put my money on something around 12 months, including ATLAS and some more app optimizations.
Ok, I'd bet on ~10 months.
However we recently found that we might be able to increase the sensitivity of the S5R4 search a little further with a new run, so we might not fully complete S5R4 if that turns out to be true (simulations are running right now to investigate this). This would just require a re-design of the workunits, no changes to the Apps needed.
Back to the original topic: ATLAS has been attached to Einstein@home; the user account #342084 will not join any team.
Back to the original topic: ATLAS has been attached to Einstein@home; the user account #342084 will not join any team.
Now go fiddle with the "On multiprocessors, use x % of the processors" and set it to 50 CPUs... see if BOINC can be set to 0.9314456% and if it then indeed uses 50 CPUs only. ;-)
(Posted in jest. I was using ATLAS as an example in an email to David Anderson to compare what is more intuitive to use when you want to set it to use 50 CPUs only... to put in 50 as a number, or to put in that percentage... waiting for an answer. ;-))
Hi All,
My vote is for teamless. Otherwise I bet there would have to be a whole subset of Crunchers Corner devoted to fielding issues arising from "random" selections. I think Bernd's time amongst others would be better spent speeding up wu production than tweaking the ATLAS scoring system.
There's also some room for further optimization in the Windows app (already implemented in the Linux and OSX apps) that might also shorten the run by a few weeks.
Given the volume of Pentium 4 and newer systems, I'd say there'd be at least a 10% reduction. If expected runtime was 300 days, then that would be 30 days reduced. My anecdotal observations (non-proven / theorized "evidence") based on my Athlon64 system though indicates that there should be around a 20% improvement, and that was without any additional boost that SSE2 might bring.
Anyway, the point you were trying to make is very true...that the systems of the general user base are still very "powerful", and thus the reason for distributed computing in the first place... :-)
Quote:
OTOH, it's possible that the current projection of S5R4 length is, however, still too optimistic. At the start of the run (because of the DCF problem), some clients received more work than their fair share and went into EDF / high priority mode. Those hosts will afterwards pause to request E@H work.
Only if those systems were allocated to other projects. Some people concentrate on a single project. For example, my Pentium 4 is attached to 3 projects now, but it is currently only doing Einstein work, as SETI and Cosmology are both set to not get any new work.
One other thing to remember is that you also have rotating "team project of the month" clubs...and that too could reduce the amount of time.
How about a third option: Not
)
How about a third option: Not crunch Einstein at all, but attach it to all the other projects out there that need their work done? (As part of a team on its own)
If that's a totally laughable solution, then add me to the no team option.
Is Bruce still switching between random teams with his cluster, or did he stop doing that?
Hello! I vote for
)
Hello!
I vote for ATLAS@"Einstein at work".
RE: How about a third
)
Oh, I like the idea. But ATALS was built and funded for data analysis @AEI, which includes Einstein@home as an AEI project, but no other (BOINC) projects.
As I recently wrote in some other thread the contribution of Bruce/"Nemo" to Einstein@home is negligible by now. I think he (i.e. his account) is still stuck with team "Ireland". He once wanted to write a script that does the team change automatically, but never got around to that. He asked us to do this for ATLAS, but I thought I'd better ask first what people want.
BM
BM
RE: From what it currently
)
Berndt,
Doesn't this mean that the current (11,165,000) wu will be processed significantly sooner than the 1 year estimate and the project will all be over in a matter of months?
Are there more runs of data yet to be processed (eg S5R5, S5R6 or S6R1 perhaps)?
BOINC blog
RE: RE: From what it
)
While ATLAS is a hell of a machine, teh E@H volunteer network itself is also huge. According to a rough estimate, I'd say that if ATLAS was running 24/7 exclusively for E@H, it would increase the output E@H by a factor of 1.5, which would cut the S5R4 runtime to ca. 8 - 9 month.
But.... "only" some idle time of ATLAS will be used, so the speed-up will be smaller. But I would not be surprised it ATLAS could cut the run short by maybe 2 or 3 month.
There's also some room for further optimization in the Windows app (already implemented in the Linux and OSX apps) that might also shorten the run by a few weeks.
OTOH, it's possible that the current projection of S5R4 length is, however, still too optimistic. At the start of the run (because of the DCF problem), some clients received more work than their fair share and went into EDF / high priority mode. Those hosts will afterwards pause to request E@H work.
So the total duration of S5R4 is still everybody's guess, but if I had to make a bet I'd put my money on something around 12 months, including ATLAS and some more app optimizations.
CU
Bikeman
RE: So the total duration
)
Ok, I'd bet on ~10 months.
However we recently found that we might be able to increase the sensitivity of the S5R4 search a little further with a new run, so we might not fully complete S5R4 if that turns out to be true (simulations are running right now to investigate this). This would just require a re-design of the workunits, no changes to the Apps needed.
Back to the original topic: ATLAS has been attached to Einstein@home; the user account #342084 will not join any team.
BM
BM
RE: Back to the original
)
Now go fiddle with the "On multiprocessors, use x % of the processors" and set it to 50 CPUs... see if BOINC can be set to 0.9314456% and if it then indeed uses 50 CPUs only. ;-)
(Posted in jest. I was using ATLAS as an example in an email to David Anderson to compare what is more intuitive to use when you want to set it to use 50 CPUs only... to put in 50 as a number, or to put in that percentage... waiting for an answer. ;-))
In my opinion, I would agree
)
In my opinion, I would agree with most that ATLAS should remain teamless.
Dave
oops... guess I was a little late!
Hi All, My vote is for
)
Hi All,
My vote is for teamless. Otherwise I bet there would have to be a whole subset of Crunchers Corner devoted to fielding issues arising from "random" selections. I think Bernd's time amongst others would be better spent speeding up wu production than tweaking the ATLAS scoring system.
Joe B
RE: There's also some room
)
Given the volume of Pentium 4 and newer systems, I'd say there'd be at least a 10% reduction. If expected runtime was 300 days, then that would be 30 days reduced. My anecdotal observations (non-proven / theorized "evidence") based on my Athlon64 system though indicates that there should be around a 20% improvement, and that was without any additional boost that SSE2 might bring.
Anyway, the point you were trying to make is very true...that the systems of the general user base are still very "powerful", and thus the reason for distributed computing in the first place... :-)
Only if those systems were allocated to other projects. Some people concentrate on a single project. For example, my Pentium 4 is attached to 3 projects now, but it is currently only doing Einstein work, as SETI and Cosmology are both set to not get any new work.
One other thing to remember is that you also have rotating "team project of the month" clubs...and that too could reduce the amount of time.