optimized applications
2 Oct 2006 16:56:45 UTC
Topic 191891
(moderation:
Hello,
I'm new to Einstein@Home (but not to boinc).
Is using an optimized science application considered as cheating or just better use of the CPU?
Is there a performance increase compared with the current regular version of the client (I read it had been optimized lately)?
(I'm talking about the SSE3 41.05 optimized application)
Thank you!
optimized applications
)
That optimized client was for the client albert 4.37 that was used before and want function today with the new client einstein_S5R1 4.24, if I´m properly informed.
RE: Hello, I'm new to
)
The current Einstein applications are listed on this page: you're looking at the bottom section, "Faster all-sky pulsar search". The application numbers are currently in the range 4.08 to 4.28
Where are you reading 41.05? What site offered it to you for download? Are you sure it's intended for the current - S5 - search on the Einstein project? It doesn't sound like it.
In general, third party applications are currently strongly discouraged on Einstein: there is concern that the scientific results may not be accepted by the scientific community through the 'peer review' process.
If you look at this post, you'll see that:
a) a lot of optimisation work has already been put into the mainstream downloadable applications.
b) the main recent optimiser (Akos F.) is working on the 'inside' with the project team, and between them they reckon they've done all the easily-accessible optimisation work.
thanks for the
)
thanks for the answers.
I'm not properly informed about every aspect of einstein yet, so I'm not exactly sure what app it is.
I'm not gonna use an optimized version.
The "SSE3 41.05" App was
)
The "SSE3 41.05" App was derived from "albert 4.37" and worked fine for the previous analysis run "S4".
For the current analysis "S5R1" we are using a new application "einstein_S5R1" that is incompatible with the old "albert". However the ideas that sped things up that time have been incorporated in the current official Apps (4.17 - 4.28). There is no "optimized" App that runs faster than the current official Apps (while giving reliable results).
BM
BM
Hello ; i see you talk about
)
Hello ; i see you talk about "Applications of a third party" ; Is Einstein@home an open source project like SETI or leiden classical ?
RE: Hello ; i see you talk
)
Whereas the framework EAH runs under is GPL (BOINC) the science apps themselves are not.
HTH,
Alinator
RE: Hello ; i see you talk
)
As Alinator says, the Einstein applications are not open source.
My reference to third party applications refers to a period - roughly April-June this year - when applications optimised by Akos Fekete were available for download, for a different scientific search from the one we're currently engaged in.
Akos didn't work from the (unreleased) source: he dis-assembled the distributed binary, optimised it and re-assembled it, to the lasting admiration of many (not least the project administration - who hired him!).
However, the current search is under closer scientific scrutiny, and this (dare I say?) amateur approach is no longer welcome. As I said previously, Akos is now working on the 'inside', optimising the official project apps downloaded automatically from the project. Which has the happy side-effect that every cruncher on Einstein is now using an optimised app, rather than just the inquisitive or proactive few.
[Apologies to any users of minority operating systems, to whom the preceeding sentence may not apply].
RE: However, the current
)
I'd refine that further - by deduction from my observation of events - a 'non-collaborative' approach is no longer welcome, meaning that blessing is definitely required from the 'inside'. Those who seek to redefine the validity of the science product ( including what 'validity' means ) - however well-intentioned, correct, skillful, clever or otherwise meritorious - ought first seek/attain collaborative status. :-)
And no, I'm not on the 'inside' ( nor will I pretend or seek to be ), so please let no one assume that from my 'gift' of moderator role.... :-)
And yes, participation in E@H is definitely 'collaborative' in the everyday sense, and ever so valuable for the project ( so keep crunching team! ), but it's not the sense I am referring to here.
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
RE: However, the current
)
That's me. I am running Einstein 4.16 on my Pentium II/SuSE 9.3 Linux box, after briefly trying 4.17, and I never had an invalid result. Slow but sure, or "Festina lente" in Latin.
Tullio
ok , thanks.
)
ok , thanks.