Hey Folks this is not a Alpha Test its more Stable Then
15 Nov 2004 13:16:16 UTC
Topic 187094
(moderation:
another Boinc Project they say this is a final release.
never see before a Project that run Stable like this in a alpha phase ,client running fine--up and down load without errors for days,--fast answers from the admins ...
Hey Folks this is not a Alpha Test its more Stable Then
)
I agree. This thing has been stable as a rock. On my windoze XP smp box, I'm running S@H and Einstein@Home 50/50 and BOINC switches between them about every 60 minutes. At times it has even split between them, running one cpu on s@h and one on einstein@home.
Thanks for the kind words! If
)
Thanks for the kind words! If only you could see the chaos, sleepless nights and coffee consumption behind the scenes... !!
We still have a number of things to do. We are still making changes to the validator. Right now, some valid results are actually being marked 'invalid'. We are tracking down the reasons. We think there may be a small problem with 'unordered floating point comparisons' in our application, rather than with the validator.
Another important thing is that we want to modify our application to do twice as much work. Instead of simply searching the output from a single interferometer, as currently, we want to search the output of two different interferometers and then only return the 'candidates' that appear in BOTH.
We also want to compress our return files to make them smaller.
Finally, we need to make a linux executable that runs more efficiently. Our key routine (LALDemod) is faster in Win32 than Linux. We need to find the compiler switches and other options that will make this faster.
So I expect that we'll still be doing 'alpha testing' for another few weeks while we get these things sorted out. I intend to add another bunch of users once we have these things done. But I want to proceed with caution and not 'open up' Einstein@home to the general public until we are really 100% ready to do science with hit. Hopefully this will be sooner rather than later!
Director, Einstein@Home
> Thanks for the kind words!
)
> Thanks for the kind words! If only you could see the chaos, sleepless nights
> and coffee consumption behind the scenes... !!
A great project non the less in spite of the fact that boinc is still bleading edge.
>
> We also want to compress our return files to make them smaller.
I am wondering if you should also look at using the boinc zip library to compress the download stream as well.
I did a test compress of the E@H boinc files and most compress very well.
This may start to help with the low speed modem issues.
The other issue with low speed users is that boinc opens a numuber of downlado connections, i have seen 6 open tranfers which is bad on low speed modem links.
>
> Finally, we need to make a linux executable that runs more efficiently. Our
> key routine (LALDemod) is faster in Win32 than Linux. We need to find the
> compiler switches and other options that will make this faster.
Good, the linxu crunchers will eb happy with better performance of the linux side.
>
> So I expect that we'll still be doing 'alpha testing' for another few weeks
> while we get these things sorted out. I intend to add another bunch of users
> once we have these things done. But I want to proceed with caution and not
> 'open up' Einstein@home to the general public until we are really 100% ready
> to do science with hit. Hopefully this will be sooner rather than later!
>
As one might say, i little change then a look back...
73 de Peter VK3AVE
... and the thing I love
)
... and the thing I love about Einstein is that it has *lots* of work (so far) and reliable up/downloads (so far), unlike certain others which have been offline in recent weeks (LHC and Predictor, you know who you are...). Even CPDN has (minor) problems - I've been unable to upload a completed WU (640+ hours on a 3GHz P4) for two days due to a uni Berne server being offline. Perhaps future versions of BOINC will be more robust, eg after 3 failed upload tries with the default server it tries a fallback server... and then another... something better than the current single-point-of-failure. But I must admit its the first such occurrence in my 14 months with CPDN.
Cheers,
PeterV.
I agree too. the project is
)
I agree too. the project is really great . When do you think to be in final/public release ?
Congratulations e@h team ;)
> I agree too. the project is
)
> I agree too. the project is really great . When do you think to be in
> final/public release ?
> Congratulations e@h team ;)
>
Hmmm... 2005 is the World Year of Physics... so would January 1st 2005 be feasible? I know that's only a couple of weeks away, but it might help the E@H project on the PR side. If not it would be 2nd best to at least have a firm public release date by January 1st 2005.
http://www.physics2005.org/events/einsteinathome/
- Marco
- Marco
Team Canada
I would rather they wait to
)
I would rather they wait to make sure that the system is 100% ready before they open it up. I was part of the initial 2000 users at LHC and then it was running great, even still when they upped the limits it still ran great. They thought they were ready, they opened it up to the public and it worked great for a few weeks, then it didn't.