BOINC & TeAm Anandtech ... where are we?

[saabcaptain]

i joined the TeAm some time ago and with my little part time Pentium have racked up 830+ WU on Seti@Home for the TeAm. BOINC is very exciting because #1 it allows me to have my computing time split between two projects and #2 it likely will provide new data sets for SETI to crunch, not the same data over and over again.

up until now my computer has run only SETI@Home via the non-screen saver version with SETI Driver and SETI Spy. i would like to switch to BOINC because i feel the time has come that to do something other than SETI so i have switched to BOINC and have put 33% of computing with BOINC SETI and 66% with ClimatePrediction.net ... both with TeAM Anandtech.

am i doing the right thing? are we still make a final push for the TeAm with the classic SETI? i am just so tired of doing data sets from 2003 and earlier that i know SETI has computed 100's of times over. my understanding is BOINC will never recompute the same sets more than 3 times. what are the problems with BOINC right now? also will BOINC ever update their snap shot from March which indicated my classic SETI had 600+ WUs to my more current 800+?

thanks... who has made the switch?

 

[Pokey]

saabcaptain

Setiboinc has been having "growing pains"..............

Look for posts by Rattledagger.

Also look at the Seti at Home main page. Below is from that site:

September 3, 2004
The SnapApp folks were here for most of the afternoon. They replaced the processors with ones that are faster and have larger caches. More importantly, we talked about how to stabilize the system. The plan is to simplify both the HW and service mix as much as possible in the hopes of removing the variable that is causing the hangs (and of course they will be going for the root cause at their lab). We altered the startup so that a bare minimal set of services are running. And we are going to only use the 18000 head unit for now, keeping no data on the sd30 expansion. This means moving some data around. We first have to move the upload/download data to a temporary volume on the sd30, then reconfigure the space on the 18000 head, then move the uldl to the head. The move to the temp volume is happening now. Tomorrow morning we will do the reconfiguration and start final move. With luck we will have the project up by EOD tomorrow.


So it looks like they are out of commission for a while longer.

 

[RaySun2Be]

I tried BOINC SETI for a bit, but because of all the continuing issues, I went back 100% to SETI Classic, primarily because our output has dropped significantly and if SETI Classic doesn't end soon, BBR may upset our #5 position. And I can't stand by and let that happen.

 

[Rattledagger]

Since you're sharing with CPDN, there shouldn't be any problems getting enough work.

And yes, the BOINC-version will only give the same wu to more than 3 people if someone doesn't return before their deadline, or crunches wrong/error. "Classic" as a comparison has now 7.233 result/wu.

As already mentioned, the Snap-box is still not working as it should, and therefore they've still got some performance-problems.
The v4.05-client also has some bugs, mainly in win9x, but also under NT. Most of these is fixed in v4.08 that should be up in alpha then the Snap is back up, but if client-problems under NT look at http://setiweb.ssl.berkeley.ed...rum_thread.php?id=3703

They're apparently planning to update the "classic"-score after "classic" is shut down...

As long as we're outcrunched by BBR in Seti for BOINC, it's doubtful these crunchers can help by going back to "classic"...

 

[saabcaptain]

Thanks,

So far BOINC is running well on my XP Home P4 2.4 machine. Sadly SETI is down right now due to the SNAP issue but should be back tonight they say. CPDN data downloaded and immediately began processing so that is a good thing and shows the major advantage of BOINC. Once SETI returns BOINC should automatically begin the 33/66% split of work load.

WHAT WE REALLY NEED is a BOINC OVERALL RACE. In other words they should have BOINC rank OVERALL computing power used towards ANY form of DC and give credit to a master race for TeAm AnandTech. That would not penalize any team if their users intend, like me, to devote their computing power to several groups at a given time.

 

[Rattledagger]

Some stats-sites is making cross-project-stats in BOINC, but they've got a couple of things to tweak yet...

 

[imported_Thunder]

saabcaptain,

I'm also still sort of trying to settle on my 'mix' of the BOINC projects. At the moment, I've biased my settings toward S@H, with enough towards CPDN to make sure even my slower machines don't fail to finish a WU in time (them are some big suckers, eh?!) to have them expire.

I'm trying to use the custom parameters for the 3 different 'groups' of computers (work, home & school) to get the right mix so as to keep all going the way I want and I think I'll end up with about 80% of my 'power' going to S@H and the rest going to a mix of CPDN and Predictor to cover the inevitable outages that S@H will have.

 

[imported_Thunder]

For what it's worth, S@H BOINC showed a brief glimmer of life for me about 1 hour ago.

1 of my clients just happened to get an answer back from the scheduler and attempted to d/l 8 WU's. 3 of them succeeded a 5 of them failed for various weird reasons.

I know it's not much, but I think it indicates that they're doing something and making progress.

No worries though... in the meantime, I'm predicting a heckuva lot about climate change.