Seti is available for BOINC, but wu crashes immediately.

[Rattledagger]

Copy from Boinc-beta mailing list:

***********************************
David Anderson
[Boinc-beta] SETI@home on BOINC


We added the Windows SETI@home app to the beta test.
It will be disseminated as we add work units -
Eric Korpela is finishing up the BOINC version
of the SETI@home splitter. We?ve been delayed by several days because of last-minute changes to the formats of WU and result files.

We?re aware of the bugs in our recent release -
no AP graphics, right-click options missing -
and are working on them. We?ll have another release soon.
Thanks for your patience.

-- David
**************************

Note, it's still just the current beta-testers that's got access, and currently none seti-wu is available... But atleast the windows-exe is released.

 

[PraetorianGuards]

Goods news! Thanks Rattledagger for keeping us updated.

 

[Assimilator1]

Thanks for the latest news RD
I hope someone is building a SETIQ replacement!

Btw its spelt 'available'

 

[LANMAN]

Originally posted by: Assimilator1
Thanks for the latest news RD
I hope someone is building a SETIQ replacement!

Btw its spelt 'available'

LMAO !!!

--LANMAN

 

[lobadobadingdong]

Originally posted by: Assimilator1
Thanks for the latest news RD
I hope someone is building a SETIQ replacement!

Btw its spelt 'available'

LOL, not if your elmer fudd

 

[Rattledagger]

There's some seti-wu now, but looks like there's a fatal bug, since all wu I've tested terminates immediately with "Unrecoverable error for result (process exited with a non-zero exit code)"

Better to wait for next version...


Grumble... since the BBR-forum didn't spit out some red on awailable, I didn't look closer on the spelling...

 

[Assimilator1]

hehe ,no probs

Btw ,shame about the 1st batch of S@H2 WUs ,I hope they fix it soon!.

Any indications of WU times for S@H2 yet?

 

[Rattledagger]

Well, since all wu is bad, no-one have crunched anything...
But if the sceduling-servers expected time is something to go by, this is 4h41m29s, and this machine uses 3h40m on normal wu under v3.03...
Then again, some of the wu has angle-range of 2,7 and one at 0,598, but all has the same "expected" time... I can't say anything certain before they fixes the wu.

The actual data-size of the current bad wu is 354.990 bytes, while the header is some longer than current seti-wu, so total lenght is 361.967 bytes give or take a few...

The latest info is this (from the beta-list):

"As you're probably aware, there were problems with the
SETI@home work units we sent out today.
They may put the application into an infinite loop,
in which case you should "Reset Project" to get rid of them.
Eric Korpela is currently hard at work fixing this problem,
and we should be producing correct work units soon.

David" (Anderson)

 

[Assimilator1]

Scheduling servers?

 

[Rattledagger]

Time to read up on BOINC.

Anyway, then connecting to BOINC, you're normally talking to the Scheduling-server, and based on your machine-specs & the benchmark you get assigned wu for download from other server(s). From the benchmark, the scheduling-server give a more or less accurate guess on how long the wu will take, and also based on this it controls how many wu to make ready for download or none if example the time-limit of all awailable wu is 7 days and your machine is too slow...
After crunching a wu, the result is returned to yet another server.
If the queued wu in BOINC dips below the min-#-cache-days, or if one of the wu is nearing the time-limit, the scheduling-server is contacted again, reporting on it's progress of current wu, returned wu, failed wu, and gets assigned new wu.
Normally, the scheduling-server isn't contacted for every uploaded result.
The scheduling-server also controls if wu needs to be re-issued to others in case not enough results returned.

 

[lane42]

If ia'm not mistaken, this sounds a little like the way Folding @ Home works with time limits and benchmarks. I bet each machine will have a machine ID#.

 

[Assimilator1]

Interesting ,though it sounds like there will no longer be a 'typical WU' to compare WU times?

 

[Rattledagger]

Yes, every BOINC-install has a machine-id.
Every wu has also a size-requirement & memory-requirement, so it's not just the speed that's counted...

Well, atleast in seti, my guess is the crunch-time will just like before be influenced by the angle-range, so getting a range of wu-times just like now.
The biggest difference from seti is that instead of 1 "point" for every crunched wu regardless of VLAR or VHAR, in BOINC the 5-hour VLAR will give 25% more points than the 4-hour VHAR. Unfortunately, this also will influence the credit of turbo-wu. Oh well, getting 0,001 or something for a 5-second wu shouldn't be a large problem.

 

[lane42]

Thanks RD

 

[Smoke]

Originally posted by: lane42
Thanks RD

And don't sweat the spelling ...

 

[Rattledagger]

The latest update: (beta-list)

"We've disabled SETI@home 2.00 due to a bug -- the next version
will be released soon.

It is normal that you download any application versions (whether
Astropulse or SETI@home) associated with this project
automatically.

--
Karl Chen 2003-10-25 17:24"


BTW, "get preferences" is the method to connect to the scheduling-server immediately. Atleast if not always connected, this is the solution to get assigned more wu, and reporting back on the returned / bad wu you've done. Of course also then changed the preferences.