KWSN Orbiting Fortress

[Nuadormrac]

It won't even send me non-beta WUs anymore; but yesterday, for me was the kicker wrt their beta units. Now I've run SETI beta, RALPH, and a bunch of other beta or even alpha projects before; and never had quite this experience. If we weren't talking POTM, for which we're trying to out shrub other teams though, I'd just write it up to a poor project setup, and make beta reports.

The project sent me a 109 estimated time to completion WU, BUT ONLY GAVE 1 DAY and 6 HOURS TO DEADLINE FROM THE TIME OF DOWNLOAD. Now I know this laptop's 1.8 GHz core 2 duo processor isn't exactly the newest thing on the block; but it isn't that out-dated either. A proc which is 2 or even 3 times faster couldn't have processed it, and given it only used 1 CPU core, not both, a quad core CPU would have been of no benefit.

So umm Eon admins; here's a clue. If you want longer running tasks, so people don't hit up your servers for tasks as often increase the fracken deadlines, to match up with the longer time to completion. Running 24/7 it shouldn't take like the lattest proc out there, crunching that unit 100% of the time, without the computer even being used like 2-3 times longer then the deadline on the task. Bad, bad, bad.... For that size task, 1 week deadline (which is incidently 168 hours) would be reasonable.

Oh, and the joys; in getting BOINC to unload other tasks, without exiting and reloading BOINC (due to the lack of check pointing) so the VM wasn't running in the background sucking up CPU cycles; it obliterated a 50 hour LHC 2 task, which thankfully at least wasn't half done. Umm, if they want long tasks, they really need to take a look at deadlines and adding checkpoints, BEFORE rolling them out....

I disabled beta projects after this; either it's gonna give me regular tasks, where lack of check pointing but shorter task is all I have to concern myself with on when I do/don't load it, or it's not sending one. After going to bed, and waking up again, the progress indicator was still 0.000% complete and over 100 hours left on estimated completion, with the deadline like 20 hours or so away...
_________________


.

[Gemjunkie]

The regular WUs were taking 2-3 hrs. but downloaded with a 200 hr. estimated completion time. It would become realistic over time but might bounce back up again, especially if you stopped crunching the project for a day and restarted.

The project admins are living in their own little dream world where all computers crunch their project 24/7/365.
_________________




(older, before split CPID)

[Nuadormrac]

That wasn't going to complete in no 2-3 hours. The thing ran like 7 or 8 hours, and didn't make any progress on the progress bar. I actually wouldn't have a hard time believing it was going to do that. But tbh, I didn't want my credit quotent for the day to be that severly hampered; and after the LHC unit got trashed no thanks to that task, I wasn't in a mood to lose even more credit/potential

Now I did have a beta task estimate at 8 hours and complete in 1, so it's possible for it to shrink; but even a 10:1 ratio (it was more like 8:1 for the other beta) means there should have been some progress.... I could have gone to work and given it another 8 hours, and then the computer would have HAD to shut down to move it, so meh....

Oh, and just to compare, the 51 hour LHC task, they had a 1 month deadline on the thing... Oh, and they checkpoint also... But then they've been doing BOINC projects since the early days of BOINC; the admins on Eon, not sure they have the experience to have figured this one out
_________________


.

[Nuadormrac]

If finally sent some non-beta units (or should I say it isn't set to get beta units); and they're all giving the problem mentioned at the end of this thread:

http://theory.cm.utexas.edu/forum/viewtopic.php?f=9&t=1115#p3839

They all go to computation error immediately, and every time it gives an improper function call, error. That's like a compiler type error one gets when trying to compile source code; there's something seriously wrong with the coding of those WUs it keeps sending to me...

I've NNT the project for the time being, just in case they penalize clients returning errored results by lowering the download quota on tasks. The problem is on the project's end, when it's task source code errors that are getting spit out on the task results text...
_________________


.

[LanDroid]

We passed Russia and should pass Sicituradastra and 3 Million points tomorrow... Currently in 15th place

[LanDroid]

Either I was groggy in the previous post or there was a large stat revision - KWSN is about 700K away from Sicituradastra and 3 Million points.

'Stones:
DoW passes 1,200,000
Sir Stooper passes 40,000
drasmussen passes 8,000

[Killerrabbit]

Maybe you were drinking too many PGGBs.


Ni
_________________

[LanDroid]

It does not appear that eOn was a very popular PotM...

Final Stats:
15th Place +5
2,483,371 Points +715,485
720,430 Points crunched last 28 days +688,943
109,449 Points shrubbed yesterday
5th Place Formula Boinc +32

[Nuadormrac]

Well, it did have low credit returns for work crunched (so what we did, didn't pay as well as other projects); and then there was a lot of headaches and problems, project downs, beta work, that seemed less stable then some projects alphas, etc....

Well, we tried.... Here's hoping this month's PotM holds up better/is less headache causing wrt project issues....
_________________


.

[Gemjunkie]