Message boards :
Number crunching :
OK, So I think I'm getting this....
Message board moderation
Author | Message |
---|---|
Michael Brennecke Send message Joined: 2 Apr 04 Posts: 33 Credit: 205,887 RAC: 0 |
I was a S@H1 user a few years ago and now I just started again about 2 months ago. Managed to crank out just under 1000 in 2 months. Not bad I guess. Anyway... switched over to BOINC right away last night when I saw it went live. Now I think I understand that 2 other results need to be returned by others for any of us to get credit. I've returned about 45 results worth 450+ credits "pending". I think I've got that straight. Problem... when I look at the details of the work units it seems like I'm the only one listed to have returned one of the parts. How could this be? I have only found one WU where one of the others were also returned. If I'm just being impatient I guess I wait it out but is the current userbase so low or are they receiving units and not returning them? Someone give me confidence that I'll at least see some of these pending credits become real. I mean, is there a predicted loss of pending vs. total? 50%? Are we all seeing this? Sorry... lots of rambling. |
Michael Brennecke Send message Joined: 2 Apr 04 Posts: 33 Credit: 205,887 RAC: 0 |
OK, I read a few comment in other topics that it can take weeks sometimes for a credit to be given. However it still seeems like I'm the only one returning part 1 of 3 of my WU. Huh? |
Michael Lockhart Send message Joined: 17 Apr 03 Posts: 2 Credit: 15,459 RAC: 0 |
Hi. Wow, you must have several machines to have pumped out 45 results that quickly. I've got one computer, and I've managed 9 work units with almost 300 credits pending. Of the 9 WUs I turned in, two of them have been returned by one of the two other people to whom the work unit was assigned. Although that means only 23% or so have been returned by two people, remember that it's only been a little over 24 hours that BOINC has been up. Maybe there are a lot of people out there whose computers aren't as fast as ours. Maybe lots of them are busy running other programs, so BOINC can't get the full CPU load much of the time. I'd suggest checking after a couple days and seeing whether the situation has improved. I'm just happy I caught that notice the moment it went up. I think I was switched over and running within about a half hour. :-) Now if they could just get something like SetiSpy or SetiMonitor for BOINC. Michael |
EclipseHA Send message Joined: 28 Jul 99 Posts: 1018 Credit: 530,719 RAC: 0 |
> OK, I read a few comment in other topics that it can take weeks sometimes for > a credit to be given. However it still seeems like I'm the only one returning > part 1 of 3 of my WU. Huh? > > Other users (2 or 3) need to complete the WU with the validator saying "OK". Then, the lowest "Requested Credit" is given to those users (I claimed 20, you claim 10, we both get 10 - say user 3 claimed 21). Now, let's say that users 4 and 5 did the same WU - they get ZERO! Seems that the only happy camper will be you, who claimed 10 and got 10! But the other 4 users expended a bunch of cycles, and get credit for part or none of what they did! That's how screwy this whole scheme is! "It's about the science and not the users!" Seti Classic = Coke Classic BOINC = New Coke Where's new Coke today? |
Michael Brennecke Send message Joined: 2 Apr 04 Posts: 33 Credit: 205,887 RAC: 0 |
Just 3 P4 machines. (1) 2.4/512 and (2) 2.8HT w/512 so I suppose it's like 5 CPU's working but based on some other benchmarks I read I'm only average or less. You have 300 credits pending for only 9 WU's? Wow, this system is going to be hard to predict for sure. I don't think we need anything like Driver, Spy or Monitor, it seems to me that those features are already built in. Michael |
Byron Leigh Hatch @ team Carl Sagan Send message Joined: 5 Jul 99 Posts: 4548 Credit: 35,667,570 RAC: 4 |
|
Byron Leigh Hatch @ team Carl Sagan Send message Joined: 5 Jul 99 Posts: 4548 Credit: 35,667,570 RAC: 4 |
OPPS , sorry Double post , erased , apologizes |
Michael Lockhart Send message Joined: 17 Apr 03 Posts: 2 Credit: 15,459 RAC: 0 |
The trouble with Boinc Spy is that it requires that you have Microsoft NET Framework 1.1 installed, which I do not. I try to avoid MS products whenever possible, although I have XP. I have a 2.8GHz HT PC with 1GB of RAM. The work units take differing amounts of time, so I have anywhere from 28 to 37 credits per work unit. How do you consider those features "built in"? I like seeing which Gaussians, for example, have been found in a work unit, the angle range, where the WU is in the sky, and I like having a list of this for all wus done. I don't see *any* of those as it is now. Michael |
AthlonRob Send message Joined: 18 May 99 Posts: 378 Credit: 7,041 RAC: 0 |
> Other users (2 or 3) need to complete the WU with the validator saying "OK". > Then, the lowest "Requested Credit" is given to those users (I claimed 20, you > claim 10, we both get 10 - say user 3 claimed 21). No, that's incorrect. The way it works is, with more than two results returned, the highest and lowest numbers are dropped. The average of the remaining requested credit numbers are averaged and *that* is how much credit is granted. As these units are all set for a quorum of 3, the middle person's claimed credit is what is granted. > Now, let's say that users 4 and 5 did the same WU - they get ZERO! Seems that > the only happy camper will be you, who claimed 10 and got 10! But the other 4 > users expended a bunch of cycles, and get credit for part or none of what they > did! That's how screwy this whole scheme is! If users 4 and 5 are given the same workunit, they'll get the same credit granted as everybody else. As it stands right now, though, I don't think any of the workunits are set to send themselves out that many times. Please get your facts in order before calling everything screwy. Rob |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.