|
|
I'm running several machines on Rosetta via BOINC and was looking through my results. I notice that almost a quarter of the WU's sent to me are being returned with errors and no credit given. Most of them return after less than a minute of processing but a couple used up 2 hours. My initial thought was that the problem was with my machines, although none of them are O/c'd and are perfectly stable otherwise. So I looked further at other's results and I find I'm not in a minority with having results returning with an error and no credit. Can anyone explain this to me? I realise that this is a project that is basically testing the Rosetta s/ware so are these errors useful in some way? 
|
|
-------------------------------
|
|
|
|
|
Ignore this, answered my own question - I really should learn to use the forums more Seems to be a temporary glitch in the WU's 
|
|
-------------------------------
|
|
|
|
|
Ignore this, answered my own question - I really should learn to use the forums more  ....  Yup you should....bin going on for days now 
|
|
-------------------------------
|
|
|
|
|
I'e had literally hundreds of them fail.... Just had another 5 or 6 awhile ago.... 
|
|
-------------------------------
|
|
|
|
|
Yup most are out of the network now but I also still get a burst of 3 or 4 a couple of time a day! 
|
|
-------------------------------
|
|
|
|
|
Ignore this, answered my own question - I really should learn to use the forums more  ....  Yup you should....bin going on for days now  I'm still a Rosetta n00b Looks like I'm in good company though 
|
|
-------------------------------
|
|
|
|
|
What worries me more is the repeated 'No work from project' messages. Thankfully I keep a couple of days work cached so haven't yet run out completely but tonight, for instance, my Boinc Manager has failed to get any new work for over 5 hours. In the same time I have returned 6 results. It's currently requesting 21476 seconds of work. That's approaching 6 hrs of work.
|
|
-------------------------------
|
|
| IP: [
Logged ]
|
|
|
|
|
It is stated that it is a boinc problem, not a rosetta problem, and like you I have never run out of work as it eventually gets some before I do...
|
|
-------------------------------
|
|
|
|
|
It is stated that it is a boinc problem, not a rosetta problem, and like you I have never run out of work as it eventually gets some before I do... Thanks Scribe. Perhaps it should be noted that anyone entering a BOINC based project should always carry a couple of days worth of cache. Finally got some work when I was up to 50,000+ secs requested. 
|
|
-------------------------------
|
|
| IP: [
Logged ]
|
|
|
| IP: [
Logged ]
|
|
|
|
|
Project was down this morning but all seems OK after about 11:00 UTC
|
|
-------------------------------
|
|
|
|
|
Strange enough I have Rossette WU's reporting a 'client error' on only one of my machines. The other ones are having no problems at all. This seemingly is a machine problem, but I don't really understand. It's a normal laptop, no overclocking or other this, plain and simple standard. It's got 512 Mb of internal memory and a 1,73 GHz Intel Centrino processor.
What seems to be my problem?
|
|
-------------------------------
|
|
|
|
|
It could be that the "Shared" Memory being used for the video on your notebook has dropped your available ram below the point that boinc will send it rosetta work.
|
|
-------------------------------
|
|
|
|
|
That could be something, because it does have shared memory.
|
|
-------------------------------
|
|
|
|
|
I don't recall ever having Boinc loaded on my notebook, but I do know the UD server wouldn't send it any HPF work when I was working Grid with it. I'm still toying with the idea of increasing my notebooks ram to 1 gig, but don't really use it for much except F@H. Puts the purchase in the catagory of "it happens when there is a really good deal at the store". 
|
|
-------------------------------
|
|
|
|
|
The strange thing is: the laptop worked fine with HPF from UD. No memory problems, no problems calculating and no problems getting credited for the results (which indicates that there are no corrupted WU returned). This can only mean that the Rosetta (boinc) is even more memory intensive than HPF 
|
|
-------------------------------
|
|
|
|
|
If I remember correctly, Rosetta@home wants 512 MB while grid.org wants 256 MB for HPF. I have a Dell with 512 MB that uses 128 MB for video (shared) so it only has 384 MB available, which is not enough for Rosetta@home. I keep telling myself that I ought to upgrade it, but I am not in any hurry.
|
|
|
|
|
|
|
Don't be tempted to run as is. I burned out my Dell's hard drive w/in the week.  It really DOES need more. 
|
|
-------------------------------
|
|
| IP: [
Logged ]
|
|
|
|
|
For the moment I'm letting Rossetta rest for a while Somehow none of my machines seem to be 'strong' enough te produce usable results. Only one accepted result against many 'computer errors'. No more, until I have a more powerfull machine, or the jobs are lighter 
|
|
-------------------------------
|
|
|
|
|
If you're still around etienne, there is a new BOINC client available that gives you much more control over size of files being sent to you and, consequently, the length of time your PC takes to process them. (It's actually the time you set and the job size is adjusted accordingly but Hey! it's all the same end result) Check out the R@H website/forum for more info. PS:- For all you crunchers who insisst on a level playing field, No it isn't an Optimised Client 
|
|
-------------------------------
|
|
| IP: [
Logged ]
|
|
|
« previous next »
Forum Index Unread Posts
|
| Print
|
|