Not receiving Gamma ray Pulsar tasks
23 Jun 2015 13:52:41 UTC
Topic 198134
(moderation:
Hi, I'm new to BOINC (been using since mid-May) so I apologise if my question has already been answered.
I have not received any Gamma ray pulsar search tasks since I joined the project. Do I have some kind of problem, or there simply isn't any gamma ray work at the moment ?
Thanks.
Not receiving Gamma ray Pulsar tasks
)
As long as you have it selected in your Einstein@home preferences you should get some eventually.
The scheduler prefers to send out Gravity wave search tasks more than Gamma ray tasks as stated here by Bernd.
In his post GW = Gravity wave search and FGRP = Gammar ray
I see. I just thought
)
I see.
I just thought something might be wrong since I got none on my other computer, and my team's public account (working on several hosts at our uni's computer lab) hasn't received any either. I'll try disabling GWs for a while and see if I get any.
Okay, they run, but I'm
)
Okay, they run, but I'm losing progress. I'm not switching my laptop off yet my progress dropped from about 14% to 0.714% while I was away for a while. This hasn't happened to the other task types.
RE: ... my progress dropped
)
Three main factors cause this, firstly, FGRP4 tasks take a while between checkpoints, secondly, BOINC will simulate progress until a first checkpoint is actually written at which point the real progress will be known, and thirdly, When you start tasks for a different science run, the estimates for crunch time may be wildly inaccurate.
If your very first FGRP4 task had a wildly low estimate, BOINC's simulated progress would be unrealistically rapid. When the first checkpoint was saved to disk, BOINC would have corrected the 'simulated' progress to 'real' progress. 14% seems unusually high but if that number was ticking over continuously (an increment every second) that is a sure sign of simulated progress. I have seen 0.714% as a progress value when the first checkpoint is written so that value doesn't surprise me at all. There are other things going on in the early calculations that interfere with a proper estimate of progress in the very early stages. After 0.714% has been static for a while (no more simulated progress) it should jump to about 3% or so when the next checkpoint is saved. After that, each successive checkpoint causes a jump of about the same magnitude.
The full story is rather more complicated and I don't know for sure all the details so I wont speculate further. However, you shouldn't see any further regressions but it will take a while between checkpoints (perhaps 10-30mins, depending on the speed of your processor). You shouldn't see the drop in progress on subsequent tasks, you will see more accurate simulated progress with the final simulated value much closer to the real value when the first checkpoint is written. You can always tell the difference between simulated progress and real progress. Only simulated progress changes every second.
Cheers,
Gary.
That seems to be the case - I
)
That seems to be the case - I observed most of the things mentioned and the tasks are now running normally. Thanks !