Well, I mostly understand that the OM types are difficult to estimate. Though I think the most aggravating parts for me about this situation is the LOW estimate assigned, the LONG time it takes to process the WU, and then the low to extremely low credit that is delivered.

I think assigning a real world estimate would at least make me feel a little better - that number actually being whatever the current formula results in times six or seven.

Oh, good, an explanation. I was starting to worry something was wrong with my machine when I saw one WU stuck at 75% after 15 hours, and another at 63% after 11.

Given that I currently look to be getting ~1% progress an hour and the deadline's in about 18 hours, hopefully it'll speed up again at some point. If not, I guess I'm going to have to babysit this for awhile to make sure any short deadline ones that come up are started immediately. It'd be really nice if the estimated times were at all reliable.

I am having the same issue as powr ..getting more repairs than the machines can run despite a .5 day cache, driven by still having a much smaller estimate of time to complete than is realistic and some of these are running close to 28 hours. Would it be reasonable during the transition to bump the time to complete errors to 5 or 7 days? I would hate to lose the 39.4 points I will likely get when the tasks complete :-) OK so I too am a bit unhappy with point scores!

This is what I initially thought too!
I was away for a while and only now noticed boinc had silently closed on an older i7-920 with 15 OET WUs in progress. One is 9.5 hours in and is 54% completed. Hopefully it can complete these in time.

If I get a too late flag does this class me as unreliable for all hosts or only that one until a certain number of successful and on time work units are sent in? Headed somewhere with questionable internet for a while and would rather at least try and get these done.

Edit: Oh dear. Intel Atom processors taking 40 hours to successfully complete WUs and only getting 39 points. I hope this will improve, though as someone else pointed out, at least the work is getting done.

Sorry, but I think that you are one of those that get this all wrong. At least from my POV...

I don't care about getting points per se, if I would, I would rather have my hosts crunch all for Einstein@Home. One host of mine that does gets alone up to twice as many points per day than all my WCG hosts combined.

But the points are the most obvious statistic that is visible in BOINC and a serious drop off in points usually indicates that there is a problem with any of the hosts. That might not be an issue to quickly verify if you just have one or two hosts at home, but it is an issue if you have a few dozens (like I do, with a lot of them remote). Than this requires at least some additional time to verify, in particular for me, as I had issues with some OET WUs randomly 'getting stuck" (no check point, running up to 99.xx% but no CPU time accumulated), taking up "a slot", keeping other WUs from getting processed, on top of the issue with at times vastly incorrect runtime estimates, which overall is just wasting time and resources.

So it isn't as simple as branding people complaining/being concerned as point ****, as some other regulars are doing as well...

Ralf

*edited to appropriate forum language - ErikaT