Hi.
I'm quite new to Einstein@Home but have been BOINCing for well over a decade.
Please have a look at a post I made here - http://boinc.berkeley.edu/dev/forum_thread.php?id=8663&postid=50940#50940
On the m/c in question I now have other BOINC projects happily running on miltpile cores, but E@H stubbornly refuses to, whilst (of course) very happily running multiple tasks on my Win machines - it's just the MacBook that seems to cause this problem.
Any ideas?
Many thanks
Copyright © 2024 Einstein@Home. All rights reserved.
Problem running miltiple E@H tasks on a 4-core MacBook
)
Not really, apart from the obvious one of thermal throttling. Even so, I'd still expect to see all threads making some progress. FGRP2 tasks can take a while to write their first checkpoint so can appear not to be making progress initially. This could last for many minutes, even tens of minutes if the frequency was throttled right down. Other types of tasks should show progress in smaller increments and a lot sooner than FGRP2. In any case, it's very unlikely to be something specific to the tasks. Aborting the ones you have and getting a fresh batch will not make the problem go away.
If a task shows no percentage progress at all (only elapsed time incrementing), the time figure will always reset to zero if BOINC is stopped. It's only if a task can be restarted from a saved checkpoint that the time will reset to the checkpoint time and not to zero. How long did you see tasks running for with zero percentage progress still showing? If it was longer than 30 minutes then all I can assume is that your machine just can't handle E@H tasks in its current configuration. What happens if you set BOINC to use 50% of the available CPUs?
All Einstein tasks are quite compute intensive, which generates a lot of heat. The cooling systems on laptops are not really able to cope with this. How hot does your machine seem to be running?
I run this project on lots of different hardware and heat is a real issue for everything. Because of rapid failure rates, I gave up using any laptops a long time ago. I am running the project on a Mac mini, which scared me at first because of the small, monolithic, completely sealed looking enclosure. It looks more like a time capsule or an external hard drive enclosure rather than a complete system. It has a hyperthreaded quad core CPU so when BOINC fired up there were 8 tasks running, all making progress. The enclosure quickly started getting too warm for comfort so I reduced it to 4 cores. I did this by setting BOINC to use only 50% of the available cores. BOINC still sees it as an 8 core machine. With only 4 tasks running, the enclosure temperature felt a lot more reasonable. Its main job is an office file server and it's been running for almost 2 years without any issues so far. Even though it's heavily occupied crunching, it has no trouble keeping up with its file serving duties.
Sorry I can't be more helpful. Maybe someone else actually running the project on a similar Macbook to yours may be able to shed more light on this.
Cheers,
Gary.
Gary, hi. Many thanks for
)
Gary, hi.
Many thanks for your note.
Just to be contrary... my own experience of BOINC over a decade plus - possibly approaching two - is almost totally on laptops! I read the early cautions about overheating with care, but just pressed ahead anyway. For most of that time I've just run Climate Prediction, and only branched out recently with World Community Grid then Einstein.
I guess it's likely that earlier apps might have been less cpu-intensive and so generated less heat - I've never experienced heat problems in all those years - so this might be the first.
I'll keep trying and see what happens.
Cheers - Richard