And another thread to match.

Just noticed that the new version of the application, deployed on 03 January 2013, is writing its checkpoint files into the project directory rather than the slot directory.

No effect on processing, because they're read back from there when needed too. But normally, they are automatically deleted at the end of the run when BOINC cleans and recycles the slot directory for re-use. So there's no need for explicit deletion of the checkpoint file.

But there's no automatic recycling of the project directory - so the files, one per task, are beginning to accumulate and use a lot of space.

Thank you very much for reporting this!

Just in time ...

BM

Should be fixed in the 1.04 release. Please have a look!

BM

OK, that looks better. With v1.04, I'm getting a file called checkpoint.cpt in the slot folder. With v1.03, they were names like 'h1_0447.15_S6GC1__S6BucketLVEa_447.263682292Hz_2819_1_0.cpt' and in the project folder - at least with the distinct names, they didn't get in each other's way when I was running multiple tasks.

Running task is h1_0447.95_S6GC1__S6BucketLVEa_448.063682292Hz_2830_1, if you want to check the output - should be done by 19:00 UTC.

Thanks a lot!

BM

Running task is h1_0447.95_S6GC1__S6BucketLVEa_448.063682292Hz_2830_1, if you want to check the output - should be done by 19:00 UTC.

Completed and reported.

Strangely enough, they complete in about 17 hours in Einstein@home running on my HP laptop with an AMD APU E-450 at 1.65 GHz and in about 20 hours on my main Linux box with its Opteron 1210 at 1.8 GHz here. Is there a difference between the WUs?
Tullio

---