FYI : to any researcher that cares, problem with WU 618924673

Message boards : Number crunching : FYI : to any researcher that cares, problem with WU 618924673

To post messages, you must log in.

AuthorMessage
Richard de Lhorbe

Send message
Joined: 17 Aug 09
Posts: 5
Credit: 3,013,955
RAC: 0
Message 77370 - Posted: 21 Aug 2014, 1:54:16 UTC

This workunit created the following condition on my Mac ... it gradually increased it's RAM use until it used all available memory, and then proceeded to gradually use all available virtual memory (something approaching 100 Gb), eventually stopping the other three processes on the CPU (a 2 core CPU with hyperthreading = 4 cores). The indicated CPU time was about 7 hours to reach only about 2/3rds completion ... but I don't think this was accurate, it was actually quite a bit longer in relation to the impact on my computer's ability to work on WU's, but not registered properly due to the memory issue. Recovery took some time, including re indexing the hard drive before I could run any other WU's. Hopefully someone can learn something from this to make future WU's behave better. The other volunteer's computer on this WU also had a failed WU, but did not seem to have the same computer issue, as they only took 498 seconds.

Cheers !
ID: 77370 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Chunfu Xu

Send message
Joined: 2 Oct 13
Posts: 2
Credit: 8,816
RAC: 0
Message 77372 - Posted: 21 Aug 2014, 5:30:23 UTC - in response to Message 77370.  

I am sorry for the inconvenience that my work unit brought you. We have identified the problem that caused the memory usage grew exponentially as the increase of protein sizes, and will try to avoid this problem in the future.

This workunit created the following condition on my Mac ... it gradually increased it's RAM use until it used all available memory, and then proceeded to gradually use all available virtual memory (something approaching 100 Gb), eventually stopping the other three processes on the CPU (a 2 core CPU with hyperthreading = 4 cores). The indicated CPU time was about 7 hours to reach only about 2/3rds completion ... but I don't think this was accurate, it was actually quite a bit longer in relation to the impact on my computer's ability to work on WU's, but not registered properly due to the memory issue. Recovery took some time, including re indexing the hard drive before I could run any other WU's. Hopefully someone can learn something from this to make future WU's behave better. The other volunteer's computer on this WU also had a failed WU, but did not seem to have the same computer issue, as they only took 498 seconds.

Cheers !

ID: 77372 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Message boards : Number crunching : FYI : to any researcher that cares, problem with WU 618924673



©2024 University of Washington
https://www.bakerlab.org