Dear All,
I have a parfor loop with n iterations (n between 20 and 200), and using input matrices depending n^2 on the size of n. I use matlab R2013b and run the parfor on 12 workers on my laptop, no other computers involved. With higher n (~n>100), I invariably get this error
"Warning: A worker aborted during execution of the parfor loop. The parfor loop will now run again on the remaining workers... The client lost connection to lab 12. This might be due to network problems, or the interactive communicating job might have errored."
I am not sure of the correlation between size of n and the error, as the processing time through the loop also increases with n, but ... here I am. Is this a known issue, and how to work around it? Decrease the no of workers? Thank you, as always,
Octavian

댓글 수: 4

Edric Ellis
Edric Ellis 2015년 4월 8일
Unfortunately this error almost always means that a worker simply crashed. Do you have any crash dump files around that might shed some light?
Octavian
Octavian 2015년 4월 8일
Not sure, where I would find such files? I am ready to run the routine again to get the error and those files if necessary. Thank you, please advise,
Octavian
Edric Ellis
Edric Ellis 2015년 4월 9일
This answer describes where the files should be located - there may well already be some there.
RONG YU
RONG YU 2019년 2월 7일
편집: RONG YU 2019년 2월 7일
we got the same issue again and again when using parfor reading many hdf files. It turns out we had one hdf file corrupted and "try catch" could not avoid the issue.

댓글을 달려면 로그인하십시오.

답변 (0개)

카테고리

도움말 센터File Exchange에서 Parallel Computing Fundamentals에 대해 자세히 알아보기

질문:

2015년 4월 7일

편집:

2019년 2월 7일

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by