Reference based motion correction error ====== Job process terminated abnormally

Motion Correction
41

I have some dataset that ran reference based motioncor without an issue and some fails.

Successful run 1:
EER format, 129k particles -22800 movies, 600pix box binned to 300pix. default parameters
Successful run 2:
EER format, 554k particles - 3230 movies, 380pix box size. default parameters
Successful run 3:
F4i, Tiff format, 560k particles -2929 movies, 380pix box size, -Target number of particles 8000, 1 GPU, GPU threshold 40GB, in memory cache 240GB

Failed run:
EER format, 660k particles, 3463 movies, 380pix box, default parameters - job hangs randomly during hyperparameter tuning run 1.

F4i Tiff format (same set as successful run 3) 560k particles -2929 movies, 380pix box size, default parameters hang randomly during hyperparameter tuning run 1. When reducing particle number to 8000, it would run for a few iteration, than hang.

1 Like
42

Iā€™m confused because it doesnā€™t seems to be related to particle numbers, as I had successful runs using 554k particles with the same dimensions.

43

I worry that there may be some corrupted images that are causing this issue.
But I donā€™t have evidence of that. So, letā€™s try to get the developerā€™s attention to this matter. :slight_smile:

Kind regards,
Dmitry

44

Just as an FYI, the useless TIFFReadDirectory: Unknown field with tag 65002 (0xfdea) encountered messages will no longer appear as of CryoSPARC v4.5, released today.

ā€“ Harris

3 Likes