-
Notifications
You must be signed in to change notification settings - Fork 1
Open
Description
Currently the reconstruction program outputs multiple individual files, one for each frame. This is a inefficient way of doing things, It's rough on the filesystem especially as a single scan can produce many thousands of files.
This will require
- Modifying the C and CUDA implementations to write out to a single file. Might need to change memory management too to support that, especially in a multiprocessing setting.
- Modifying the peaksearch program to read through read an individual file. Jon suggested keeping the single-layer processing capability and to pass in an index as an additional parameter to keep process-level parallelization. A more robust solution would involve passing in a range but would be a significantly larger refactor.
Metadata
Metadata
Assignees
Labels
No labels