Skip to content

Convert double to float data type in C code to reduce memory usage (issue #45)

Yang, Diyu requested to merge double_to_float into master

Created by: dyang37

This PR is to handle issue #45 (closed). Recently I've been running large scale CT experiments. I constantly run into out of memory issues on Brown (maximum allowed memory is 91 GB). By converting double to float in C and Cython interface, we are going to be able to save half of the memory space. Tested with 3D shepp Logan demo. Results are visually the same as before. Note that you will need to clean up cached system matrix before testing, otherwise there will be a seg fault.

Merge request reports