Closed TomMaullin closed 3 years ago
Following PR #45 , this now exists as the backdoor option diskMem
. When the disk memory is too low, it allows splitting of the analysis mask into chunks with each separate chunk being run one after the other, via a BLMM analysis, in serial. This issue will remain open until it is decided whether to document this option or delete it.
Following PR #49 , this issue is obsolete as the batch stage accounts for the later voxelwise batching stages by outputting to several files, one for each of the following voxelwise batches.
There should be an extra big mode which allows for parallelization over voxels even at the batch stage. This is for the case where the transfer cost is too high.