Open jrhemstad opened 1 month ago
Priority-wise, I think we want to focus on the algorithms that currently do not support large number of items yet (see overview). That is,
device_select.cuh
, device_partition.cuh
, device_scan.cuh
, device_segmented_sort.cuh
, device_segmented_radix_sort.cuh
, device_run_length_encode.cuh
In order to make an informed decision about the offset type solution, we would like to have a complete understanding of the impact on each CUB algorithm. Therefore, for each algorithm, we'd like a table that summarizes the following: