Does this fit current hardware, and should it be for both sequences or just the primary parallel axis?
It may be good to revisit if this cutoff is still a reasonable size for current hardware, and if it should be based upon the "primary" axis of parallel work or if it should be based upon the sum of both sequences. However, I would propose to do this separately from this PR, as it should be addressed across all usages of __set_algo_cut_off, which spans all algorithms for both iterator and ranges APIs.
Originally posted by @danhoeflinger in #2583
Does this fit current hardware, and should it be for both sequences or just the primary parallel axis?
Originally posted by @danhoeflinger in #2583