This proposal leaves some algorithms of <memory>
, <algorithm>
, and <numeric>
unchanged
because they either 1. have no parallelism, 2. have explicitly sequential semantics, 3. have an uncertain parallel implementation and for which we feel parallelization would be low priority.
For example, instead of parallelizing the standard heap algorithms, a better strategy may be to provide concurrent priority queues as a data structure as some have suggested.
- No parallelism
addressof
-
No parallelism
binary_search
equal_range
iter_swap
lower_bound
min
max
minmax
upper_bound
-
Uncertain / low priority
is_permutation
push_heap
pop_heap
make_heap
sort_heap
is_heap
is_heap_until
next_permutation
prev_permutation
-
Explicitly sequential
copy_backward
move_backward
random_shuffle
shuffle
- Explicitly sequential
accumulate
partial_sum
iota