You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copying or moving data has (at least) the following variants:
Element size in bytes
1.1 size is a power of 2
1.2 size is a natively-supported power of 2 (1, 2, 4, 8)
1.3 size is a multiple of the physical write quantum
Destination alignment:
2.1 w.r.t. the element size
2.2 w.r.t. natively-supported sizes (2, 4, 8)
Source alignment:
3.1 w.r.t. the element size
3.2 w.r.t. natively-supported sizes (2, 4, 8)
Trivial copiability/movability
Alignment effects of the number of elements
5.1 in itself
5.2 relative to the beginning of the source
5.3 relative to the beginning of the destination
6.1 Number of elements % warp size
6.2. Number of elements % block size
Participation cohesion in the copying: Full grid, full block, full warp, no constraint.
We currently ignore most of these, in some cases to detrimental effect. We are particularly not mindful enough of alignment and its effects.
This also has effects on other functions in cuda_kat, e.g. append_to_global_memory(), which may be forced to opt for a detail::naive_copy() because of issues with copy() or copy_n().
The text was updated successfully, but these errors were encountered:
Copying or moving data has (at least) the following variants:
1.1 size is a power of 2
1.2 size is a natively-supported power of 2 (1, 2, 4, 8)
1.3 size is a multiple of the physical write quantum
2.1 w.r.t. the element size
2.2 w.r.t. natively-supported sizes (2, 4, 8)
3.1 w.r.t. the element size
3.2 w.r.t. natively-supported sizes (2, 4, 8)
5.1 in itself
5.2 relative to the beginning of the source
5.3 relative to the beginning of the destination
6.2. Number of elements % block size
We currently ignore most of these, in some cases to detrimental effect. We are particularly not mindful enough of alignment and its effects.
This also has effects on other functions in cuda_kat, e.g.
append_to_global_memory()
, which may be forced to opt for adetail::naive_copy()
because of issues withcopy()
orcopy_n()
.The text was updated successfully, but these errors were encountered: