Skip to content

Skip stride ranking when the memory estimate fits #92

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
lkdvos merged 1 commit into from
Oct 7, 2026
Merged

Skip stride ranking when the memory estimate fits #92

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
18 changes: 11 additions & 7 deletions src/mapreduce.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -200,11 +200,14 @@ function _mapreduce_block!(
f::F, op::OP, initop::I, dims, strides, offsets, costs, arrays
) where {F, OP, I}
bytestrides = map((s, stride) -> s .* stride, sizeof.(eltype.(arrays)), strides)
strideorders = map(indexorder, strides)
blocks = _computeblocks(dims, costs, bytestrides, strideorders)

# t = @elapsed _computeblocks(dims, costs, bytestrides, strideorders)
# println("_computeblocks time: $t")
memoryregion = totalmemoryregion(dims, bytestrides)
# When the estimated footprint fits, the full dimensions are already optimal.
blocks = if memoryregion <= BLOCKMEMORYSIZE
dims
else
strideorders = map(indexorder, strides)
_computeblocks(dims, costs, bytestrides, strideorders, BLOCKMEMORYSIZE, memoryregion)
end

if get_num_threads() == 1 || prod(dims) <= MINTHREADLENGTH
_mapreduce_kernel!(f, op, initop, dims, blocks, arrays, strides, offsets)
Expand Down Expand Up @@ -595,9 +598,10 @@ function _computeblocks(
dims::NTuple{N, Int}, costs::NTuple{N, Int},
bytestrides::Tuple{Vararg{NTuple{N, Int}}},
strideorders::Tuple{Vararg{NTuple{N, Int}}},
blocksize::Int = BLOCKMEMORYSIZE
blocksize::Int = BLOCKMEMORYSIZE,
memoryregion::Int = totalmemoryregion(dims, bytestrides)
) where {N}
if totalmemoryregion(dims, bytestrides) <= blocksize
if memoryregion <= blocksize
return dims
end
minstrideorder = minimum(minimum.(strideorders))
Expand Down
Loading