Skip to content

Convert pos_diff to NHS eltype if coordinates have a different eltype #132

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
svchb merged 7 commits into from
Nov 17, 2025
Merged

Convert pos_diff to NHS eltype if coordinates have a different eltype #132

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
09e2eaf
Convert `pos_diff` to NHS eltype if coordinates have a different eltype
efaulhaber Oct 15, 2025
4d57002
Merge branch 'main' into ef/coordinates-eltype
svchb Oct 17, 2025
99559cc
Merge branch 'main' into ef/coordinates-eltype
efaulhaber Oct 20, 2025
d4f1ca8
Convert `pos_diff` for trivial and precomputed NHS
efaulhaber Oct 30, 2025
4d7774b
Simplify
efaulhaber Oct 31, 2025
21adb65
Merge branch 'ef/coordinates-eltype' of github.com:trixi-framework/Tr…
efaulhaber Oct 31, 2025
f2f33be
Merge branch 'main' into ef/coordinates-eltype
svchb Nov 17, 2025
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
3 changes: 2 additions & 1 deletion src/neighborhood_search.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,6 +1,7 @@
abstract type AbstractNeighborhoodSearch end

@inline search_radius(search::AbstractNeighborhoodSearch) = search.search_radius
@inline Base.eltype(search::AbstractNeighborhoodSearch) = eltype(search_radius(search))

"""
requires_update(search::AbstractNeighborhoodSearch)
Expand Down Expand Up @@ -220,7 +221,7 @@ end
neighbor_coords = extract_svector(neighbor_system_coords,
Val(ndims(neighborhood_search)), neighbor)

pos_diff = point_coords - neighbor_coords
pos_diff = convert.(eltype(neighborhood_search), point_coords - neighbor_coords)
distance2 = dot(pos_diff, pos_diff)

pos_diff,
Expand Down
2 changes: 1 addition & 1 deletion src/nhs_grid.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -523,7 +523,7 @@ end
neighbor_coords = extract_svector(neighbor_system_coords,
Val(ndims(neighborhood_search)), neighbor)

pos_diff = point_coords - neighbor_coords
pos_diff = convert.(eltype(neighborhood_search), point_coords - neighbor_coords)
distance2 = dot(pos_diff, pos_diff)

pos_diff,
Expand Down
15 changes: 9 additions & 6 deletions src/nhs_precomputed.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -110,12 +110,15 @@ end
for neighbor_ in eachindex(neighbors)
neighbor = @inbounds neighbors[neighbor_]

# Making the following `@inbounds` yields a ~2% speedup on an NVIDIA H100.
# But we don't know if `neighbor` (extracted from the cell list) is in bounds.
neighbor_coords = extract_svector(neighbor_system_coords,
Val(ndims(neighborhood_search)), neighbor)

pos_diff = point_coords - neighbor_coords
# Making this `@inbounds` is not perfectly safe because
# `neighbor` (extracted from the neighbor list) is only guaranteed to be in bounds
# if the neighbor lists were constructed correctly and have not been corrupted.
# However, adding this `@inbounds` yields a ~20% speedup for TLSPH on GPUs (A4500).
neighbor_coords = @inbounds extract_svector(neighbor_system_coords,
Val(ndims(neighborhood_search)),
neighbor)

pos_diff = convert.(eltype(neighborhood_search), point_coords - neighbor_coords)
distance2 = dot(pos_diff, pos_diff)

pos_diff,
Expand Down