Skip to content
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.

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

Eliminate literals passed to device_uvector::set_element_async #367

Merged
merged 2 commits into from
Mar 18, 2021
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
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 cpp/src/join/detail/traversal.cuh
Original file line number Diff line number Diff line change
Expand Up @@ -72,7 +72,8 @@ descend_quadtree(LengthsIter counts,
thrust::inclusive_scan(
rmm::exec_policy(stream), parent_counts, parent_counts + num_quads, parent_offsets.begin() + 1);

parent_offsets.set_element_async(0, 0, stream);
uint32_t init{0};
parent_offsets.set_element_async(0, init, stream);
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Why eliminate one 0 literal and not the other?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Great question. Only the value parameter (second parameter) is pass-by-reference, so that's the only one we are disallowing to be a literal. The first parameter, the index, is pass-by-value so it's perfectly safe to use a literal since it will be copied rather than referenced, and so we don't have to worry about the lifetime of the temporary inside set_element_async.

Copy link
Contributor

@thomcom thomcom Mar 18, 2021

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

That answers my question, yes. But why have one argument be pass by reference and one pass by value? They're both primitives. The 0 is only going to be used by the callee, and the init is going to be passed around and modified by others but has to be initialized here? Thanks!


auto num_children = parent_offsets.back_element(stream); // synchronizes stream

Expand Down
3 changes: 2 additions & 1 deletion cpp/src/join/quadtree_point_in_polygon.cu
Original file line number Diff line number Diff line change
Expand Up @@ -134,7 +134,8 @@ struct compute_quadtree_point_in_polygon {
local_point_offsets.begin() + 1);

// Ensure local point offsets starts at 0
local_point_offsets.set_element_async(0, 0, stream);
uint32_t init{0};
local_point_offsets.set_element_async(0, init, stream);

// The last element is the total number of points to test against any polygon.
auto num_total_points = local_point_offsets.back_element(stream);
Expand Down