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.

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

Jump to bottom

Fix Louvain hang in multi-GPU testing #2028

Merged
merged 1 commit into from
Jan 21, 2022
Merged

Fix Louvain hang in multi-GPU testing #2028

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
22 changes: 12 additions & 10 deletions cpp/include/cugraph/prims/copy_v_transform_reduce_key_aggregated_out_nbr.cuh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -336,11 +336,10 @@ void copy_v_transform_reduce_key_aggregated_out_nbr(
handle.get_stream());

if (matrix_partition.get_number_of_edges() > 0) {
auto segment_offsets = graph_view.get_local_adj_matrix_partition_segment_offsets(i);

detail::decompress_matrix_partition_to_fill_edgelist_majors(
handle,
matrix_partition,
tmp_major_vertices.data(),
graph_view.get_local_adj_matrix_partition_segment_offsets(i));
handle, matrix_partition, tmp_major_vertices.data(), segment_offsets);

auto minor_key_first = thrust::make_transform_iterator(
matrix_partition.get_indices(),
Expand All @@ -351,12 +350,15 @@ void copy_v_transform_reduce_key_aggregated_out_nbr(
// to limit memory footprint ((1 << 20) is a tuning parameter)
auto approx_edges_to_sort_per_iteration =
static_cast<size_t>(handle.get_device_properties().multiProcessorCount) * (1 << 20);
auto [h_vertex_offsets, h_edge_offsets] =
detail::compute_offset_aligned_edge_chunks(handle,
matrix_partition.get_offsets(),
matrix_partition.get_major_size(),
matrix_partition.get_number_of_edges(),
approx_edges_to_sort_per_iteration);
auto [h_vertex_offsets, h_edge_offsets] = detail::compute_offset_aligned_edge_chunks(
handle,
matrix_partition.get_offsets(),
matrix_partition.get_dcs_nzd_vertices()
? (*segment_offsets)[detail::num_sparse_segments_per_vertex_partition] +
*(matrix_partition.get_dcs_nzd_vertex_count())
: matrix_partition.get_major_size(),
matrix_partition.get_number_of_edges(),
approx_edges_to_sort_per_iteration);
auto num_chunks = h_vertex_offsets.size() - 1;

size_t max_chunk_size{0};
Expand Down