-
Notifications
You must be signed in to change notification settings - Fork 71
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
Reduce cagra build binary size #334
base: branch-24.10
Are you sure you want to change the base?
Conversation
Currently we have around 13 MiB reduction in |
@@ -357,8 +353,7 @@ void build_knn_graph( | |||
raft::host_matrix_view<IdxT, int64_t, raft::row_major> knn_graph, | |||
cuvs::neighbors::nn_descent::index_params build_params) | |||
{ | |||
auto nn_descent_idx = cuvs::neighbors::nn_descent::index<IdxT>(res, knn_graph); | |||
cuvs::neighbors::nn_descent::build<DataT, IdxT>(res, build_params, dataset, nn_descent_idx); | |||
auto nn_descent_idx = cuvs::neighbors::nn_descent::build(res, build_params, dataset); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This is relevant: #264 (comment)
Most of these changes were already contained in #264. We keep this open to test further improvements in build binary size. |
During
cagra::build
we use eitherivf_pq
ornn_descent
. Until now we used these algorithm through thedetail
namespace, which resulted in recompilation. This PR changes to the public interface ofivf_pq
andnn_descent
to avoid recompilation.