Merge pull request #1982 from rapidsai/release/26.04

GPUtester · web-flow · commit 7311ade1d88c · 2026-04-01T19:19:04.000-05:00
Forward-merge release/26.04 into main
diff --git a/c/include/cuvs/neighbors/all_neighbors.h b/c/include/cuvs/neighbors/all_neighbors.h
@@ -1,5 +1,5 @@
 /*
- * SPDX-FileCopyrightText: Copyright (c) 2025, NVIDIA CORPORATION.
+ * SPDX-FileCopyrightText: Copyright (c) 2025-2026, NVIDIA CORPORATION.
  * SPDX-License-Identifier: Apache-2.0
  */
 
@@ -17,7 +17,7 @@ extern "C" {
 #endif
 
 /**
- * @defgroup all_neighbors_c_build All-neighbors C-API build
+ * @defgroup all_neighbors_c_params All-neighbors C-API build parameters
  * @{
  *
  * All-neighbors constructs an approximate k-NN graph for all vectors in a dataset.
@@ -78,6 +78,13 @@ cuvsError_t cuvsAllNeighborsIndexParamsCreate(cuvsAllNeighborsIndexParams_t* ind
  */
 cuvsError_t cuvsAllNeighborsIndexParamsDestroy(cuvsAllNeighborsIndexParams_t index_params);
 
+/** @} */
+
+/**
+ * @defgroup all_neighbors_c_build All-neighbors C-API build
+ * @{
+ */
+
 /**
  * @brief Build an all-neighbors k-NN graph automatically detecting host vs device dataset.
  *
diff --git a/cpp/include/cuvs/cluster/kmeans.hpp b/cpp/include/cuvs/cluster/kmeans.hpp
@@ -1573,15 +1573,14 @@ void cluster_cost(
  * @}
  */
 
+namespace helpers {
 /**
  * @defgroup kmeans_helpers k-means API helpers
  * @{
  */
 
-namespace helpers {
-
 /**
- * Automatically find the optimal value of k using a binary search.
+ * @brief Automatically find the optimal value of k using a binary search.
  * This method maximizes the Calinski-Harabasz Index while minimizing the per-cluster inertia.
  *
  *  @code{.cpp}
@@ -1626,10 +1625,9 @@ void find_k(raft::resources const& handle,
             int kmin    = 1,
             int maxiter = 100,
             float tol   = 1e-3);
-}  // namespace helpers
-
 /**
  * @}
  */
+}  // namespace helpers
 
 }  // namespace  cuvs::cluster::kmeans
diff --git a/cpp/include/cuvs/preprocessing/quantize/binary.hpp b/cpp/include/cuvs/preprocessing/quantize/binary.hpp
@@ -1,5 +1,5 @@
 /*
- * SPDX-FileCopyrightText: Copyright (c) 2024-2025, NVIDIA CORPORATION.
+ * SPDX-FileCopyrightText: Copyright (c) 2024-2026, NVIDIA CORPORATION.
  * SPDX-License-Identifier: Apache-2.0
  */
 
@@ -30,11 +30,10 @@ enum class bit_threshold { zero, mean, sampling_median };
  * @brief quantizer parameters.
  */
 struct params {
+  /** Threshold method for binarization. */
   bit_threshold threshold = bit_threshold::mean;
 
-  /*
-   * specifies the sampling ratio
-   */
+  /** Specifies the sampling ratio. */
   float sampling_ratio = 0.1;
 };
 
@@ -48,8 +47,10 @@ struct params {
  */
 template <typename T>
 struct quantizer {
+  /** Threshold vector used for binarization. */
   raft::device_vector<T, int64_t> threshold;
 
+  /** @brief Construct a quantizer with an empty threshold vector. */
   quantizer(raft::resources const& res) : threshold(raft::make_device_vector<T, int64_t>(res, 0)) {}
 };
 
diff --git a/cpp/include/cuvs/preprocessing/quantize/pq.hpp b/cpp/include/cuvs/preprocessing/quantize/pq.hpp
@@ -84,7 +84,9 @@ struct params {
  */
 template <typename T>
 struct quantizer {
+  /** Parameters used to build this quantizer. */
   params params_quantizer;
+  /** VPQ codebooks produced during training. */
   cuvs::neighbors::vpq_dataset<T, int64_t> vpq_codebooks;
 };
 
diff --git a/cpp/include/cuvs/preprocessing/quantize/scalar.hpp b/cpp/include/cuvs/preprocessing/quantize/scalar.hpp
@@ -1,5 +1,5 @@
 /*
- * SPDX-FileCopyrightText: Copyright (c) 2024-2025, NVIDIA CORPORATION.
+ * SPDX-FileCopyrightText: Copyright (c) 2024-2026, NVIDIA CORPORATION.
  * SPDX-License-Identifier: Apache-2.0
  */
 
@@ -24,9 +24,9 @@ namespace cuvs::preprocessing::quantize::scalar {
  * @brief quantizer parameters.
  */
 struct params {
-  /*
-   * specifies how many outliers at top & bottom will be ignored
-   * needs to be within range of (0, 1]
+  /**
+   * Specifies how many outliers at top & bottom will be ignored.
+   * Needs to be within range of (0, 1].
    */
   float quantile = 0.99;
 };
@@ -42,7 +42,9 @@ struct params {
  */
 template <typename T>
 struct quantizer {
+  /** Minimum value of the quantization range. */
   T min_;
+  /** Maximum value of the quantization range. */
   T max_;
 };
 
diff --git a/docs/source/c_api/neighbors.rst b/docs/source/c_api/neighbors.rst
@@ -9,11 +9,11 @@ Nearest Neighbors
    :maxdepth: 2
    :caption: Contents:
 
+   neighbors_all_neighbors_c.rst
    neighbors_bruteforce_c.rst
-   neighbors_ivf_flat_c.rst
-   neighbors_ivf_pq_c.rst
    neighbors_cagra_c.rst
    neighbors_hnsw_c.rst
+   neighbors_ivf_flat_c.rst
+   neighbors_ivf_pq_c.rst
    neighbors_mg.rst
-   neighbors_all_neighbors_c.rst
    neighbors_vamana_c.rst
diff --git a/docs/source/c_api/neighbors_all_neighbors_c.rst b/docs/source/c_api/neighbors_all_neighbors_c.rst
@@ -9,6 +9,14 @@ The all-neighbors method constructs a k-NN graph for all vectors in a dataset. I
 
 ``#include <cuvs/neighbors/all_neighbors.h>``
 
+Build parameters
+----------------
+
+.. doxygengroup:: all_neighbors_c_params
+    :project: cuvs
+    :members:
+    :content-only:
+
 Build
 -----
 
diff --git a/docs/source/cpp_api/cluster.rst b/docs/source/cpp_api/cluster.rst
@@ -9,6 +9,6 @@ Cluster
    :maxdepth: 2
    :caption: Contents:
 
-   cluster_kmeans.rst
    cluster_agglomerative.rst
+   cluster_kmeans.rst
    cluster_spectral.rst
diff --git a/docs/source/cpp_api/cluster_agglomerative.rst b/docs/source/cpp_api/cluster_agglomerative.rst
@@ -5,8 +5,8 @@ Agglomerative
    :language: c++
    :class: highlight
 
-Params
-------
+Parameters
+----------
 
 ``#include <cuvs/cluster/agglomerative.hpp>``
 
@@ -18,10 +18,10 @@ namespace *cuvs::cluster::agglomerative*
    :content-only:
 
 
-Single-linkage
---------------
+Agglomerative
+-------------
 
-``include <cuvs/cluster/agglomerative.hpp>``
+``#include <cuvs/cluster/agglomerative.hpp>``
 
 namespace *cuvs::cluster::agglomerative*
 
diff --git a/docs/source/cpp_api/cluster_kmeans.rst b/docs/source/cpp_api/cluster_kmeans.rst
@@ -5,8 +5,8 @@ K-Means
    :language: c++
    :class: highlight
 
-Params
-------
+Parameters
+----------
 
 ``#include <cuvs/cluster/kmeans.hpp>``
 
@@ -21,7 +21,7 @@ namespace *cuvs::cluster::kmeans*
 K-means
 -------
 
-``include <cuvs/cluster/kmeans.hpp>``
+``#include <cuvs/cluster/kmeans.hpp>``
 
 namespace *cuvs::cluster::kmeans*
 
@@ -34,7 +34,7 @@ namespace *cuvs::cluster::kmeans*
 K-means Helpers
 ---------------
 
-``include <cuvs/cluster/kmeans.hpp>``
+``#include <cuvs/cluster/kmeans.hpp>``
 
 namespace *cuvs::cluster::kmeans::helpers*
 
diff --git a/docs/source/cpp_api/distance.rst b/docs/source/cpp_api/distance.rst
@@ -22,7 +22,7 @@ namespace *cuvs::distance*
 Pairwise Distances
 ------------------
 
-``include <cuvs/distance/distance.hpp>``
+``#include <cuvs/distance/distance.hpp>``
 
 namespace *cuvs::distance*
 
diff --git a/docs/source/cpp_api/neighbors.rst b/docs/source/cpp_api/neighbors.rst
@@ -18,7 +18,7 @@ Nearest Neighbors
    neighbors_hnsw.rst
    neighbors_ivf_flat.rst
    neighbors_ivf_pq.rst
+   neighbors_mg.rst
    neighbors_nn_descent.rst
    neighbors_refine.rst
-   neighbors_mg.rst
    neighbors_vamana.rst
diff --git a/docs/source/cpp_api/neighbors_all_neighbors.rst b/docs/source/cpp_api/neighbors_all_neighbors.rst
@@ -1,7 +1,7 @@
-All-neighbors
+All-Neighbors
 =============
 
-All-neighbors allows building an approximate all-neighbors knn graph. Given a full dataset, it finds nearest neighbors for all the training vectors in the dataset.
+All-Neighbors allows building an approximate all-neighbors knn graph. Given a full dataset, it finds nearest neighbors for all the training vectors in the dataset.
 
 .. role:: py(code)
    :language: c++
@@ -11,8 +11,8 @@ All-neighbors allows building an approximate all-neighbors knn graph. Given a fu
 
 namespace *cuvs::neighbors::all_neighbors*
 
-All neighbors knn graph build parameters
-----------------------------------------
+Build Parameters
+----------------
 
 .. doxygengroup:: all_neighbors_cpp_params
     :project: cuvs
diff --git a/docs/source/cpp_api/neighbors_mg.rst b/docs/source/cpp_api/neighbors_mg.rst
@@ -1,7 +1,7 @@
-Distributed ANN
-===============
+Multi-GPU Nearest Neighbors
+===========================
 
-The SNMG (single-node multi-GPUs) ANN API provides a set of functions to deploy ANN indexes on multiple GPUs.
+The Multi-GPU (SNMG - single-node multi-GPUs) nearest neighbors API provides a set of functions to deploy ANN indexes across multiple GPUs for improved performance and scalability.
 
 .. role:: py(code)
    :language: c++
diff --git a/docs/source/cpp_api/preprocessing_pca.rst b/docs/source/cpp_api/preprocessing_pca.rst
@@ -11,7 +11,7 @@ Principal Component Analysis (PCA) is a linear dimensionality reduction techniqu
 
 namespace *cuvs::preprocessing::pca*
 
-Params
+Parameters
 ----------
 
 .. doxygenstruct:: cuvs::preprocessing::pca::params
diff --git a/docs/source/cpp_api/preprocessing_quantize.rst b/docs/source/cpp_api/preprocessing_quantize.rst
@@ -8,25 +8,17 @@ This page provides C++ class references for the publicly-exposed elements of the
    :language: c++
    :class: highlight
 
-Scalar
-------
-
-``#include <cuvs/preprocessing/quantize/scalar.hpp>``
-
-namespace *cuvs::preprocessing::quantize::scalar*
-
-.. doxygengroup:: scalar
-   :project: cuvs
-
-Binary
-------
+Binary Quantizer
+----------------
 
 ``#include <cuvs/preprocessing/quantize/binary.hpp>``
 
 namespace *cuvs::preprocessing::quantize::binary*
 
 .. doxygengroup:: binary
    :project: cuvs
+   :members:
+   :content-only:
 
 Product Quantizer
 -----------------
@@ -37,3 +29,17 @@ namespace *cuvs::preprocessing::quantize::pq*
 
 .. doxygengroup:: pq
    :project: cuvs
+   :members:
+   :content-only:
+
+Scalar Quantizer
+----------------
+
+``#include <cuvs/preprocessing/quantize/scalar.hpp>``
+
+namespace *cuvs::preprocessing::quantize::scalar*
+
+.. doxygengroup:: scalar
+   :project: cuvs
+   :members:
+   :content-only:
diff --git a/docs/source/cpp_api/preprocessing_spectral_embedding.rst b/docs/source/cpp_api/preprocessing_spectral_embedding.rst
@@ -31,8 +31,8 @@ namespace *cuvs::preprocessing::spectral_embedding*
    :project: cuvs
    :members:
 
-Functions
----------
+Spectral Embedding
+------------------
 
 ``#include <cuvs/preprocessing/spectral_embedding.hpp>``
 
diff --git a/docs/source/getting_started.rst b/docs/source/getting_started.rst
@@ -86,8 +86,9 @@ Blogs
 
 We frequently publish blogs on GPU-enabled vector search, which can provide great deep dives into various important topics and breakthroughs:
 
+#. `See all cuVS blogs <https://developer.nvidia.com/blog/recent-posts/?products=cuVS>`_
 #. `Accelerated Vector Search: Approximating with cuVS IVF-Flat <https://developer.nvidia.com/blog/accelerated-vector-search-approximating-with-rapids-raft-ivf-flat/>`_
-#. `Accelerating Vector Search with cuVS IVF-PQ <https://developer.nvidia.com/blog/accelerating-vector-search-rapids-cuvs-ivf-pq-deep-dive-part-1/>`_
+#. Accelerating Vector Search with cuVS IVF-PQ (`Part 1 <https://developer.nvidia.com/blog/accelerating-vector-search-rapids-cuvs-ivf-pq-deep-dive-part-1/>`_, `Part 2 <https://developer.nvidia.com/blog/accelerating-vector-search-nvidia-cuvs-ivf-pq-performance-tuning-part-2/>`_)
 
 Research
 --------
@@ -99,6 +100,7 @@ For the interested reader, many of the accelerated implementations in cuVS are a
 #. `Fast K-NN Graph Construction by GPU Based NN-Descent <https://dl.acm.org/doi/abs/10.1145/3459637.3482344?casa_token=O_nan1B1F5cAAAAA:QHWDEhh0wmd6UUTLY9_Gv6c3XI-5DXM9mXVaUXOYeStlpxTPmV3nKvABRfoivZAaQ3n8FWyrkWw>`_
 #. `cuSLINK: Single-linkage Agglomerative Clustering on the GPU <https://arxiv.org/abs/2306.16354>`_
 #. `GPU Semiring Primitives for Sparse Neighborhood Methods <https://arxiv.org/abs/2104.06357>`_
+#. `VecFlow: A High-Performance Vector Data Management System for Filtered-Search on GPUs <https://arxiv.org/abs/2506.00812>`_
 
 
 Get involved
diff --git a/docs/source/index.rst b/docs/source/index.rst
@@ -63,7 +63,7 @@ There are several benefits to using cuVS and GPUs for vector search, including
 6. Multiple language support
 7. Building blocks for composing new or accelerating existing algorithms
 
-In addition to the items above, cuVS shoulders the responsibility of keeping non-trivial accelerated code up to date as new NVIDIA architectures and CUDA versions are released. This provides a deslightful development experimence, guaranteeing that any libraries, databases, or applications built on top of it will always be receiving the best performance and scale.
+In addition to the items above, cuVS shoulders the responsibility of keeping non-trivial accelerated code up to date as new NVIDIA architectures and CUDA versions are released. This provides a delightful development experience, guaranteeing that any libraries, databases, or applications built on top of it will always be receiving the best performance and scale.
 
 cuVS Technology Stack
 #####################
diff --git a/docs/source/python_api/neighbors.rst b/docs/source/python_api/neighbors.rst
@@ -5,26 +5,15 @@ Nearest Neighbors
    :language: python
    :class: highlight
 
-Single-GPU Algorithms
-#####################
-
 .. toctree::
    :maxdepth: 2
-   :caption: Single-GPU ANN Algorithms:
+   :caption: Contents:
 
+   neighbors_all_neighbors.rst
    neighbors_brute_force.rst
    neighbors_cagra.rst
    neighbors_hnsw.rst
    neighbors_ivf_flat.rst
    neighbors_ivf_pq.rst
-   neighbors_nn_decent.rst
-
-Multi-GPU Algorithms
-####################
-
-.. toctree::
-   :maxdepth: 2
-   :caption: Multi-GPU Distributed ANN:
-
    neighbors_multi_gpu.rst
-   neighbors_all_neighbors.rst
+   neighbors_nn_decent.rst
diff --git a/docs/source/python_api/neighbors_all_neighbors.rst b/docs/source/python_api/neighbors_all_neighbors.rst
diff --git a/docs/source/python_api/neighbors_multi_gpu.rst b/docs/source/python_api/neighbors_multi_gpu.rst
