Merge pull request #24877 from vespa-engine/toregge/move-inner-class-…

…hnswindex-config-to-hnswindexconfig Move inner class HnswIndex::Config to HnswIndexConfig.
vespa-engine · Nov 15, 2022 · 1581058 · 1581058
2 parents 9a35cc0 + 7310a44
commit 1581058
Show file tree

Hide file tree

Showing 6 changed files with 51 additions and 38 deletions.
diff --git a/searchlib/src/tests/tensor/hnsw_index/hnsw_index_test.cpp b/searchlib/src/tests/tensor/hnsw_index/hnsw_index_test.cpp
@@ -103,7 +103,7 @@ class HnswIndexTest : public ::testing::Test {
         level_generator = generator.get();
         index = std::make_unique<HnswIndex>(vectors, std::make_unique<SquaredEuclideanDistance>(vespalib::eval::CellType::FLOAT),
                                             std::move(generator),
-                                            HnswIndex::Config(5, 2, 10, 0, heuristic_select_neighbors));
+                                            HnswIndexConfig(5, 2, 10, 0, heuristic_select_neighbors));
     }
     void add_document(uint32_t docid, uint32_t max_level = 0) {
         level_generator->level = max_level;

diff --git a/searchlib/src/tests/tensor/hnsw_index/stress_hnsw_mt.cpp b/searchlib/src/tests/tensor/hnsw_index/stress_hnsw_mt.cpp
@@ -252,7 +252,7 @@ class Stressor : public ::testing::Test {
         uint32_t m = 16;
         index = std::make_unique<HnswIndex>(vectors, std::make_unique<FloatSqEuclideanDistance>(),
                                             std::make_unique<InvLogLevelGenerator>(m),
-                                            HnswIndex::Config(2*m, m, 200, 10, true));
+                                            HnswIndexConfig(2*m, m, 200, 10, true));
     }
     size_t get_rnd(size_t size) {
         return rng.nextUniform() * size;

diff --git a/searchlib/src/vespa/searchlib/tensor/default_nearest_neighbor_index_factory.cpp b/searchlib/src/vespa/searchlib/tensor/default_nearest_neighbor_index_factory.cpp
@@ -33,11 +33,11 @@ DefaultNearestNeighborIndexFactory::make(const DocVectorAccess& vectors,
 {
     (void) vector_size;
     uint32_t m = params.max_links_per_node();
-    HnswIndex::Config cfg(m * 2,
-                          m,
-                          params.neighbors_to_explore_at_insert(),
-                          10000,
-                          true);
+    HnswIndexConfig cfg(m * 2,
+                        m,
+                        params.neighbors_to_explore_at_insert(),
+                        10000,
+                        true);
     return std::make_unique<HnswIndex>(vectors,
                                        make_distance_function(params.distance_metric(), cell_type),
                                        make_random_level_generator(m),

diff --git a/searchlib/src/vespa/searchlib/tensor/hnsw_index.cpp b/searchlib/src/vespa/searchlib/tensor/hnsw_index.cpp
@@ -326,7 +326,7 @@ HnswIndex::search_layer(const TypedCells& input, uint32_t neighbors_to_find,
 }
 
 HnswIndex::HnswIndex(const DocVectorAccess& vectors, DistanceFunction::UP distance_func,
-                     RandomLevelGenerator::UP level_generator, const Config& cfg)
+                     RandomLevelGenerator::UP level_generator, const HnswIndexConfig& cfg)
     : _graph(),
       _vectors(vectors),
       _distance_func(std::move(distance_func)),

diff --git a/searchlib/src/vespa/searchlib/tensor/hnsw_index.h b/searchlib/src/vespa/searchlib/tensor/hnsw_index.h
@@ -2,6 +2,7 @@
 
 #pragma once
 
+#include "hnsw_index_config.h"
 #include "distance_function.h"
 #include "doc_vector_access.h"
 #include "hnsw_identity_mapping.h"
@@ -36,33 +37,6 @@ namespace search::tensor {
  */
 class HnswIndex : public NearestNeighborIndex {
 public:
-    class Config {
-    private:
-        uint32_t _max_links_at_level_0;
-        uint32_t _max_links_on_inserts;
-        uint32_t _neighbors_to_explore_at_construction;
-        uint32_t _min_size_before_two_phase;
-        bool _heuristic_select_neighbors;
-
-    public:
-        Config(uint32_t max_links_at_level_0_in,
-               uint32_t max_links_on_inserts_in,
-               uint32_t neighbors_to_explore_at_construction_in,
-               uint32_t min_size_before_two_phase_in,
-               bool heuristic_select_neighbors_in)
-            : _max_links_at_level_0(max_links_at_level_0_in),
-              _max_links_on_inserts(max_links_on_inserts_in),
-              _neighbors_to_explore_at_construction(neighbors_to_explore_at_construction_in),
-              _min_size_before_two_phase(min_size_before_two_phase_in),
-              _heuristic_select_neighbors(heuristic_select_neighbors_in)
-        {}
-        uint32_t max_links_at_level_0() const { return _max_links_at_level_0; }
-        uint32_t max_links_on_inserts() const { return _max_links_on_inserts; }
-        uint32_t neighbors_to_explore_at_construction() const { return _neighbors_to_explore_at_construction; }
-        uint32_t min_size_before_two_phase() const { return _min_size_before_two_phase; }
-        bool heuristic_select_neighbors() const { return _heuristic_select_neighbors; }
-    };
-
     class HnswIndexCompactionSpec {
         CompactionSpec _level_arrays;
         CompactionSpec _link_arrays;
@@ -109,7 +83,7 @@ class HnswIndex : public NearestNeighborIndex {
     DistanceFunction::UP _distance_func;
     RandomLevelGenerator::UP _level_generator;
     IdMapping _id_mapping; // mapping from docid to nodeid vector
-    Config _cfg;
+    HnswIndexConfig _cfg;
     HnswIndexCompactionSpec _compaction_spec;
 
     uint32_t max_links_for_level(uint32_t level) const;
@@ -212,10 +186,10 @@ class HnswIndex : public NearestNeighborIndex {
     void internal_complete_add_node(uint32_t nodeid, uint32_t docid, uint32_t subspace, PreparedAddNode &prepared_node);
 public:
     HnswIndex(const DocVectorAccess& vectors, DistanceFunction::UP distance_func,
-              RandomLevelGenerator::UP level_generator, const Config& cfg);
+              RandomLevelGenerator::UP level_generator, const HnswIndexConfig& cfg);
     ~HnswIndex() override;
 
-    const Config& config() const { return _cfg; }
+    const HnswIndexConfig& config() const { return _cfg; }
 
     // Implements NearestNeighborIndex
     void add_document(uint32_t docid) override;

diff --git a/searchlib/src/vespa/searchlib/tensor/hnsw_index_config.h b/searchlib/src/vespa/searchlib/tensor/hnsw_index_config.h
@@ -0,0 +1,39 @@
+// Copyright Yahoo. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
+
+#pragma once
+
+#include <cstdint>
+
+namespace search::tensor {
+
+/*
+ * Class containing config for HnswIndex.
+ */
+class HnswIndexConfig {
+private:
+    uint32_t _max_links_at_level_0;
+    uint32_t _max_links_on_inserts;
+    uint32_t _neighbors_to_explore_at_construction;
+    uint32_t _min_size_before_two_phase;
+    bool     _heuristic_select_neighbors;
+
+public:
+    HnswIndexConfig(uint32_t max_links_at_level_0_in,
+                    uint32_t max_links_on_inserts_in,
+                    uint32_t neighbors_to_explore_at_construction_in,
+                    uint32_t min_size_before_two_phase_in,
+                    bool heuristic_select_neighbors_in)
+        : _max_links_at_level_0(max_links_at_level_0_in),
+          _max_links_on_inserts(max_links_on_inserts_in),
+          _neighbors_to_explore_at_construction(neighbors_to_explore_at_construction_in),
+          _min_size_before_two_phase(min_size_before_two_phase_in),
+          _heuristic_select_neighbors(heuristic_select_neighbors_in)
+    {}
+    uint32_t max_links_at_level_0() const { return _max_links_at_level_0; }
+    uint32_t max_links_on_inserts() const { return _max_links_on_inserts; }
+    uint32_t neighbors_to_explore_at_construction() const { return _neighbors_to_explore_at_construction; }
+    uint32_t min_size_before_two_phase() const { return _min_size_before_two_phase; }
+    bool heuristic_select_neighbors() const { return _heuristic_select_neighbors; }
+};
+
+}