From 0967e8a974fbd9b1a6c5942d5daab12bb44a0c0e Mon Sep 17 00:00:00 2001
From: Arne Juul <arnej@verizonmedia.com>
Date: Tue, 31 Mar 2020 13:04:54 +0000
Subject: add some simple statistics when exploring state

---
 searchlib/src/vespa/searchlib/tensor/hnsw_graph.cpp | 19 +++++++++++++++++++
 searchlib/src/vespa/searchlib/tensor/hnsw_graph.h   |  2 ++
 searchlib/src/vespa/searchlib/tensor/hnsw_index.cpp |  7 +++++++
 3 files changed, 28 insertions(+)

(limited to 'searchlib')

diff --git a/searchlib/src/vespa/searchlib/tensor/hnsw_graph.cpp b/searchlib/src/vespa/searchlib/tensor/hnsw_graph.cpp
index db8d1067980..6f902a30861 100644
--- a/searchlib/src/vespa/searchlib/tensor/hnsw_graph.cpp
+++ b/searchlib/src/vespa/searchlib/tensor/hnsw_graph.cpp
@@ -50,4 +50,23 @@ HnswGraph::set_link_array(uint32_t docid, uint32_t level, const LinkArrayRef& ne
     links.remove(old_links_ref);
 }
 
+std::vector<uint32_t>
+HnswGraph::level_histogram() const
+{
+    std::vector<uint32_t> result;
+    size_t num_nodes = node_refs.size();
+    for (size_t i = 0; i < num_nodes; ++i) {
+        uint32_t levels = 0;
+        auto node_ref = node_refs[i].load_acquire();
+        if (node_ref.valid()) {
+            levels = nodes.get(node_ref).size();
+        }
+        while (result.size() <= levels) {
+            result.push_back(0);
+        }
+        ++result[levels];
+    }
+    return result;
+}
+
 } // namespace
diff --git a/searchlib/src/vespa/searchlib/tensor/hnsw_graph.h b/searchlib/src/vespa/searchlib/tensor/hnsw_graph.h
index 64892d06f09..233b9087af7 100644
--- a/searchlib/src/vespa/searchlib/tensor/hnsw_graph.h
+++ b/searchlib/src/vespa/searchlib/tensor/hnsw_graph.h
@@ -69,6 +69,8 @@ struct HnswGraph {
     }
 
     size_t size() const { return node_refs.size(); }
+
+    std::vector<uint32_t> level_histogram() const;
 };
 
 }
diff --git a/searchlib/src/vespa/searchlib/tensor/hnsw_index.cpp b/searchlib/src/vespa/searchlib/tensor/hnsw_index.cpp
index de6daba650c..10be90a7f30 100644
--- a/searchlib/src/vespa/searchlib/tensor/hnsw_index.cpp
+++ b/searchlib/src/vespa/searchlib/tensor/hnsw_index.cpp
@@ -381,6 +381,13 @@ HnswIndex::get_state(const vespalib::slime::Inserter& inserter) const
 {
     auto& object = inserter.insertObject();
     StateExplorerUtils::memory_usage_to_slime(memory_usage(), object.setObject("memory_usage"));
+    object.setLong("nodes", _graph.size());
+    auto& level_histogram = object.setArray("level_histogram");
+    for (uint32_t hist_val : _graph.level_histogram()) {
+        level_histogram.addLong(hist_val);
+    }
+    object.setLong("entry_docid", _graph.entry_docid);
+    object.setLong("entry_level", _graph.entry_level);
 }
 
 std::unique_ptr<NearestNeighborIndexSaver>
-- 
cgit v1.2.3