feat: refine HNSW greedy search by adjusting history buffer capacity, increasing max iterations, and improving early termination logic.

Author: shisoft

src/apps/hnsw/partition/search.rs

@@ -645,13 +645,14 @@ impl HnswOnlinePartition {
 
         //println!("Starting search with vertex: {:?}", starting_vertices);
         // CRITICAL: Always start with a fresh visited set for each new search
+        let history_capacity = std::cmp::max(1, std::cmp::max(ef, k)).next_power_of_two();
         let partition_search = PartitionSearch {
             schema,
             field_id,
             metadata: SearchMetadata {
                 level: max_level,
                 visited: CompactVisitedSet::with_capacity(ef * 10),
-                history: RingBuffer::new(k.next_power_of_two()),
+                history: RingBuffer::new(history_capacity),
                 remote_frontiers: HashSet::with_hasher(IdBuildHasher),
                 last_distance: Distance::INFINITY,
                 level_entries: AHashMap::default(),
@@ -664,7 +665,7 @@ impl HnswOnlinePartition {
             frontier: starting_vertices,
             server_id: self.server_id,
             iteration: 0,
-            history: RingBuffer::new(k.next_power_of_two()),
+            history: RingBuffer::new(history_capacity),
             last_distance: Distance::INFINITY,
             metric,
             max_level,
@@ -1948,7 +1949,7 @@ impl HnswOnlinePartition {
         const VERTICES_PER_ITERATION: usize = 4;
 
         // Maximum iterations (safety limit, usually converges much faster)
-        let max_iterations = 50;
+        let max_iterations = 100;
         let mut iterations = 0;
 
         // Early termination tracking: stop when top-k results are stable
@@ -1977,7 +1978,7 @@ impl HnswOnlinePartition {
             last_best_distance = current_best;
 
             // Early termination: if best distance is stable for 3 iterations and we have enough candidates
-            if stable_count >= 3 && search.history.len() >= ef {
+            if stable_count >= 3 && search.history.len() >= ef && priority_frontier.len() >= ef {
                 break;
             }
 
@@ -2025,15 +2026,13 @@ impl HnswOnlinePartition {
 
             iterations += 1;
 
-            // Check if we have enough candidates
-            if search.history.len() >= ef
-                && priority_frontier
+            if search.history.len() >= ef && priority_frontier.len() >= ef {
+                let frontier_best = priority_frontier
                     .peek_min_distance()
-                    .unwrap_or(Distance::INFINITY)
-                    > last_best_distance
-            {
-                // We have enough candidates and frontier has worse candidates, stop
-                break;
+                    .unwrap_or(Distance::INFINITY);
+                if frontier_best > last_best_distance {
+                    break;
+                }
             }
         }