From 4f34cc095875af8d0a7b4366c622cd4de5fb3998 Mon Sep 17 00:00:00 2001
From: Anton Korobeynikov <anton@korobeynikov.info>
Date: Fri, 20 Dec 2024 11:40:57 -0800
Subject: [PATCH] Ensure we're not losing variance in paired info improver. Use
 better unordered data structures

---
 src/common/paired_info/concurrent_pair_info_buffer.hpp |  6 ++++++
 src/common/paired_info/pair_info_improver.hpp          | 10 +++++-----
 src/common/paired_info/paired_info.hpp                 |  6 ++++++
 3 files changed, 17 insertions(+), 5 deletions(-)

diff --git a/src/common/paired_info/concurrent_pair_info_buffer.hpp b/src/common/paired_info/concurrent_pair_info_buffer.hpp
index 9048ceb2b..5d4305732 100644
--- a/src/common/paired_info/concurrent_pair_info_buffer.hpp
+++ b/src/common/paired_info/concurrent_pair_info_buffer.hpp
@@ -128,6 +128,12 @@ class ConcurrentPairedBuffer : public PairedBufferBase<ConcurrentPairedBuffer<G,
 template<class Graph>
 using ConcurrentPairedInfoBuffer = ConcurrentPairedBuffer<Graph, RawPointTraits, btree_map>;
 
+template<class Graph>
+using ConcurrentClusteredPairedInfoBuffer = ConcurrentPairedBuffer<Graph, PointTraits, btree_map>;
+
+template<class Graph>
+using ConcurrentUnorderedClusteredPairedInfoBuffer = ConcurrentPairedBuffer<Graph, PointTraits, phmap_map>;
+
 } // namespace de
 
 } // namespace omnigraph
diff --git a/src/common/paired_info/pair_info_improver.hpp b/src/common/paired_info/pair_info_improver.hpp
index eb19ae82d..b19ab6d51 100644
--- a/src/common/paired_info/pair_info_improver.hpp
+++ b/src/common/paired_info/pair_info_improver.hpp
@@ -45,7 +45,7 @@ class PairInfoImprover {
     typedef std::vector<omnigraph::de::PairInfo<EdgeId> > PairInfos;
     typedef std::pair<EdgeId, EdgeId> EdgePair;
     typedef omnigraph::de::PairedInfoIndexT<Graph> Index;
-    typedef omnigraph::de::ConcurrentPairedInfoBuffer<Graph> Buffer;
+    typedef omnigraph::de::ConcurrentUnorderedClusteredPairedInfoBuffer<Graph> Buffer;
 
   public:
     PairInfoImprover(const Graph& g,
@@ -123,7 +123,7 @@ class PairInfoImprover {
     }
 
     size_t RemoveContradictional(unsigned nthreads) {
-        omnigraph::de::ConcurrentPairedInfoBuffer<Graph> buf(graph_);
+        Buffer buf(graph_);
 
         omnigraph::IterationHelper<Graph, EdgeId> edges(graph_);
         auto ranges = edges.Ranges(nthreads * 16);
@@ -142,7 +142,7 @@ class PairInfoImprover {
         DEBUG("Merging maps");
         // FIXME: This is a bit crazy, but we do not have a sane way to iterate
         // over buffer. In any case, this is better than it used to be before
-        omnigraph::de::UnclusteredPairedInfoIndexT<Graph> to_remove(graph_);
+        omnigraph::de::UnorderedPairedInfoIndexT<Graph> to_remove(graph_);
         to_remove.MoveAssign(buf);
 
         DEBUG("Resulting size " << to_remove.size());
@@ -161,7 +161,7 @@ class PairInfoImprover {
 
     size_t FillMissing(unsigned nthreads) {
         DEBUG("Fill missing: Creating indexes");
-        omnigraph::de::ConcurrentPairedInfoBuffer<Graph> buf(graph_);
+        Buffer buf(graph_);
 
         SplitPathConstructor<Graph> spc(graph_);
 
@@ -189,7 +189,7 @@ class PairInfoImprover {
         DEBUG("Merging maps");
         // FIXME: This is a bit crazy, but we do not have a sane way to iterate
         // over buffer. In any case, this is better than it used to be before
-        omnigraph::de::UnclusteredPairedInfoIndexT<Graph> to_add(graph_);
+        omnigraph::de::UnorderedPairedInfoIndexT<Graph> to_add(graph_);
         to_add.MoveAssign(buf);
 
         DEBUG("Resulting size " << to_add.size());
diff --git a/src/common/paired_info/paired_info.hpp b/src/common/paired_info/paired_info.hpp
index ad9455034..7a2ac6af6 100644
--- a/src/common/paired_info/paired_info.hpp
+++ b/src/common/paired_info/paired_info.hpp
@@ -603,6 +603,12 @@ using const_btree_map = NoLockingConstAdapter<phmap::btree_map<K, V>>; //Two-par
 template<typename Graph>
 using PairedInfoIndexT = PairedIndex<Graph, PointTraits, const_btree_map>;
 
+template<typename K, typename V>
+using phmap_map = NoLockingAdapter<phmap::parallel_flat_hash_map<K, V>>; //Two-parameters wrapper
+
+template<typename Graph>
+using UnorderedPairedInfoIndexT = PairedIndex<Graph, PointTraits, phmap_map>;
+
 template<typename K, typename V>
 using btree_map = NoLockingAdapter<phmap::btree_map<K, V>>; //Two-parameters wrapper