speedup get_output_histogram for all amounts when min_count > 0

This skips the vast majority of "dust" output amounts with just
one instance on the chain. Clocks in at 0.15% of the original
time on testnet.

2 files changed, 9 insertions, 5 deletions

diff --git a/src/blockchain_db/lmdb/db_lmdb.cpp b/src/blockchain_db/lmdb/db_lmdb.cpp
index 9a755fb0c..5672a34f6 100644
--- a/src/blockchain_db/lmdb/db_lmdb.cpp
+++ b/src/blockchain_db/lmdb/db_lmdb.cpp
@@ -3086,7 +3086,7 @@ void BlockchainLMDB::get_output_tx_and_index(const uint64_t& amount, const std::
   LOG_PRINT_L3("db3: " << db3);
 }
 
-std::map<uint64_t, std::tuple<uint64_t, uint64_t, uint64_t>> BlockchainLMDB::get_output_histogram(const std::vector<uint64_t> &amounts, bool unlocked, uint64_t recent_cutoff) const
+std::map<uint64_t, std::tuple<uint64_t, uint64_t, uint64_t>> BlockchainLMDB::get_output_histogram(const std::vector<uint64_t> &amounts, bool unlocked, uint64_t recent_cutoff, uint64_t min_count) const
 {
   LOG_PRINT_L3("BlockchainLMDB::" << __func__);
   check_open();
@@ -3112,7 +3112,8 @@ std::map<uint64_t, std::tuple<uint64_t, uint64_t, uint64_t>> BlockchainLMDB::get
       mdb_size_t num_elems = 0;
       mdb_cursor_count(m_cur_output_amounts, &num_elems);
       uint64_t amount = *(const uint64_t*)k.mv_data;
-      histogram[amount] = std::make_tuple(num_elems, 0, 0);
+      if (num_elems >= min_count)
+        histogram[amount] = std::make_tuple(num_elems, 0, 0);
     }
   }
   else
@@ -3123,13 +3124,15 @@ std::map<uint64_t, std::tuple<uint64_t, uint64_t, uint64_t>> BlockchainLMDB::get
       int ret = mdb_cursor_get(m_cur_output_amounts, &k, &v, MDB_SET);
       if (ret == MDB_NOTFOUND)
       {
-        histogram[amount] = std::make_tuple(0, 0, 0);
+        if (0 >= min_count)
+          histogram[amount] = std::make_tuple(0, 0, 0);
       }
       else if (ret == MDB_SUCCESS)
       {
         mdb_size_t num_elems = 0;
         mdb_cursor_count(m_cur_output_amounts, &num_elems);
-        histogram[amount] = std::make_tuple(num_elems, 0, 0);
+        if (num_elems >= min_count)
+          histogram[amount] = std::make_tuple(num_elems, 0, 0);
       }
       else
       {
diff --git a/src/blockchain_db/lmdb/db_lmdb.h b/src/blockchain_db/lmdb/db_lmdb.h
index 0aa4bb86e..9a20c0f1e 100644
--- a/src/blockchain_db/lmdb/db_lmdb.h
+++ b/src/blockchain_db/lmdb/db_lmdb.h
@@ -287,10 +287,11 @@ public:
    * @param amounts optional set of amounts to lookup
    * @param unlocked whether to restrict count to unlocked outputs
    * @param recent_cutoff timestamp to determine which outputs are recent
+   * @param min_count return only amounts with at least that many instances
    *
    * @return a set of amount/instances
    */
-  std::map<uint64_t, std::tuple<uint64_t, uint64_t, uint64_t>> get_output_histogram(const std::vector<uint64_t> &amounts, bool unlocked, uint64_t recent_cutoff) const;
+  std::map<uint64_t, std::tuple<uint64_t, uint64_t, uint64_t>> get_output_histogram(const std::vector<uint64_t> &amounts, bool unlocked, uint64_t recent_cutoff, uint64_t min_count) const;
 
 private:
   void do_resize(uint64_t size_increase=0);