Merge pull request #293 from smehringer/minor_things

[MISC] Reduce complexity of insert_into_ibf.

include/raptor/build/hibf/insert_into_ibf.hpp

@@ -23,14 +23,11 @@ namespace raptor::hibf
 {
 
 // automatically does naive splitting if number_of_bins > 1
-void insert_into_ibf(robin_hood::unordered_flat_set<size_t> & parent_kmers,
-                     robin_hood::unordered_flat_set<size_t> const & kmers,
+void insert_into_ibf(robin_hood::unordered_flat_set<size_t> const & kmers,
                      size_t const number_of_bins,
                      size_t const bin_index,
                      seqan3::interleaved_bloom_filter<> & ibf,
-                     bool is_root,
-                     timer<concurrent::yes> & fill_ibf_timer,
-                     timer<concurrent::yes> & merge_kmers_timer);
+                     timer<concurrent::yes> & fill_ibf_timer);
 
 void insert_into_ibf(build_arguments const & arguments,
                      chopper_pack_record const & record,

include/raptor/build/hibf/update_parent_kmers.hpp

@@ -0,0 +1,26 @@
+// --------------------------------------------------------------------------------------------------
+// Copyright (c) 2006-2023, Knut Reinert & Freie Universität Berlin
+// Copyright (c) 2016-2023, Knut Reinert & MPI für molekulare Genetik
+// This file may be used, modified and/or redistributed under the terms of the 3-clause BSD-License
+// shipped with this file and also available at: https://github.com/seqan/raptor/blob/main/LICENSE.md
+// --------------------------------------------------------------------------------------------------
+
+/*!\file
+ * \brief Provides raptor::hibf::update_parent_kmers.
+ * \author Svenja Mehringer <svenja.mehringer AT fu-berlin.de>
+ */
+
+#pragma once
+
+#include <robin_hood.h>
+
+#include <raptor/argument_parsing/timer.hpp>
+
+namespace raptor::hibf
+{
+
+void update_parent_kmers(robin_hood::unordered_flat_set<size_t> & parent_kmers,
+                         robin_hood::unordered_flat_set<size_t> const & kmers,
+                         timer<concurrent::yes> & merge_kmers_timer);
+
+} // namespace raptor::hibf

src/build/hibf/CMakeLists.txt

@@ -13,6 +13,7 @@ if (NOT TARGET raptor_build_hibf)
                  parse_chopper_pack_header.cpp
                  parse_chopper_pack_line.cpp
                  read_chopper_pack_file.cpp
+                 update_parent_kmers.cpp
                  update_user_bins.cpp
     )
 

src/build/hibf/construct_ibf.cpp

@@ -15,6 +15,7 @@
 #include <raptor/build/hibf/bin_size_in_bits.hpp>
 #include <raptor/build/hibf/construct_ibf.hpp>
 #include <raptor/build/hibf/insert_into_ibf.hpp>
+#include <raptor/build/hibf/update_parent_kmers.hpp>
 
 namespace raptor::hibf
 {
@@ -41,14 +42,9 @@ seqan3::interleaved_bloom_filter<> construct_ibf(robin_hood::unordered_flat_set<
     local_index_allocation_timer.stop();
     arguments.index_allocation_timer += local_index_allocation_timer;
 
-    insert_into_ibf(parent_kmers,
-                    kmers,
-                    number_of_bins,
-                    node_data.max_bin_index,
-                    ibf,
-                    is_root,
-                    arguments.fill_ibf_timer,
-                    arguments.merge_kmers_timer);
+    insert_into_ibf(kmers, number_of_bins, node_data.max_bin_index, ibf, arguments.fill_ibf_timer);
+    if (!is_root)
+        update_parent_kmers(parent_kmers, kmers, arguments.merge_kmers_timer);
 
     return ibf;
 }

src/build/hibf/hierarchical_build.cpp

@@ -17,6 +17,7 @@
 #include <raptor/build/hibf/hierarchical_build.hpp>
 #include <raptor/build/hibf/insert_into_ibf.hpp>
 #include <raptor/build/hibf/loop_over_children.hpp>
+#include <raptor/build/hibf/update_parent_kmers.hpp>
 #include <raptor/build/hibf/update_user_bins.hpp>
 
 namespace raptor::hibf
@@ -86,14 +87,13 @@ size_t hierarchical_build(robin_hood::unordered_flat_set<size_t> & parent_kmers,
         else
         {
             compute_kmers(kmers, arguments, record);
-            insert_into_ibf(parent_kmers,
-                            kmers,
+            insert_into_ibf(kmers,
                             record.number_of_bins.back(),
                             record.bin_indices.back(),
                             ibf,
-                            is_root,
-                            arguments.fill_ibf_timer,
-                            arguments.merge_kmers_timer);
+                            arguments.fill_ibf_timer);
+            if (!is_root)
+                update_parent_kmers(parent_kmers, kmers, arguments.merge_kmers_timer);
         }
 
         update_user_bins(data, filename_indices, record);

src/build/hibf/insert_into_ibf.cpp

@@ -20,14 +20,11 @@ namespace raptor::hibf
 {
 
 // automatically does naive splitting if number_of_bins > 1
-void insert_into_ibf(robin_hood::unordered_flat_set<size_t> & parent_kmers,
-                     robin_hood::unordered_flat_set<size_t> const & kmers,
+void insert_into_ibf(robin_hood::unordered_flat_set<size_t> const & kmers,
                      size_t const number_of_bins,
                      size_t const bin_index,
                      seqan3::interleaved_bloom_filter<> & ibf,
-                     bool is_root,
-                     timer<concurrent::yes> & fill_ibf_timer,
-                     timer<concurrent::yes> & merge_kmers_timer)
+                     timer<concurrent::yes> & fill_ibf_timer)
 {
     size_t const chunk_size = kmers.size() / number_of_bins + 1;
     size_t chunk_number{};
@@ -44,13 +41,6 @@ void insert_into_ibf(robin_hood::unordered_flat_set<size_t> & parent_kmers,
     }
     local_fill_ibf_timer.stop();
     fill_ibf_timer += local_fill_ibf_timer;
-
-    timer<concurrent::no> local_merge_kmers_timer{};
-    local_merge_kmers_timer.start();
-    if (!is_root)
-        parent_kmers.insert(kmers.begin(), kmers.end());
-    local_merge_kmers_timer.stop();
-    merge_kmers_timer += local_merge_kmers_timer;
 }
 
 void insert_into_ibf(build_arguments const & arguments,

src/build/hibf/loop_over_children.cpp

@@ -17,6 +17,7 @@
 #include <raptor/build/hibf/hierarchical_build.hpp>
 #include <raptor/build/hibf/insert_into_ibf.hpp>
 #include <raptor/build/hibf/loop_over_children.hpp>
+#include <raptor/build/hibf/update_parent_kmers.hpp>
 
 namespace raptor::hibf
 {
@@ -55,14 +56,9 @@ void loop_over_children(robin_hood::unordered_flat_set<size_t> & parent_kmers,
                 size_t const mutex_id{parent_bin_index / 64};
                 std::lock_guard<std::mutex> guard{local_ibf_mutex[mutex_id]};
                 ibf_positions[parent_bin_index] = ibf_pos;
-                insert_into_ibf(parent_kmers,
-                                kmers,
-                                1,
-                                parent_bin_index,
-                                ibf,
-                                is_root,
-                                arguments.fill_ibf_timer,
-                                arguments.merge_kmers_timer);
+                insert_into_ibf(kmers, 1, parent_bin_index, ibf, arguments.fill_ibf_timer);
+                if (!is_root)
+                    update_parent_kmers(parent_kmers, kmers, arguments.merge_kmers_timer);
             }
         }
     };

src/build/hibf/update_parent_kmers.cpp

@@ -0,0 +1,29 @@
+// --------------------------------------------------------------------------------------------------
+// Copyright (c) 2006-2023, Knut Reinert & Freie Universität Berlin
+// Copyright (c) 2016-2023, Knut Reinert & MPI für molekulare Genetik
+// This file may be used, modified and/or redistributed under the terms of the 3-clause BSD-License
+// shipped with this file and also available at: https://github.com/seqan/raptor/blob/main/LICENSE.md
+// --------------------------------------------------------------------------------------------------
+
+/*!\file
+ * \brief Implements raptor::hibf::update_parent_kmers.
+ * \author Svenja Mehringer <svenja.mehringer AT fu-berlin.de>
+ */
+
+#include <raptor/build/hibf/update_parent_kmers.hpp>
+
+namespace raptor::hibf
+{
+
+void update_parent_kmers(robin_hood::unordered_flat_set<size_t> & parent_kmers,
+                         robin_hood::unordered_flat_set<size_t> const & kmers,
+                         timer<concurrent::yes> & merge_kmers_timer)
+{
+    timer<concurrent::no> local_merge_kmers_timer{};
+    local_merge_kmers_timer.start();
+    parent_kmers.insert(kmers.begin(), kmers.end());
+    local_merge_kmers_timer.stop();
+    merge_kmers_timer += local_merge_kmers_timer;
+}
+
+} // namespace raptor::hibf