Merge pull request #403 from eseiler/misc/shuffle

[MISC] Shuffle records in search

include/raptor/search/search_singular_ibf.hpp

@@ -12,6 +12,8 @@
 
 #pragma once
 
+#include <random>
+
 #include <seqan3/search/views/minimiser_hash.hpp>
 
 #include <raptor/adjust_seed.hpp>
@@ -140,6 +142,8 @@ void search_singular_ibf(search_arguments const & arguments, index_t && index)
         records.clear();
         arguments.query_file_io_timer.start();
         std::ranges::move(chunked_records, std::back_inserter(records));
+        // Very fast, improves parallel processing when chunks of the query belong to the same bin.
+        std::ranges::shuffle(records, std::mt19937_64{0u});
         arguments.query_file_io_timer.stop();
 
         cereal_future.get();

src/search/search_partitioned_ibf.cpp

@@ -10,6 +10,8 @@
  * \author Enrico Seiler <enrico.seiler AT fu-berlin.de>
  */
 
+#include <random>
+
 #include <seqan3/search/views/minimiser_hash.hpp>
 
 #include <raptor/adjust_seed.hpp>
@@ -55,6 +57,8 @@ void search_partitioned_ibf(search_arguments const & arguments)
         records.clear();
         arguments.query_file_io_timer.start();
         std::ranges::move(chunked_records, std::back_inserter(records));
+        // Very fast, improves parallel processing when chunks of the query belong to the same bin.
+        std::ranges::shuffle(records, std::mt19937_64{0u});
         arguments.query_file_io_timer.stop();
 
         cereal_future.get();

test/include/raptor/test/cli_test.hpp

@@ -375,7 +375,7 @@ struct raptor_base : public cli_test
         ASSERT_EQ(line, "#QUERY_NAME\tUSER_BINS");
 
         std::string const query_prefix{"query"};
-        for (char i : {'1', '2', '3'})
+        for (char i : {'3', '1', '2'})
         {
             ASSERT_TRUE(std::getline(search_result, line));
             std::string_view line_view{line};