Search vs consolidation time

eaasna · Sep 26, 2023 · a1e8527 · a1e8527
1 parent 8899b64
commit a1e8527
Show file tree

Hide file tree

Showing 3 changed files with 15 additions and 10 deletions.
diff --git a/include/valik/search/iterate_queries.hpp b/include/valik/search/iterate_queries.hpp
@@ -43,7 +43,7 @@ void iterate_distributed_queries(search_arguments const & arguments,
         start = std::chrono::high_resolution_clock::now();
         prefilter_queries_parallel(ibf, arguments, query_records, thresholder, queue);
         end = std::chrono::high_resolution_clock::now();
-        time_statistics.prefilter_time += std::chrono::duration_cast<std::chrono::duration<double>>(end - start).count();
+        time_statistics.search_time += std::chrono::duration_cast<std::chrono::duration<double>>(end - start).count();
     }
 }
 
@@ -101,7 +101,7 @@ void iterate_short_queries(search_arguments const & arguments,
             auto start = std::chrono::high_resolution_clock::now();
             prefilter_queries_parallel<shared_query_record<seqan2::String<seqan2::Dna>>>(ibf, arguments, query_records, thresholder, queue);
             auto end = std::chrono::high_resolution_clock::now();
-            time_statistics.prefilter_time += std::chrono::duration_cast<std::chrono::duration<double>>(end - start).count();
+            time_statistics.search_time += std::chrono::duration_cast<std::chrono::duration<double>>(end - start).count();
             query_records.clear();
         }
     }
@@ -112,7 +112,7 @@ void iterate_short_queries(search_arguments const & arguments,
     auto start = std::chrono::high_resolution_clock::now();
     prefilter_queries_parallel<shared_query_record<seqan2::String<seqan2::Dna>>>(ibf, arguments, query_records, thresholder, queue);
     auto end = std::chrono::high_resolution_clock::now();
-    time_statistics.prefilter_time += std::chrono::duration_cast<std::chrono::duration<double>>(end - start).count();
+    time_statistics.search_time += std::chrono::duration_cast<std::chrono::duration<double>>(end - start).count();
 }
 
 /**
@@ -175,7 +175,7 @@ void iterate_split_queries(search_arguments const & arguments,
                 auto start = std::chrono::high_resolution_clock::now();
                 prefilter_queries_parallel<shared_query_record<seqan2::String<seqan2::Dna>>>(ibf, arguments, query_records, thresholder, queue);
                 auto end = std::chrono::high_resolution_clock::now();
-                time_statistics.prefilter_time += std::chrono::duration_cast<std::chrono::duration<double>>(end - start).count();
+                time_statistics.search_time += std::chrono::duration_cast<std::chrono::duration<double>>(end - start).count();
                 query_records.clear();
             }
         }
@@ -188,7 +188,7 @@ void iterate_split_queries(search_arguments const & arguments,
     auto start = std::chrono::high_resolution_clock::now();
     prefilter_queries_parallel<shared_query_record<seqan2::String<seqan2::Dna>>>(ibf, arguments, query_records, thresholder, queue);
     auto end = std::chrono::high_resolution_clock::now();
-    time_statistics.prefilter_time += std::chrono::duration_cast<std::chrono::duration<double>>(end - start).count();
+    time_statistics.search_time += std::chrono::duration_cast<std::chrono::duration<double>>(end - start).count();
 }
 
 }   // namespace valik::app
diff --git a/include/valik/search/search_time_statistics.hpp b/include/valik/search/search_time_statistics.hpp
@@ -13,7 +13,8 @@ struct search_time_statistics
     double index_io_time{0.0};
     std::vector<double> cart_processing_times;
     double reads_io_time{0.0};
-    double prefilter_time{0.0};
+    double search_time{0.0};
+    double consolidation_time{0.0};
 
     double get_cart_min() const
     {
@@ -41,22 +42,23 @@ inline void write_time_statistics(search_time_statistics const & time_statistics
     std::filesystem::path file_path{time_file};
     std::ofstream file_handle(file_path, std::ofstream::app);
 
-    file_handle << "Ref I/O\tIBF I/O\tReads I/O\tPrefilter\tMin cart time\tAvg cart time\tMax cart time\tNr carts\n";
+    file_handle << "Ref I/O\tIBF I/O\tReads I/O\tSearch\tMin cart time\tAvg cart time\tMax cart time\tNr carts\tConsolidation\n";
     file_handle << std::fixed
                 << std::setprecision(2)
                 << time_statistics.ref_io_time << '\t'
                 << time_statistics.index_io_time << '\t'
                 << time_statistics.reads_io_time << '\t'
-                << time_statistics.prefilter_time << '\t';
+                << time_statistics.search_time << '\t';
     if (!time_statistics.cart_processing_times.empty())
     {
         file_handle << time_statistics.get_cart_min() << '\t'
                     << time_statistics.get_cart_avg() << '\t'
                     << time_statistics.get_cart_max() << '\t'
-                    << time_statistics.cart_processing_times.size() << '\n';
-
+                    << time_statistics.cart_processing_times.size() << '\t';
     }
 
+    file_handle << time_statistics.consolidation_time << '\n';
+
 }
 
 } // namespace valik::app
diff --git a/src/valik_search.cpp b/src/valik_search.cpp
@@ -45,6 +45,7 @@ void valik_search(search_arguments const & arguments)
     }
 
     // Consolidate matches (not necessary when searching a metagenomic database)
+    auto start = std::chrono::high_resolution_clock::now();
     if (!arguments.ref_meta_path.empty())
     {
         consolidate_matches(arguments);
@@ -53,6 +54,8 @@ void valik_search(search_arguments const & arguments)
             std::cerr << "Could not clean up intermediate file: \t" << std::string(arguments.all_matches) << '\n';
         failed = failed || error_in_delete;
     }
+    auto end = std::chrono::high_resolution_clock::now();
+    time_statistics.consolidation_time += std::chrono::duration_cast<std::chrono::duration<double>>(end - start).count();
 
     if (arguments.write_time)
         write_time_statistics(time_statistics, arguments.out_file.string() + ".time");