From 7e7fd9f7339eee2f1ae974c662a447532dfb1b1a Mon Sep 17 00:00:00 2001
From: "Douglas B. Rumbaugh" <dbr4@psu.edu>
Date: Tue, 26 Mar 2024 16:35:12 -0400
Subject: Updated FSTrie benchmark and some minor fixes

---
 benchmarks/string_insertion_tput.cpp | 73 ++++++++++++++++++++++--------------
 include/query/pointlookup.h          |  4 +-
 include/shard/FSTrie.h               |  2 +-
 3 files changed, 49 insertions(+), 30 deletions(-)

diff --git a/benchmarks/string_insertion_tput.cpp b/benchmarks/string_insertion_tput.cpp
index d205175..e41e996 100644
--- a/benchmarks/string_insertion_tput.cpp
+++ b/benchmarks/string_insertion_tput.cpp
@@ -9,7 +9,7 @@
 
 #include "framework/DynamicExtension.h"
 #include "shard/FSTrie.h"
-#include "query/rangequery.h"
+#include "query/pointlookup.h"
 #include "framework/interface/Record.h"
 
 #include "psu-util/timer.h"
@@ -18,8 +18,8 @@
 
 typedef de::Record<std::string, uint64_t> Rec;
 typedef de::FSTrie<Rec> Trie;
-typedef de::rq::Query<Rec, Trie> Q;
-typedef de::DynamicExtension<Rec, Trie, Q> Ext; //, de::LayoutPolicy::TEIRING, de::DeletePolicy::TAGGING, de::SerialScheduler> Ext;
+typedef de::pl::Query<Rec, Trie> Q;
+typedef de::DynamicExtension<Rec, Trie, Q, de::LayoutPolicy::TEIRING, de::DeletePolicy::TAGGING, de::SerialScheduler> Ext;
 
 std::vector<std::string> strings;
 
@@ -47,45 +47,62 @@ void read_data(std::string fname, size_t n=10000000) {
     }
 }
 
+void usage(char *name) {
+    fprintf(stderr, "Usage:\n%s datafile record_count\n", name);
+}
+
 int main(int argc, char **argv) {
-    size_t n = 100000000;
 
-    std::vector<int> counts = {1 , 2, 4, 8}; //, 16, 32, 64};
-    //
-    read_data("benchmarks/data/ursa-genome.txt", n);
+    if (argc < 3) {
+        usage(argv[0]);
+        exit(EXIT_FAILURE);
+    }
 
-    fprintf(stderr, "Finished reading from file.\n");
+    std::string fname = std::string(argv[1]);
+    size_t n = atol(argv[2]);
 
-    for (auto thread_count : counts) {
+    read_data(fname, n);
 
-        auto extension = new Ext(1000, 12000, 8);
+    if (strings.size() == 0) {
+        fprintf(stderr, "[E]: No string data read from file. Aborting execution.\n");
+    } else {
+        fprintf(stderr, "Finished reading from file.\n");
+    }
 
-        size_t per_thread = n / thread_count;
+    auto extension = new Ext(1000, 12000, 8);
 
-        std::thread threads[thread_count];
+    TIMER_INIT();
+    TIMER_START();
+    insert_thread(0, strings.size(), extension);
+    TIMER_STOP();
 
-        TIMER_INIT();
-        TIMER_START();
-        for (size_t i=0; i<thread_count; i++) {
-            threads[i] = std::thread(insert_thread, i*per_thread, 
-                                     i*per_thread+per_thread, extension);
-        }
+    auto total_time = TIMER_RESULT();
 
-        for (size_t i=0; i<thread_count; i++) {
-            threads[i].join();
-        }
+    size_t m = 100;
+    TIMER_START();
+    for (size_t i=0; i<m; i++) {
+        size_t j = rand() % strings.size();
+        de::pl::Parms<Rec> parms;
+        parms.search_key = strings[j];
 
-        TIMER_STOP();
+        auto res = extension->query(&parms);
+        auto ans = res.get();
 
-        auto total_time = TIMER_RESULT();
+        assert(ans[0].value == j);
+    }
+    TIMER_STOP();
 
-        double tput = (double) n / (double) total_time * 1e9;
+    auto query_time = TIMER_RESULT();
 
-        fprintf(stdout, "%ld\t%d\t%lf\n", extension->get_record_count(), 
-                thread_count, tput);
 
-        delete extension;
-    }
+    double i_tput = (double) n / (double) total_time * 1e9;
+    size_t q_lat = total_time / m;
+
+    fprintf(stdout, "%ld\t\t%lf\t%ld\n", extension->get_record_count(), 
+            i_tput, q_lat);
+
+
+    delete extension;
 
     fflush(stderr);
 }
diff --git a/include/query/pointlookup.h b/include/query/pointlookup.h
index caaa320..35d38e3 100644
--- a/include/query/pointlookup.h
+++ b/include/query/pointlookup.h
@@ -97,10 +97,12 @@ public:
                     return output;
                 }
 
-                output.append(r[0].rec);
+                output.push_back(r[0].rec);
                 return output;
             }
         }
+
+        return output;
     }
 
     static void delete_query_state(void *state) {
diff --git a/include/shard/FSTrie.h b/include/shard/FSTrie.h
index 50bf982..95f396f 100644
--- a/include/shard/FSTrie.h
+++ b/include/shard/FSTrie.h
@@ -61,7 +61,7 @@ public:
         std::sort(base, stop, std::less<Wrapped<R>>());
 
         for (size_t i=0; i<buffer.get_record_count(); i++) {
-            if (temp_buffer[i].is_deleted() || !temp_buffer[i].is_visible()) {
+            if (temp_buffer[i].is_deleted() || !temp_buffer[i].is_visible() || temp_buffer[i].rec.key == "") {
                 continue;
             }
 
-- 
cgit v1.2.3