Added QuART_tail stuff

Author: cangokmen

.gitignore

@@ -1,3 +1,4 @@
 **/*.csv
 build/
-.DS_Store
+.DS_Store
+results/

CMakeLists.txt

@@ -17,3 +17,4 @@ set(CMAKE_CXX_FLAGS_DEBUG "-g -O0")
 set(CMAKE_CXX_FLAGS_RELEASE "-O3")
 add_executable(main main.cpp)
 add_executable(test_range test_range_query.cpp)
+add_executable(run run.cpp)

run.cpp

@@ -0,0 +1,108 @@
+#include <algorithm>
+#include <chrono>
+#include <fstream>
+#include <iostream>
+#include <vector>
+
+#include "ART.h"
+#include "ArtNode.h"
+#include "ArtNodeNewMethods.cpp"
+#include "Chain.h"
+#include "Helper.h"
+#include "trees/QuART_tail.h"
+
+using namespace std;
+
+template <typename key_type>
+std::vector<key_type> read_bin(const char* filename) {
+    std::ifstream inputFile(filename, std::ios::binary);
+    inputFile.seekg(0, std::ios::end);
+    const std::streampos fileSize = inputFile.tellg();
+    inputFile.seekg(0, std::ios::beg);
+    std::vector<key_type> data(fileSize / sizeof(key_type));
+    inputFile.read(reinterpret_cast<char*>(data.data()), fileSize);
+    return data;
+}
+
+int main(int argc, char** argv) {
+    bool verbose = false;      // optional argument
+    int N = 5000000;           // optional argument
+    string input_file;         // required argument
+    string tree_type = "ART";  // default tree type
+
+    // Parse arguments; make sure to increment i by 2 if you consume an argument
+    for (int i = 1; i < argc;) {
+        if (string(argv[i]) == "-v") {
+            verbose = true;
+            i++;
+        } else if (string(argv[i]) == "-N") {
+            N = atoi(argv[i + 1]);
+            i += 2;
+        } else if (string(argv[i]) == "-f") {
+            input_file = argv[i + 1];
+            i += 2;
+        } else if (string(argv[i]) == "-t") {
+            tree_type = argv[i + 1];
+            i += 2;
+        } else {
+            i++;
+        }
+    }
+
+    // read data
+    auto keys = read_bin<uint32_t>(input_file.c_str());
+
+    // Build tree
+    ART::ART* tree = nullptr;
+    if (tree_type == "ART") {
+        tree = new ART::ART();
+    } else if (tree_type == "QuART_tail") {
+        tree = new ART::QuART_tail();
+    } else {
+        cerr << "Unknown tree type: " << tree_type << endl;
+        return 1;
+    }
+
+    if (verbose) {
+        cout << "Tree type: " << tree_type << endl;
+    }
+
+    long long insertion_time = 0;
+    for (uint64_t i = 0; i < N; i++) {
+        uint8_t key[4];
+        ART::loadKey(keys[i], key);
+        auto start = chrono::high_resolution_clock::now();
+        tree->insert(key, keys[i]);
+        auto stop = chrono::high_resolution_clock::now();
+        auto duration =
+            chrono::duration_cast<chrono::nanoseconds>(stop - start);
+        insertion_time += duration.count();
+    }
+
+    if (verbose) {
+        cout << "Insertion time: " << insertion_time << " ns" << endl;
+    }
+
+    // Query tree
+    long long query_time = 0;
+    for (uint64_t i = 0; i < N; i++) {
+        uint8_t key[4];
+        ART::loadKey(keys[i], key);
+        auto start = chrono::high_resolution_clock::now();
+        ART::ArtNode* leaf = tree->lookup(key);
+        auto stop = chrono::high_resolution_clock::now();
+        auto duration =
+            chrono::duration_cast<chrono::nanoseconds>(stop - start);
+        query_time += duration.count();
+        assert(ART::isLeaf(leaf) && ART::getLeafValue(leaf) == keys[i]);
+    }
+
+    if (verbose) {
+        cout << "Query time: " << query_time << " ns" << endl;
+    }
+
+    // Output the times in csv format, including tree type
+    cout << insertion_time << "," << query_time << endl;
+
+    return 0;
+}

run_experiments.sh

@@ -0,0 +1,41 @@
+#!/bin/bash
+SUFFIX=$(date +"%Y%m%d_%H%M%S")
+RESULTSDIR="results"
+RESULTS="${RESULTSDIR}/results_${SUFFIX}.txt"
+LOGDIR="${RESULTSDIR}/logs"
+
+mkdir -p "$LOGDIR"
+
+echo "N,K,L,type_of_tree,avg_insert_time,avg_query_time" > "$RESULTS"
+
+REPEAT=1
+
+for FILE in ../bods/workloads/workload_N*_K*_L*.bin; do
+    [ -f "$FILE" ] || continue
+
+    BASENAME=$(basename "$FILE")
+    N=$(echo "$BASENAME" | sed -n 's/.*_N\([0-9]*\)_K[0-9]*_L[0-9]*.bin/\1/p')
+    K=$(echo "$BASENAME" | sed -n 's/.*_N[0-9]*_K\([0-9]*\)_L[0-9]*.bin/\1/p')
+    L=$(echo "$BASENAME" | sed -n 's/.*_N[0-9]*_K[0-9]*_L\([0-9]*\).bin/\1/p')
+    LOGFILE="${LOGDIR}/log_${BASENAME%.txt}_${SUFFIX}.txt"
+
+    for TREE in ART QuART_tail; do
+        INSERT_SUM=0
+        QUERY_SUM=0
+
+        for ((i=1; i<=REPEAT; i++)); do
+            echo "Running $TREE on $FILE (run $i/$REPEAT)" >> "$LOGFILE"
+            OUTPUT=$(./build/run -v -f "$FILE" -N "$N" -t "$TREE" 2>>"$LOGFILE")
+            echo "$OUTPUT" >> "$LOGFILE"
+            CSV_LINE=$(echo "$OUTPUT" | tail -1)
+            INSERT_TIME=$(echo "$CSV_LINE" | cut -d',' -f1 | xargs)
+            QUERY_TIME=$(echo "$CSV_LINE" | cut -d',' -f2 | xargs)
+            INSERT_SUM=$((INSERT_SUM + INSERT_TIME))
+            QUERY_SUM=$((QUERY_SUM + QUERY_TIME))
+        done
+
+        AVG_INSERT_TIME=$((INSERT_SUM / REPEAT))
+        AVG_QUERY_TIME=$((QUERY_SUM / REPEAT))
+        echo "$N,$K,$L,$TREE,$AVG_INSERT_TIME,$AVG_QUERY_TIME" >> "$RESULTS"
+    done
+done

trees/QuART_tail.h

@@ -1,10 +1,6 @@
 #pragma once
 
-#include "ART.h"
-#include "ArtNode.h"
-
-namespace ART {
-
+namespace ART {    
     class QuART_tail : public ART {
         public: