From 675cf7f7558ebaef15f398d90cc3d1d91457b219 Mon Sep 17 00:00:00 2001
From: Douglas Rumbaugh <dbr4@psu.edu>
Date: Fri, 3 May 2024 11:01:47 -0400
Subject: FST benchmarks

---
 benchmarks/include/file_util.h    |  45 ++++++++++++++++-
 benchmarks/vldb/fst_bench.cpp     | 100 ++++++++++++++++++++++++++++++++++++++
 benchmarks/vldb/fst_bsm_bench.cpp | 100 ++++++++++++++++++++++++++++++++++++++
 3 files changed, 244 insertions(+), 1 deletion(-)
 create mode 100644 benchmarks/vldb/fst_bench.cpp
 create mode 100644 benchmarks/vldb/fst_bsm_bench.cpp

(limited to 'benchmarks')

diff --git a/benchmarks/include/file_util.h b/benchmarks/include/file_util.h
index ebcf17e..586b44f 100644
--- a/benchmarks/include/file_util.h
+++ b/benchmarks/include/file_util.h
@@ -1,3 +1,5 @@
+#pragma once
+
 #include <cstdlib>
 #include <cstdio>
 #include <iostream>
@@ -7,8 +9,10 @@
 #include <gsl/gsl_rng.h>
 #include <cstring>
 #include <vector>
+#include <memory>
+
+#include "psu-util/progress.h"
 
-#pragma once
 
 template <typename QP>
 static std::vector<QP> read_lookup_queries(std::string fname, double selectivity) {
@@ -35,6 +39,20 @@ static std::vector<QP> read_lookup_queries(std::string fname, double selectivity
     return queries;
 }
 
+template <typename QP>
+static std::vector<QP> generate_string_lookup_queries(std::vector<std::unique_ptr<char[]>> &strings, size_t cnt, gsl_rng *rng) {
+    std::vector<QP> queries;
+
+    for (size_t i=0; i<cnt; i++) {
+        auto idx = gsl_rng_uniform_int(rng, strings.size());
+        QP q;
+        q.search_key = strings[idx].get();
+        queries.push_back(q);
+    }
+
+    return queries;
+}
+
 template <typename QP>
 static std::vector<QP> read_range_queries(std::string &fname, double selectivity) {
     std::vector<QP> queries;
@@ -173,3 +191,28 @@ static std::vector<R> read_vector_file(std::string &fname, size_t n) {
 
     return records;
 }
+
+
+static std::vector<std::unique_ptr<char[]>>read_string_file(std::string fname, size_t n=10000000) {
+
+    std::fstream file;
+    file.open(fname, std::ios::in);
+
+    if (!file.is_open()) {
+        fprintf(stderr, "ERROR: Failed to open file %s\n", fname.c_str());
+        exit(EXIT_FAILURE);
+    }
+
+    std::vector<std::unique_ptr<char[]>> strings;
+    strings.reserve(n);
+
+    size_t i=0;
+    std::string line;
+    while (i < n && std::getline(file, line, '\n')) {
+        strings.emplace_back(std::unique_ptr<char[]>(strdup(line.c_str())));
+        i++;
+        psudb::progress_update((double) i / (double) n, "Reading file:");
+    }
+
+    return strings;
+}
diff --git a/benchmarks/vldb/fst_bench.cpp b/benchmarks/vldb/fst_bench.cpp
new file mode 100644
index 0000000..276a922
--- /dev/null
+++ b/benchmarks/vldb/fst_bench.cpp
@@ -0,0 +1,100 @@
+/*
+ *
+ */
+
+#define ENABLE_TIMER
+#define TS_TEST
+
+#include <thread>
+
+#include "framework/DynamicExtension.h"
+#include "shard/FSTrie.h"
+#include "query/pointlookup.h"
+#include "framework/interface/Record.h"
+#include "file_util.h"
+#include "standard_benchmarks.h"
+
+#include <gsl/gsl_rng.h>
+
+#include "psu-util/timer.h"
+
+
+typedef de::Record<const char *, uint64_t> Rec;
+typedef de::FSTrie<Rec> Shard;
+typedef de::pl::Query<Rec, Shard> Q;
+typedef de::DynamicExtension<Rec, Shard, Q, de::LayoutPolicy::TEIRING, de::DeletePolicy::TOMBSTONE, de::SerialScheduler> Ext;
+typedef de::pl::Parms<Rec> QP;
+
+void usage(char *progname) {
+    fprintf(stderr, "%s reccnt datafile\n", progname);
+}
+
+int main(int argc, char **argv) {
+
+    if (argc < 3) {
+        usage(argv[0]);
+        exit(EXIT_FAILURE);
+    }
+
+    size_t n = atol(argv[1]);
+    std::string d_fname = std::string(argv[2]);
+
+    auto extension = new Ext(12000, 12001, 8, 0, 64);
+    gsl_rng * rng = gsl_rng_alloc(gsl_rng_mt19937);
+    
+    auto strings = read_string_file(d_fname, n);
+    auto queries = generate_string_lookup_queries<QP>(strings, 1000, rng);
+
+    std::vector<Rec> data;
+    for (size_t i=0; i<strings.size(); i++) {
+        data.push_back({strings[i].get(), i, strlen(strings[i].get())});
+    }
+
+    std::vector<size_t> to_delete(n * delete_proportion);
+    size_t j=0;
+    for (size_t i=0; i<strings.size() && j<to_delete.size(); i++) {
+        if (gsl_rng_uniform(rng) <= delete_proportion) {
+            to_delete[j++] = i;
+        } 
+    }
+
+    /* warmup structure w/ 10% of records */
+    size_t warmup = .1 * n;
+    size_t delete_idx = 0;
+    insert_records<Ext, Rec>(extension, 0, warmup, data, to_delete, delete_idx, false, rng);
+
+    extension->await_next_epoch();
+
+    TIMER_INIT();
+
+    TIMER_START();
+    insert_records<Ext, Rec>(extension, warmup, data.size(), data, to_delete, delete_idx, true, rng);
+    TIMER_STOP();
+
+    auto insert_latency = TIMER_RESULT();
+    size_t insert_throughput = (size_t) ((double) (n - warmup) / (double) insert_latency * 1e9);
+
+    TIMER_START();
+    run_queries<Ext, QP>(extension, queries);
+    TIMER_STOP();
+
+    auto query_latency = TIMER_RESULT() / queries.size();
+
+    auto shard = extension->create_static_structure();
+
+    TIMER_START();
+    run_static_queries<Shard, QP, Q>(shard, queries);
+    TIMER_STOP();
+
+    auto static_latency = TIMER_RESULT() / queries.size();
+
+    auto ext_size = extension->get_memory_usage() + extension->get_aux_memory_usage();
+    auto static_size = shard->get_memory_usage(); //+ shard->get_aux_memory_usage();
+
+    fprintf(stdout, "%ld\t%ld\t%ld\t%ld\t%ld\n", insert_throughput, query_latency, ext_size, static_latency, static_size);
+
+    gsl_rng_free(rng);
+    delete extension;
+    fflush(stderr);
+}
+
diff --git a/benchmarks/vldb/fst_bsm_bench.cpp b/benchmarks/vldb/fst_bsm_bench.cpp
new file mode 100644
index 0000000..15a441a
--- /dev/null
+++ b/benchmarks/vldb/fst_bsm_bench.cpp
@@ -0,0 +1,100 @@
+/*
+ *
+ */
+
+#define ENABLE_TIMER
+#define TS_TEST
+
+#include <thread>
+
+#include "framework/DynamicExtension.h"
+#include "shard/FSTrie.h"
+#include "query/pointlookup.h"
+#include "framework/interface/Record.h"
+#include "file_util.h"
+#include "standard_benchmarks.h"
+
+#include <gsl/gsl_rng.h>
+
+#include "psu-util/timer.h"
+
+
+typedef de::Record<const char *, uint64_t> Rec;
+typedef de::FSTrie<Rec> Shard;
+typedef de::pl::Query<Rec, Shard> Q;
+typedef de::DynamicExtension<Rec, Shard, Q, de::LayoutPolicy::BSM, de::DeletePolicy::TOMBSTONE, de::SerialScheduler> Ext;
+typedef de::pl::Parms<Rec> QP;
+
+void usage(char *progname) {
+    fprintf(stderr, "%s reccnt datafile\n", progname);
+}
+
+int main(int argc, char **argv) {
+
+    if (argc < 3) {
+        usage(argv[0]);
+        exit(EXIT_FAILURE);
+    }
+
+    size_t n = atol(argv[1]);
+    std::string d_fname = std::string(argv[2]);
+
+    auto extension = new Ext(1, 12001, 2, 0, 64);
+    gsl_rng * rng = gsl_rng_alloc(gsl_rng_mt19937);
+    
+    auto strings = read_string_file(d_fname, n);
+    auto queries = generate_string_lookup_queries<QP>(strings, 1000, rng);
+
+    std::vector<Rec> data;
+    for (size_t i=0; i<strings.size(); i++) {
+        data.push_back({strings[i].get(), i, strlen(strings[i].get())});
+    }
+
+    std::vector<size_t> to_delete(n * delete_proportion);
+    size_t j=0;
+    for (size_t i=0; i<strings.size() && j<to_delete.size(); i++) {
+        if (gsl_rng_uniform(rng) <= delete_proportion) {
+            to_delete[j++] = i;
+        } 
+    }
+
+    /* warmup structure w/ 10% of records */
+    size_t warmup = .1 * n;
+    size_t delete_idx = 0;
+    insert_records<Ext, Rec>(extension, 0, warmup, data, to_delete, delete_idx, false, rng);
+
+    extension->await_next_epoch();
+
+    TIMER_INIT();
+
+    TIMER_START();
+    insert_records<Ext, Rec>(extension, warmup, data.size(), data, to_delete, delete_idx, true, rng);
+    TIMER_STOP();
+
+    auto insert_latency = TIMER_RESULT();
+    size_t insert_throughput = (size_t) ((double) (n - warmup) / (double) insert_latency * 1e9);
+
+    TIMER_START();
+    run_queries<Ext, QP>(extension, queries);
+    TIMER_STOP();
+
+    auto query_latency = TIMER_RESULT() / queries.size();
+
+    auto shard = extension->create_static_structure();
+
+    TIMER_START();
+    run_static_queries<Shard, QP, Q>(shard, queries);
+    TIMER_STOP();
+
+    auto static_latency = TIMER_RESULT() / queries.size();
+
+    auto ext_size = extension->get_memory_usage() + extension->get_aux_memory_usage();
+    auto static_size = shard->get_memory_usage(); //+ shard->get_aux_memory_usage();
+
+    fprintf(stdout, "%ld\t%ld\t%ld\t%ld\t%ld\n", insert_throughput, query_latency, ext_size, static_latency, static_size);
+
+    gsl_rng_free(rng);
+    delete extension;
+    fflush(stderr);
+}
+
-- 
cgit v1.2.3