From 3dd16320d38f6312e7f5d7164b6bb7a4e14790fa Mon Sep 17 00:00:00 2001
From: Douglas Rumbaugh <dbr4@psu.edu>
Date: Fri, 9 Feb 2024 15:35:32 -0500
Subject: Benchmark updates

---
 benchmarks/irs_bench.cpp |   4 +-
 benchmarks/pgm_bench.cpp | 123 +++++++++++++++++++++++++++++++++++++++++++++++
 benchmarks/ts_bench.cpp  | 123 +++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 248 insertions(+), 2 deletions(-)
 create mode 100644 benchmarks/pgm_bench.cpp
 create mode 100644 benchmarks/ts_bench.cpp

(limited to 'benchmarks')

diff --git a/benchmarks/irs_bench.cpp b/benchmarks/irs_bench.cpp
index ddb4220..49b1630 100644
--- a/benchmarks/irs_bench.cpp
+++ b/benchmarks/irs_bench.cpp
@@ -17,7 +17,7 @@
 #include "psu-util/timer.h"
 
 
-typedef de::Record<int64_t, int64_t> Rec;
+typedef de::Record<uint64_t, uint64_t> Rec;
 typedef de::ISAMTree<Rec> ISAM;
 typedef de::irs::Query<Rec, ISAM> Q;
 typedef de::DynamicExtension<Rec, ISAM, Q, de::LayoutPolicy::TEIRING, de::DeletePolicy::TOMBSTONE, de::SerialScheduler> Ext;
@@ -73,7 +73,7 @@ void insert_records(Ext *extension, size_t start,
 int main(int argc, char **argv) {
 
     if (argc < 4) {
-        fprintf(stderr, "insert_query_tput reccnt datafile queryfile\n");
+        fprintf(stderr, "irs_bench reccnt datafile queryfile\n");
         exit(EXIT_FAILURE);
     }
 
diff --git a/benchmarks/pgm_bench.cpp b/benchmarks/pgm_bench.cpp
new file mode 100644
index 0000000..72d3b52
--- /dev/null
+++ b/benchmarks/pgm_bench.cpp
@@ -0,0 +1,123 @@
+/*
+ *
+ */
+
+#define ENABLE_TIMER
+
+#include <thread>
+
+#include "framework/DynamicExtension.h"
+#include "shard/PGM.h"
+#include "query/rangecount.h"
+#include "framework/interface/Record.h"
+#include "include/data-proc.h"
+
+#include <gsl/gsl_rng.h>
+
+#include "psu-util/timer.h"
+
+
+typedef de::Record<uint64_t, uint64_t> Rec;
+typedef de::PGM<Rec> S;
+typedef de::rc::Query<Rec, S> Q;
+typedef de::DynamicExtension<Rec, S, Q, de::LayoutPolicy::TEIRING, de::DeletePolicy::TAGGING, de::SerialScheduler> Ext;
+typedef de::rc::Parms<Rec> QP;
+
+void run_queries(Ext *extension, std::vector<QP> &queries, gsl_rng *rng) {
+    size_t total;
+    for (size_t i=0; i<queries.size(); i++) {
+        auto q = &queries[i];
+
+        auto res = extension->query(q);
+        auto r = res.get();
+        total += r.size();
+    }
+
+    fprintf(stderr, "%ld\n", total);
+}
+
+size_t g_deleted_records = 0;
+double delete_proportion = 0.05;
+
+void insert_records(Ext *extension, size_t start, 
+                    size_t stop,
+                    std::vector<int64_t> &records,
+                    std::vector<size_t> &to_delete,
+                    size_t &delete_idx,
+                    bool delete_records,
+                    gsl_rng *rng) {
+    size_t reccnt = 0;
+    Rec r;
+    for (size_t i=start; i<stop; i++) {
+        r.key = records[i];
+        r.value = i;
+
+        while (!extension->insert(r)) {
+            usleep(1);
+        }
+
+        if (delete_records && gsl_rng_uniform(rng) <= delete_proportion && to_delete[delete_idx] <= i) {
+            r.key = records[to_delete[delete_idx]];
+            r.value = (int64_t) (to_delete[delete_idx]);
+            while (!extension->erase(r)) {
+                usleep(1);
+            }
+            delete_idx++;
+            g_deleted_records++;
+        }
+    }
+}
+
+int main(int argc, char **argv) {
+
+    if (argc < 4) {
+        fprintf(stderr, "pgm_bench reccnt datafile queryfile\n");
+        exit(EXIT_FAILURE);
+    }
+
+    size_t n = atol(argv[1]);
+    std::string d_fname = std::string(argv[2]);
+    std::string q_fname = std::string(argv[3]);
+
+    auto extension = new Ext(12000, 12001, 8, 0, 64);
+    gsl_rng * rng = gsl_rng_alloc(gsl_rng_mt19937);
+    
+    auto data = read_sosd_file(d_fname, n);
+    std::vector<size_t> to_delete(n * delete_proportion);
+    size_t j=0;
+    for (size_t i=0; i<data.size() && j<to_delete.size(); i++) {
+        if (gsl_rng_uniform(rng) <= delete_proportion) {
+            to_delete[j++] = i;
+        } 
+    }
+    auto queries = read_range_queries<QP>(q_fname, .001);
+
+    /* warmup structure w/ 10% of records */
+    size_t warmup = .1 * n;
+    size_t delete_idx = 0;
+    insert_records(extension, 0, warmup, data, to_delete, delete_idx, false, rng);
+
+    extension->await_next_epoch();
+
+    TIMER_INIT();
+
+    TIMER_START();
+    insert_records(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(extension, queries, rng);
+    TIMER_STOP();
+
+    auto query_latency = TIMER_RESULT() / queries.size();
+
+    fprintf(stdout, "T\t%ld\t%ld\t%ld\n", insert_throughput, query_latency, g_deleted_records);
+
+    gsl_rng_free(rng);
+    delete extension;
+    fflush(stderr);
+}
+
diff --git a/benchmarks/ts_bench.cpp b/benchmarks/ts_bench.cpp
new file mode 100644
index 0000000..3df3371
--- /dev/null
+++ b/benchmarks/ts_bench.cpp
@@ -0,0 +1,123 @@
+/*
+ *
+ */
+
+#define ENABLE_TIMER
+
+#include <thread>
+
+#include "framework/DynamicExtension.h"
+#include "shard/TrieSpline.h"
+#include "query/rangecount.h"
+#include "framework/interface/Record.h"
+#include "include/data-proc.h"
+
+#include <gsl/gsl_rng.h>
+
+#include "psu-util/timer.h"
+
+
+typedef de::Record<uint64_t, uint64_t> Rec;
+typedef de::TrieSpline<Rec> TS;
+typedef de::rc::Query<Rec, TS> Q;
+typedef de::DynamicExtension<Rec, TS, Q, de::LayoutPolicy::TEIRING, de::DeletePolicy::TAGGING, de::SerialScheduler> Ext;
+typedef de::rc::Parms<Rec> QP;
+
+void run_queries(Ext *extension, std::vector<QP> &queries, gsl_rng *rng) {
+    size_t total;
+    for (size_t i=0; i<queries.size(); i++) {
+        auto q = &queries[i];
+
+        auto res = extension->query(q);
+        auto r = res.get();
+        total += r.size();
+    }
+
+    fprintf(stderr, "%ld\n", total);
+}
+
+size_t g_deleted_records = 0;
+double delete_proportion = 0.05;
+
+void insert_records(Ext *extension, size_t start, 
+                    size_t stop,
+                    std::vector<int64_t> &records,
+                    std::vector<size_t> &to_delete,
+                    size_t &delete_idx,
+                    bool delete_records,
+                    gsl_rng *rng) {
+    size_t reccnt = 0;
+    Rec r;
+    for (size_t i=start; i<stop; i++) {
+        r.key = records[i];
+        r.value = i;
+
+        while (!extension->insert(r)) {
+            usleep(1);
+        }
+
+        if (delete_records && gsl_rng_uniform(rng) <= delete_proportion && to_delete[delete_idx] <= i) {
+            r.key = records[to_delete[delete_idx]];
+            r.value = (int64_t) (to_delete[delete_idx]);
+            while (!extension->erase(r)) {
+                usleep(1);
+            }
+            delete_idx++;
+            g_deleted_records++;
+        }
+    }
+}
+
+int main(int argc, char **argv) {
+
+    if (argc < 4) {
+        fprintf(stderr, "ts_bench reccnt datafile queryfile\n");
+        exit(EXIT_FAILURE);
+    }
+
+    size_t n = atol(argv[1]);
+    std::string d_fname = std::string(argv[2]);
+    std::string q_fname = std::string(argv[3]);
+
+    auto extension = new Ext(12000, 12001, 8, 0, 64);
+    gsl_rng * rng = gsl_rng_alloc(gsl_rng_mt19937);
+    
+    auto data = read_sosd_file(d_fname, n);
+    std::vector<size_t> to_delete(n * delete_proportion);
+    size_t j=0;
+    for (size_t i=0; i<data.size() && j<to_delete.size(); i++) {
+        if (gsl_rng_uniform(rng) <= delete_proportion) {
+            to_delete[j++] = i;
+        } 
+    }
+    auto queries = read_range_queries<QP>(q_fname, .001);
+
+    /* warmup structure w/ 10% of records */
+    size_t warmup = .1 * n;
+    size_t delete_idx = 0;
+    insert_records(extension, 0, warmup, data, to_delete, delete_idx, false, rng);
+
+    extension->await_next_epoch();
+
+    TIMER_INIT();
+
+    TIMER_START();
+    insert_records(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(extension, queries, rng);
+    TIMER_STOP();
+
+    auto query_latency = TIMER_RESULT() / queries.size();
+
+    fprintf(stdout, "T\t%ld\t%ld\t%ld\n", insert_throughput, query_latency, g_deleted_records);
+
+    gsl_rng_free(rng);
+    delete extension;
+    fflush(stderr);
+}
+
-- 
cgit v1.2.3