From 69f36c9b4f5df19f09156689b333afe29a017eed Mon Sep 17 00:00:00 2001
From: Douglas Rumbaugh <dbr4@psu.edu>
Date: Fri, 23 Feb 2024 15:40:49 -0500
Subject: Benchmark updates

---
 benchmarks/include/file_util.h | 130 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 130 insertions(+)
 create mode 100644 benchmarks/include/file_util.h

(limited to 'benchmarks/include/file_util.h')

diff --git a/benchmarks/include/file_util.h b/benchmarks/include/file_util.h
new file mode 100644
index 0000000..2a3300a
--- /dev/null
+++ b/benchmarks/include/file_util.h
@@ -0,0 +1,130 @@
+#include <cstdlib>
+#include <cstdio>
+#include <iostream>
+#include <fstream>
+#include <sstream>
+#include <string>
+#include <gsl/gsl_rng.h>
+#include <cstring>
+#include <vector>
+
+#include "framework/interface/Record.h"
+#include "query/irs.h"
+
+#pragma once
+
+template <typename QP>
+static std::vector<QP> read_lookup_queries(std::string fname, double selectivity) {
+    std::vector<QP> queries;
+
+    FILE *qf = fopen(fname.c_str(), "r");
+    size_t start, stop;
+    double sel;
+    while (fscanf(qf, "%zu%zu%lf\n", &start, &stop, &sel) != EOF) {
+        if (start < stop && std::abs(sel - selectivity) < 0.1) {
+            QP q;
+            q.target_key = start;
+            queries.push_back(q);
+        }
+    }
+    fclose(qf);
+
+    return queries;
+}
+
+template <typename QP>
+static std::vector<QP> read_range_queries(std::string &fname, double selectivity) {
+    std::vector<QP> queries;
+
+    FILE *qf = fopen(fname.c_str(), "r");
+    size_t start, stop;
+    double sel;
+    while (fscanf(qf, "%zu%zu%lf\n", &start, &stop, &sel) != EOF) {
+        if (start < stop && std::abs(sel - selectivity) < 0.1) {
+            QP q;
+            q.lower_bound = start;
+            q.upper_bound = stop;
+
+            queries.push_back(q);
+        }
+    }
+    fclose(qf);
+
+    return queries;
+}
+
+template <typename QP>
+static std::vector<QP> read_knn_queries(std::string fname, size_t k) {
+    std::vector<QP> queries;
+
+    FILE *qf = fopen(fname.c_str(), "r");
+    char *line = NULL;
+    size_t len = 0;
+
+    while (getline(&line, &len, qf) > 0) {
+        char *token;
+        QP query;
+        size_t idx = 0;
+
+        token = strtok(line, " ");
+        do {
+            query.point.data[idx++] = atof(token);
+        } while ((token = strtok(NULL, " ")));
+
+        query.k = k;
+        queries.emplace_back(query);
+    }
+
+    free(line);
+    fclose(qf);
+
+    return queries;
+}
+
+template<de::KVPInterface R>
+static std::vector<R> read_sosd_file(std::string &fname, size_t n) {
+    std::fstream file;
+    file.open(fname, std::ios::in | std::ios::binary);
+
+    std::vector<R> records(n);
+    for (size_t i=0; i<n; i++) {
+        uint64_t k;
+        file.read((char*) &(k), sizeof(uint64_t));
+        records[i].key = k;
+        records[i].value = i;
+    }
+
+    return records;
+}
+
+/*
+ * This function expects a plaintext file with each vector on its own line.
+ * There should be D dimensions (or more) for each record, separated by
+ * whitespace. The function will generate a vector of n records, each
+ * record built from the first D dimensions of the data in the file.
+ */
+template <typename R, size_t D>
+static std::vector<R> read_vector_file(std::string &fname, size_t n) {
+    std::fstream file;
+    file.open(fname, std::ios::in);
+
+    std::vector<R> records;
+    records.reserve(n);
+
+    for (size_t i=0; i<n; i++) {
+        std::string line;
+        if (!std::getline(file, line, '\n')) break;
+
+        std::stringstream line_stream(line);
+        R rec;
+        for (size_t j=0; j<D; j++) {
+            std::string dim;
+            if (!std::getline(line_stream, dim, ' ')) break;
+
+            rec.data[j] = atof(dim.c_str());
+        }
+        records.emplace_back(rec);
+    }
+
+    return records;
+}
-- 
cgit v1.2.3


From 0bb5b46ec2b64be17f6269631915e62d02e315e4 Mon Sep 17 00:00:00 2001
From: Douglas Rumbaugh <dbr4@psu.edu>
Date: Mon, 22 Apr 2024 12:27:43 -0400
Subject: Added plain BSM and MDSP BSM benchmark

---
 benchmarks/include/file_util.h | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)

(limited to 'benchmarks/include/file_util.h')

diff --git a/benchmarks/include/file_util.h b/benchmarks/include/file_util.h
index 2a3300a..ea8b42e 100644
--- a/benchmarks/include/file_util.h
+++ b/benchmarks/include/file_util.h
@@ -97,6 +97,22 @@ static std::vector<R> read_sosd_file(std::string &fname, size_t n) {
     return records;
 }
 
+template<typename K, typename V>
+static std::vector<std::pair<K, V>> read_sosd_file_pair(std::string &fname, size_t n) {
+    std::fstream file;
+    file.open(fname, std::ios::in | std::ios::binary);
+
+    std::vector<std::pair<K,V>> records(n);
+    for (size_t i=0; i<n; i++) {
+        uint64_t k;
+        file.read((char*) &(k), sizeof(uint64_t));
+        records[i].first = k;
+        records[i].second = i;
+    }
+
+    return records;
+}
+
 /*
  * This function expects a plaintext file with each vector on its own line.
  * There should be D dimensions (or more) for each record, separated by
-- 
cgit v1.2.3


From fb08023dea10c9a7673ceb2c8ef67718937717b8 Mon Sep 17 00:00:00 2001
From: Douglas Rumbaugh <dbr4@psu.edu>
Date: Tue, 23 Apr 2024 13:24:27 -0400
Subject: benchmarks/file_util: removed dependency on framework in prep for
 ALEX

---
 benchmarks/include/file_util.h | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)

(limited to 'benchmarks/include/file_util.h')

diff --git a/benchmarks/include/file_util.h b/benchmarks/include/file_util.h
index ea8b42e..8dc1b5f 100644
--- a/benchmarks/include/file_util.h
+++ b/benchmarks/include/file_util.h
@@ -8,9 +8,6 @@
 #include <cstring>
 #include <vector>
 
-#include "framework/interface/Record.h"
-#include "query/irs.h"
-
 #pragma once
 
 template <typename QP>
@@ -81,7 +78,7 @@ static std::vector<QP> read_knn_queries(std::string fname, size_t k) {
     return queries;
 }
 
-template<de::KVPInterface R>
+template<typename R>
 static std::vector<R> read_sosd_file(std::string &fname, size_t n) {
     std::fstream file;
     file.open(fname, std::ios::in | std::ios::binary);
-- 
cgit v1.2.3


From 160692f4b9f80c6eba7d18d3221fc1c3e3c3139e Mon Sep 17 00:00:00 2001
From: Douglas Rumbaugh <dbr4@psu.edu>
Date: Wed, 1 May 2024 13:31:24 -0400
Subject: Added error checks to file opening, and generalized key types

---
 benchmarks/include/file_util.h | 40 ++++++++++++++++++++++++++++++++++++----
 1 file changed, 36 insertions(+), 4 deletions(-)

(limited to 'benchmarks/include/file_util.h')

diff --git a/benchmarks/include/file_util.h b/benchmarks/include/file_util.h
index 8dc1b5f..b5b3417 100644
--- a/benchmarks/include/file_util.h
+++ b/benchmarks/include/file_util.h
@@ -15,6 +15,12 @@ static std::vector<QP> read_lookup_queries(std::string fname, double selectivity
     std::vector<QP> queries;
 
     FILE *qf = fopen(fname.c_str(), "r");
+
+    if (!qf) {
+        fprintf(stderr, "ERROR: Failed to open file %s\n", fname.c_str());
+        exit(EXIT_FAILURE);
+    }
+
     size_t start, stop;
     double sel;
     while (fscanf(qf, "%zu%zu%lf\n", &start, &stop, &sel) != EOF) {
@@ -34,6 +40,12 @@ static std::vector<QP> read_range_queries(std::string &fname, double selectivity
     std::vector<QP> queries;
 
     FILE *qf = fopen(fname.c_str(), "r");
+
+    if (!qf) {
+        fprintf(stderr, "ERROR: Failed to open file %s\n", fname.c_str());
+        exit(EXIT_FAILURE);
+    }
+
     size_t start, stop;
     double sel;
     while (fscanf(qf, "%zu%zu%lf\n", &start, &stop, &sel) != EOF) {
@@ -58,6 +70,11 @@ static std::vector<QP> read_knn_queries(std::string fname, size_t k) {
     char *line = NULL;
     size_t len = 0;
 
+    if (!qf) {
+        fprintf(stderr, "ERROR: Failed to open file %s\n", fname.c_str());
+        exit(EXIT_FAILURE);
+    }
+
     while (getline(&line, &len, qf) > 0) {
         char *token;
         QP query;
@@ -83,10 +100,15 @@ static std::vector<R> read_sosd_file(std::string &fname, size_t n) {
     std::fstream file;
     file.open(fname, std::ios::in | std::ios::binary);
 
+    if (!file.is_open()) {
+        fprintf(stderr, "ERROR: Failed to open file %s\n", fname.c_str());
+        exit(EXIT_FAILURE);
+    }
+
     std::vector<R> records(n);
     for (size_t i=0; i<n; i++) {
-        uint64_t k;
-        file.read((char*) &(k), sizeof(uint64_t));
+        decltype(R::key) k;
+        file.read((char*) &(k), sizeof(R::key));
         records[i].key = k;
         records[i].value = i;
     }
@@ -99,10 +121,15 @@ static std::vector<std::pair<K, V>> read_sosd_file_pair(std::string &fname, size
     std::fstream file;
     file.open(fname, std::ios::in | std::ios::binary);
 
+    if (!file.is_open()) {
+        fprintf(stderr, "ERROR: Failed to open file %s\n", fname.c_str());
+        exit(EXIT_FAILURE);
+    }
+
     std::vector<std::pair<K,V>> records(n);
     for (size_t i=0; i<n; i++) {
-        uint64_t k;
-        file.read((char*) &(k), sizeof(uint64_t));
+        K k;
+        file.read((char*) &(k), sizeof(K));
         records[i].first = k;
         records[i].second = i;
     }
@@ -121,6 +148,11 @@ static std::vector<R> read_vector_file(std::string &fname, size_t n) {
     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<R> records;
     records.reserve(n);
 
-- 
cgit v1.2.3


From 5636838a6e64760c291b00107657a90428a0f9e1 Mon Sep 17 00:00:00 2001
From: Douglas Rumbaugh <dbr4@psu.edu>
Date: Wed, 1 May 2024 15:59:48 -0400
Subject: File Util: fixed the reading in of undesired queries

---
 benchmarks/include/file_util.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'benchmarks/include/file_util.h')

diff --git a/benchmarks/include/file_util.h b/benchmarks/include/file_util.h
index b5b3417..ebcf17e 100644
--- a/benchmarks/include/file_util.h
+++ b/benchmarks/include/file_util.h
@@ -49,7 +49,7 @@ static std::vector<QP> read_range_queries(std::string &fname, double selectivity
     size_t start, stop;
     double sel;
     while (fscanf(qf, "%zu%zu%lf\n", &start, &stop, &sel) != EOF) {
-        if (start < stop && std::abs(sel - selectivity) < 0.1) {
+        if (start < stop && std::abs(sel - selectivity) < 0.00001) {
             QP q;
             q.lower_bound = start;
             q.upper_bound = stop;
-- 
cgit v1.2.3


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 +++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 44 insertions(+), 1 deletion(-)

(limited to 'benchmarks/include/file_util.h')

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;
+}
-- 
cgit v1.2.3


From 265610435e1164a9acc39ca02ea1139acd37c46c Mon Sep 17 00:00:00 2001
From: Douglas Rumbaugh <dbr4@psu.edu>
Date: Thu, 9 May 2024 14:10:29 -0400
Subject: Added benchmarks for BigANN

---
 benchmarks/include/file_util.h | 76 ++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 76 insertions(+)

(limited to 'benchmarks/include/file_util.h')

diff --git a/benchmarks/include/file_util.h b/benchmarks/include/file_util.h
index 586b44f..41eb18c 100644
--- a/benchmarks/include/file_util.h
+++ b/benchmarks/include/file_util.h
@@ -80,6 +80,46 @@ static std::vector<QP> read_range_queries(std::string &fname, double selectivity
     return queries;
 }
 
+
+template <typename QP>
+static std::vector<QP> read_binary_knn_queries(std::string fname, size_t k, size_t n) {
+    std::vector<QP> queries;
+    queries.reserve(n);
+
+    std::fstream file;
+    file.open(fname, std::ios::in | std::ios::binary);
+
+    if (!file.is_open()) {
+        fprintf(stderr, "ERROR: Failed to open file %s\n", fname.c_str());
+        exit(EXIT_FAILURE);
+    }
+
+
+    int32_t dim;
+    int32_t cnt;
+
+    file.read((char*) &(cnt), sizeof(cnt));
+    file.read((char*) &(dim), sizeof(dim));
+
+    if (n > cnt) {
+        n = cnt;
+    }
+
+    for (size_t i=0; i<n; i++) {
+        QP query;
+        for (size_t j=0; j<dim; j++) {
+            uint64_t val;
+            file.read((char*) &(val), sizeof(uint64_t));
+            query.point.data[j] = val;
+        }
+        query.k = k;
+        queries.push_back(query);
+    }
+
+    return queries;
+}
+
+
 template <typename QP>
 static std::vector<QP> read_knn_queries(std::string fname, size_t k) {
     std::vector<QP> queries;
@@ -192,6 +232,42 @@ static std::vector<R> read_vector_file(std::string &fname, size_t n) {
     return records;
 }
 
+template <typename R>
+static std::vector<R> read_binary_vector_file(std::string &fname, size_t n) {
+    std::fstream file;
+    file.open(fname, std::ios::in | std::ios::binary);
+
+    if (!file.is_open()) {
+        fprintf(stderr, "ERROR: Failed to open file %s\n", fname.c_str());
+        exit(EXIT_FAILURE);
+    }
+
+    std::vector<R> records;
+    records.reserve(n);
+
+    int32_t dim;
+    int32_t cnt;
+
+    file.read((char*) &(cnt), sizeof(cnt));
+    file.read((char*) &(dim), sizeof(dim));
+
+    if (n > cnt) {
+        n = cnt;
+    }
+
+    R rec;
+    for (size_t i=0; i<n; i++) {
+        for (size_t j=0; j<dim; j++) {
+            uint64_t val;
+            file.read((char*) &(val), sizeof(uint64_t));
+            rec.data[j] = val;
+        }
+
+        records.emplace_back(rec);
+    }
+
+    return records;
+}
 
 static std::vector<std::unique_ptr<char[]>>read_string_file(std::string fname, size_t n=10000000) {
 
-- 
cgit v1.2.3