From 37304a11b8edfef8fc627e7a3a095e77e70b07b1 Mon Sep 17 00:00:00 2001 From: Douglas Rumbaugh Date: Tue, 13 Jun 2023 13:56:17 -0400 Subject: Benchmark refactoring --- benchmarks/bench.h | 206 -------------------------------- benchmarks/include/bench.h | 67 +++++++++++ benchmarks/include/bench_utility.h | 236 +++++++++++++++++++++++++++++++++++++ benchmarks/sampling_tput.cpp | 96 ++------------- 4 files changed, 312 insertions(+), 293 deletions(-) delete mode 100644 benchmarks/bench.h create mode 100644 benchmarks/include/bench.h create mode 100644 benchmarks/include/bench_utility.h (limited to 'benchmarks') diff --git a/benchmarks/bench.h b/benchmarks/bench.h deleted file mode 100644 index c1457fc..0000000 --- a/benchmarks/bench.h +++ /dev/null @@ -1,206 +0,0 @@ -#ifndef H_BENCH -#define H_BENCH -#include "framework/DynamicExtension.h" -#include "shard/WSS.h" - -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include - -typedef uint64_t key_type; -typedef uint32_t value_type; -typedef uint64_t weight_type; - -typedef de::WeightedRecord WRec; -typedef de::DynamicExtension, de::WSSQuery> ExtendedWSS; - -static gsl_rng *g_rng; -static std::set *g_to_delete; -static bool g_osm_data; - -static key_type g_min_key = UINT64_MAX; -static key_type g_max_key = 0; - -static size_t g_max_record_cnt = 0; -static size_t g_reccnt = 0; - -static constexpr unsigned int DEFAULT_SEED = 0; - -static unsigned int get_random_seed() -{ - unsigned int seed = 0; - std::fstream urandom; - urandom.open("/dev/urandom", std::ios::in|std::ios::binary); - urandom.read((char *) &seed, sizeof(seed)); - urandom.close(); - - return seed; -} - -static key_type osm_to_key(const char *key_field) { - double tmp_key = (atof(key_field) + 180) * 10e6; - return (key_type) tmp_key; -} - -static void init_bench_rng(unsigned int seed, const gsl_rng_type *type) -{ - g_rng = gsl_rng_alloc(type); - gsl_rng_set(g_rng, seed); -} - -static void init_bench_env(size_t max_reccnt, bool random_seed, bool osm_correction=true) -{ - unsigned int seed = (random_seed) ? get_random_seed() : DEFAULT_SEED; - init_bench_rng(seed, gsl_rng_mt19937); - g_to_delete = new std::set(); - g_osm_data = osm_correction; - g_max_record_cnt = max_reccnt; - g_reccnt = 0; -} - -static void delete_bench_env() -{ - gsl_rng_free(g_rng); - delete g_to_delete; -} - -static bool next_record(std::fstream *file, WRec *record) -{ - if (g_reccnt >= g_max_record_cnt) return false; - - std::string line; - if (std::getline(*file, line, '\n')) { - std::stringstream line_stream(line); - std::string key_field; - std::string value_field; - std::string weight_field; - - std::getline(line_stream, value_field, '\t'); - std::getline(line_stream, key_field, '\t'); - std::getline(line_stream, weight_field, '\t'); - - record->key = (g_osm_data) ? osm_to_key(key_field.c_str()) : atol(key_field.c_str()); - record->value = atol(value_field.c_str()); - record->weight = atof(weight_field.c_str()); - - if (record->key < g_min_key) g_min_key = record->key; - - if (record->key > g_max_key) g_max_key = record->key; - - g_reccnt++; - - return true; - } - - return false; -} - -static bool build_insert_vec(std::fstream *file, std::vector &vec, size_t n) { - vec.clear(); - for (size_t i=0; i= 1) fprintf(stderr, "\n"); -} - -static bool warmup(std::fstream *file, ExtendedWSS *extended_wss, size_t count, double delete_prop, bool progress=true) -{ - size_t del_buf_size = 10000; - size_t delete_idx = del_buf_size; - - std::vector delbuf; - std::set deleted_keys; - - de::wss_query_parms parms; - parms.rng = g_rng; - parms.sample_size = del_buf_size; - - size_t inserted = 0; - - double last_percent = 0; - for (size_t i=0; iinsert(rec); - - if (delete_prop > 0 && i > extended_wss->get_buffer_capacity() && delete_idx >= delbuf.size()) { - delbuf = extended_wss->query(&parms); - delete_idx = 0; - deleted_keys.clear(); - } - - if (delete_prop > 0 && i > extended_wss->get_buffer_capacity() && gsl_rng_uniform(g_rng) < delete_prop && delete_idx < delbuf.size()) { - auto rec = delbuf[delete_idx]; - delete_idx++; - - if (deleted_keys.find(rec) == deleted_keys.end()) { - extended_wss->erase(rec); - deleted_keys.insert(rec); - } - } - - if (progress && ((double) i / (double) count) - last_percent > .01) { - progress_update((double) i / (double) count, "warming up:"); - last_percent = (double) i / (double) count; - } - } - - if (progress) { - progress_update(1, "warming up:"); - } - - return true; -} - - -static void reset_de_perf_metrics() { - - /* - * rejection counters are zeroed automatically by the - * sampling function itself. - */ - - RESET_IO_CNT(); -} - -#endif // H_BENCH diff --git a/benchmarks/include/bench.h b/benchmarks/include/bench.h new file mode 100644 index 0000000..d82da48 --- /dev/null +++ b/benchmarks/include/bench.h @@ -0,0 +1,67 @@ +#include "bench_utility.h" + +template +static bool insert_tput_bench(DE &de_index, std::fstream &file, size_t insert_cnt, + double delete_prop, std::vector &to_delete) { + + size_t delete_cnt = insert_cnt * delete_prop; + + size_t applied_deletes = 0; + size_t applied_inserts = 0; + + std::vector insert_vec; + std::vector delete_vec; + insert_vec.reserve(BATCH); + delete_vec.reserve(BATCH*delete_prop); + + size_t delete_idx = 0; + + bool continue_benchmark = true; + + size_t total_time = 0; + + while (applied_inserts < insert_cnt && continue_benchmark) { + continue_benchmark = build_insert_vec(file, insert_vec, BATCH, delete_prop, to_delete); + if (applied_deletes < delete_cnt) { + build_delete_vec(to_delete, delete_vec, BATCH*delete_prop); + delete_idx = 0; + } + + if (insert_vec.size() == 0) { + break; + } + + if constexpr (PROGRESS) { + progress_update((double) applied_inserts / (double) insert_cnt, "inserting:"); + } + + auto insert_start = std::chrono::high_resolution_clock::now(); + for (size_t i=0; i(insert_stop - insert_start).count(); + } + + if constexpr (PROGRESS) { + progress_update(1.0, "inserting:"); + } + + size_t throughput = (((double) (applied_inserts + applied_deletes) / (double) total_time) * 1e9); + + fprintf(stdout, "\n%ld\n", throughput); + reset_de_perf_metrics(); + + return continue_benchmark; +} + + diff --git a/benchmarks/include/bench_utility.h b/benchmarks/include/bench_utility.h new file mode 100644 index 0000000..78d6415 --- /dev/null +++ b/benchmarks/include/bench_utility.h @@ -0,0 +1,236 @@ +#ifndef H_BENCH +#define H_BENCH +#include "framework/DynamicExtension.h" +#include "shard/WSS.h" +#include "shard/MemISAM.h" +#include "shard/PGM.h" +#include "shard/TrieSpline.h" +#include "shard/WIRS.h" + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +typedef uint64_t key_type; +typedef uint32_t value_type; +typedef uint64_t weight_type; + +typedef de::WeightedRecord WRec; +typedef de::Record Rec; + +typedef de::DynamicExtension, de::WSSQuery> ExtendedWSS; +typedef de::DynamicExtension, de::TrieSplineRangeQuery> ExtendedTS; +typedef de::DynamicExtension, de::PGMRangeQuery> ExtendedPGM; + +static gsl_rng *g_rng; +static std::set *g_to_delete; +static bool g_osm_data; + +static key_type g_min_key = UINT64_MAX; +static key_type g_max_key = 0; + +static size_t g_max_record_cnt = 0; +static size_t g_reccnt = 0; + +static constexpr unsigned int DEFAULT_SEED = 0; + +static unsigned int get_random_seed() +{ + unsigned int seed = 0; + std::fstream urandom; + urandom.open("/dev/urandom", std::ios::in|std::ios::binary); + urandom.read((char *) &seed, sizeof(seed)); + urandom.close(); + + return seed; +} + +static key_type osm_to_key(const char *key_field) { + double tmp_key = (atof(key_field) + 180) * 10e6; + return (key_type) tmp_key; +} + +static void init_bench_rng(unsigned int seed, const gsl_rng_type *type) +{ + g_rng = gsl_rng_alloc(type); + gsl_rng_set(g_rng, seed); +} + +static void init_bench_env(size_t max_reccnt, bool random_seed, bool osm_correction=true) +{ + unsigned int seed = (random_seed) ? get_random_seed() : DEFAULT_SEED; + init_bench_rng(seed, gsl_rng_mt19937); + g_to_delete = new std::set(); + g_osm_data = osm_correction; + g_max_record_cnt = max_reccnt; + g_reccnt = 0; +} + +static void delete_bench_env() +{ + gsl_rng_free(g_rng); + delete g_to_delete; +} + +template +static bool next_record(std::fstream &file, R &record) +{ + if (g_reccnt >= g_max_record_cnt) return false; + + std::string line; + if (std::getline(file, line, '\n')) { + std::stringstream line_stream(line); + std::string key_field; + std::string value_field; + std::string weight_field; + + std::getline(line_stream, value_field, '\t'); + std::getline(line_stream, key_field, '\t'); + std::getline(line_stream, weight_field, '\t'); + + record.key = (g_osm_data) ? osm_to_key(key_field.c_str()) : atol(key_field.c_str()); + record.value = atol(value_field.c_str()); + + if constexpr (de::WeightedRecordInterface) { + record.weight = atof(weight_field.c_str()); + } + + if (record.key < g_min_key) g_min_key = record.key; + + if (record.key > g_max_key) g_max_key = record.key; + + g_reccnt++; + + return true; + } + + return false; +} + +template +static bool build_insert_vec(std::fstream &file, std::vector &vec, size_t n, + double delete_prop, std::vector &to_delete) { + vec.clear(); + for (size_t i=0; i +static bool build_delete_vec(std::vector &to_delete, std::vector &vec, size_t n) { + vec.clear(); + + size_t cnt = 0; + while (cnt < n) { + if (to_delete.size() == 0) { + return false; + } + + auto i = gsl_rng_uniform_int(g_rng, to_delete.size()); + vec.emplace_back(to_delete[i]); + to_delete.erase(to_delete.begin() + i); + } +td: + return true; +} + +/* + * helper routines for displaying progress bars to stderr + */ +static const char *g_prog_bar = "======================================================================"; +static const size_t g_prog_width = 50; + +static void progress_update(double percentage, std::string prompt) { + int val = (int) (percentage * 100); + int lpad = (int) (percentage * g_prog_width); + int rpad = (int) (g_prog_width - lpad); + fprintf(stderr, "\r(%3d%%) %20s [%.*s%*s]", val, prompt.c_str(), lpad, g_prog_bar, rpad, ""); + fflush(stderr); + + if (percentage >= 1) fprintf(stderr, "\n"); +} + +template +static bool warmup(std::fstream &file, DE &extended_index, size_t count, + double delete_prop, std::vector to_delete, bool progress=true) { + size_t batch = std::min(.1 * count, 25000.0); + + std::vector insert_vec; + std::vector delete_vec; + insert_vec.reserve(batch); + delete_vec.reserve(batch*delete_prop); + + size_t inserted = 0; + size_t delete_idx = 0; + + double last_percent = 0; + while (inserted < count) { + // Build vector of records to insert and potentially delete + auto continue_warmup = build_insert_vec(file, insert_vec, batch, delete_prop, to_delete); + if (inserted > batch) { + build_delete_vec(to_delete, delete_vec, batch*delete_prop); + delete_idx = 0; + } + + for (size_t i=0; i delbuf; - - std::set deleted; - - size_t applied_deletes = 0; - size_t applied_inserts = 0; - - std::vector insert_vec; - insert_vec.reserve(g_insert_batch_size); - bool continue_benchmark = true; - - size_t total_time = 0; - - de::wss_query_parms parms; - parms.rng = g_rng; - parms.sample_size = delete_batch_size; - - while (applied_inserts < insert_cnt && continue_benchmark) { - continue_benchmark = build_insert_vec(file, insert_vec, g_insert_batch_size); - - if (insert_vec.size() == 0) { - break; - } - - // if we've fully processed the delete vector, sample a new - // set of records to delete. - if (delete_idx >= delbuf.size()) { - delbuf = tree->query(&parms); - deleted.clear(); - delete_idx = 0; - } - - progress_update((double) applied_inserts / (double) insert_cnt, "inserting:"); - size_t local_inserted = 0; - size_t local_deleted = 0; - - auto insert_start = std::chrono::high_resolution_clock::now(); - for (size_t i=0; ierase(rec); - deleted.insert(rec); - local_deleted++; - } - } - - // insert the record; - tree->insert(insert_vec[i]); - local_inserted++; - } - auto insert_stop = std::chrono::high_resolution_clock::now(); - - applied_deletes += local_deleted; - applied_inserts += local_inserted; - total_time += std::chrono::duration_cast(insert_stop - insert_start).count(); - } - - progress_update(1.0, "inserting:"); - size_t throughput = (((double) (applied_inserts + applied_deletes) / (double) total_time) * 1e9); - - fprintf(stdout, "\n%ld\n", throughput); - reset_de_perf_metrics(); - - return continue_benchmark; -} - - -static void sample_benchmark(ExtendedWSS *tree, size_t k, size_t trial_cnt) +static void sample_benchmark(ExtendedWSS &de_wss, size_t k, size_t trial_cnt) { char progbuf[25]; sprintf(progbuf, "sampling (%ld):", k); @@ -101,7 +21,7 @@ static void sample_benchmark(ExtendedWSS *tree, size_t k, size_t trial_cnt) progress_update((double) (i * batch_size) / (double) trial_cnt, progbuf); auto start = std::chrono::high_resolution_clock::now(); for (int j=0; j < batch_size; j++) { - auto res = tree->query(&parms); + auto res = de_wss.query(&parms); total_samples += res.size(); } auto stop = std::chrono::high_resolution_clock::now(); @@ -137,20 +57,22 @@ int main(int argc, char **argv) init_bench_env(record_count, true, use_osm); - auto sampling_lsm = ExtendedWSS(buffer_cap, scale_factor, max_delete_prop); + auto de_wss = ExtendedWSS(buffer_cap, scale_factor, max_delete_prop); std::fstream datafile; datafile.open(filename, std::ios::in); + std::vector to_delete; + // warm up the tree with initial_insertions number of initially inserted // records size_t warmup_cnt = insert_batch * record_count; - warmup(&datafile, &sampling_lsm, warmup_cnt, delete_prop); + warmup(datafile, de_wss, warmup_cnt, delete_prop, to_delete); size_t insert_cnt = record_count - warmup_cnt; - insert_benchmark(&sampling_lsm, &datafile, insert_cnt, delete_prop); - sample_benchmark(&sampling_lsm, 1000, 10000); + insert_tput_bench(de_wss, datafile, insert_cnt, delete_prop, to_delete); + sample_benchmark(de_wss, 1000, 10000); delete_bench_env(); fflush(stdout); -- cgit v1.2.3