You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
rocksdb/table/cuckoo_table_builder.cc

517 lines
20 KiB

// Copyright (c) 2011-present, Facebook, Inc. All rights reserved.
// This source code is licensed under both the GPLv2 (found in the
// COPYING file in the root directory) and Apache 2.0 License
// (found in the LICENSE.Apache file in the root directory).
#ifndef ROCKSDB_LITE
#include "table/cuckoo_table_builder.h"
#include <assert.h>
#include <algorithm>
#include <limits>
#include <string>
#include <vector>
#include "db/dbformat.h"
#include "rocksdb/env.h"
#include "rocksdb/table.h"
#include "table/block_builder.h"
Improve Cuckoo Table Reader performance. Inlined hash function and number of buckets a power of two. Summary: Use inlined hash functions instead of function pointer. Make number of buckets a power of two and use bitwise and instead of mod. After these changes, we get almost 50% improvement in performance. Results: With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.231us (4.3 Mqps) with batch size of 0 Time taken per op is 0.229us (4.4 Mqps) with batch size of 0 Time taken per op is 0.185us (5.4 Mqps) with batch size of 0 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.108us (9.3 Mqps) with batch size of 10 Time taken per op is 0.100us (10.0 Mqps) with batch size of 10 Time taken per op is 0.103us (9.7 Mqps) with batch size of 10 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.101us (9.9 Mqps) with batch size of 25 Time taken per op is 0.098us (10.2 Mqps) with batch size of 25 Time taken per op is 0.097us (10.3 Mqps) with batch size of 25 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.100us (10.0 Mqps) with batch size of 50 Time taken per op is 0.097us (10.3 Mqps) with batch size of 50 Time taken per op is 0.097us (10.3 Mqps) with batch size of 50 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.102us (9.8 Mqps) with batch size of 100 Time taken per op is 0.098us (10.2 Mqps) with batch size of 100 Time taken per op is 0.115us (8.7 Mqps) with batch size of 100 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.201us (5.0 Mqps) with batch size of 0 Time taken per op is 0.155us (6.5 Mqps) with batch size of 0 Time taken per op is 0.152us (6.6 Mqps) with batch size of 0 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.089us (11.3 Mqps) with batch size of 10 Time taken per op is 0.084us (11.9 Mqps) with batch size of 10 Time taken per op is 0.086us (11.6 Mqps) with batch size of 10 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.087us (11.5 Mqps) with batch size of 25 Time taken per op is 0.085us (11.7 Mqps) with batch size of 25 Time taken per op is 0.093us (10.8 Mqps) with batch size of 25 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.094us (10.6 Mqps) with batch size of 50 Time taken per op is 0.094us (10.7 Mqps) with batch size of 50 Time taken per op is 0.093us (10.8 Mqps) with batch size of 50 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.092us (10.9 Mqps) with batch size of 100 Time taken per op is 0.089us (11.2 Mqps) with batch size of 100 Time taken per op is 0.088us (11.3 Mqps) with batch size of 100 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.154us (6.5 Mqps) with batch size of 0 Time taken per op is 0.168us (6.0 Mqps) with batch size of 0 Time taken per op is 0.190us (5.3 Mqps) with batch size of 0 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.081us (12.4 Mqps) with batch size of 10 Time taken per op is 0.077us (13.0 Mqps) with batch size of 10 Time taken per op is 0.083us (12.1 Mqps) with batch size of 10 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.077us (13.0 Mqps) with batch size of 25 Time taken per op is 0.073us (13.7 Mqps) with batch size of 25 Time taken per op is 0.073us (13.7 Mqps) with batch size of 25 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.076us (13.1 Mqps) with batch size of 50 Time taken per op is 0.072us (13.8 Mqps) with batch size of 50 Time taken per op is 0.072us (13.8 Mqps) with batch size of 50 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.077us (13.0 Mqps) with batch size of 100 Time taken per op is 0.074us (13.6 Mqps) with batch size of 100 Time taken per op is 0.073us (13.6 Mqps) with batch size of 100 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.190us (5.3 Mqps) with batch size of 0 Time taken per op is 0.186us (5.4 Mqps) with batch size of 0 Time taken per op is 0.184us (5.4 Mqps) with batch size of 0 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.079us (12.7 Mqps) with batch size of 10 Time taken per op is 0.070us (14.2 Mqps) with batch size of 10 Time taken per op is 0.072us (14.0 Mqps) with batch size of 10 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.080us (12.5 Mqps) with batch size of 25 Time taken per op is 0.072us (14.0 Mqps) with batch size of 25 Time taken per op is 0.071us (14.1 Mqps) with batch size of 25 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.082us (12.1 Mqps) with batch size of 50 Time taken per op is 0.071us (14.1 Mqps) with batch size of 50 Time taken per op is 0.073us (13.6 Mqps) with batch size of 50 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.080us (12.5 Mqps) with batch size of 100 Time taken per op is 0.077us (13.0 Mqps) with batch size of 100 Time taken per op is 0.078us (12.8 Mqps) with batch size of 100 Test Plan: make check all make valgrind_check make asan_check Reviewers: sdong, ljin Reviewed By: ljin Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D22539
10 years ago
#include "table/cuckoo_table_factory.h"
#include "table/format.h"
#include "table/meta_blocks.h"
#include "util/autovector.h"
#include "util/file_reader_writer.h"
#include "util/random.h"
#include "util/string_util.h"
namespace rocksdb {
const std::string CuckooTablePropertyNames::kEmptyKey =
"rocksdb.cuckoo.bucket.empty.key";
const std::string CuckooTablePropertyNames::kNumHashFunc =
"rocksdb.cuckoo.hash.num";
const std::string CuckooTablePropertyNames::kHashTableSize =
"rocksdb.cuckoo.hash.size";
const std::string CuckooTablePropertyNames::kValueLength =
"rocksdb.cuckoo.value.length";
const std::string CuckooTablePropertyNames::kIsLastLevel =
"rocksdb.cuckoo.file.islastlevel";
const std::string CuckooTablePropertyNames::kCuckooBlockSize =
"rocksdb.cuckoo.hash.cuckooblocksize";
CuckooTable: add one option to allow identity function for the first hash function Summary: MurmurHash becomes expensive when we do millions Get() a second in one thread. Add this option to allow the first hash function to use identity function as hash function. It results in QPS increase from 3.7M/s to ~4.3M/s. I did not observe improvement for end to end RocksDB performance. This may be caused by other bottlenecks that I will address in a separate diff. Test Plan: ``` [ljin@dev1964 rocksdb] ./cuckoo_table_reader_test --enable_perf --file_dir=/dev/shm --write --identity_as_first_hash=0 ==== Test CuckooReaderTest.WhenKeyExists ==== Test CuckooReaderTest.WhenKeyExistsWithUint64Comparator ==== Test CuckooReaderTest.CheckIterator ==== Test CuckooReaderTest.CheckIteratorUint64 ==== Test CuckooReaderTest.WhenKeyNotFound ==== Test CuckooReaderTest.TestReadPerformance With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.272us (3.7 Mqps) with batch size of 0, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.138us (7.2 Mqps) with batch size of 10, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.142us (7.1 Mqps) with batch size of 25, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.142us (7.0 Mqps) with batch size of 50, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.144us (6.9 Mqps) with batch size of 100, # of found keys 125829120 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.201us (5.0 Mqps) with batch size of 0, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.121us (8.3 Mqps) with batch size of 10, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.123us (8.1 Mqps) with batch size of 25, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.121us (8.3 Mqps) with batch size of 50, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.112us (8.9 Mqps) with batch size of 100, # of found keys 104857600 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.251us (4.0 Mqps) with batch size of 0, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.107us (9.4 Mqps) with batch size of 10, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.099us (10.1 Mqps) with batch size of 25, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.100us (10.0 Mqps) with batch size of 50, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.116us (8.6 Mqps) with batch size of 100, # of found keys 83886080 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.189us (5.3 Mqps) with batch size of 0, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.095us (10.5 Mqps) with batch size of 10, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.096us (10.4 Mqps) with batch size of 25, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.098us (10.2 Mqps) with batch size of 50, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.105us (9.5 Mqps) with batch size of 100, # of found keys 73400320 [ljin@dev1964 rocksdb] ./cuckoo_table_reader_test --enable_perf --file_dir=/dev/shm --write --identity_as_first_hash=1 ==== Test CuckooReaderTest.WhenKeyExists ==== Test CuckooReaderTest.WhenKeyExistsWithUint64Comparator ==== Test CuckooReaderTest.CheckIterator ==== Test CuckooReaderTest.CheckIteratorUint64 ==== Test CuckooReaderTest.WhenKeyNotFound ==== Test CuckooReaderTest.TestReadPerformance With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.230us (4.3 Mqps) with batch size of 0, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.086us (11.7 Mqps) with batch size of 10, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.088us (11.3 Mqps) with batch size of 25, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.083us (12.1 Mqps) with batch size of 50, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.083us (12.1 Mqps) with batch size of 100, # of found keys 125829120 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.159us (6.3 Mqps) with batch size of 0, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.078us (12.8 Mqps) with batch size of 10, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.080us (12.6 Mqps) with batch size of 25, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.080us (12.5 Mqps) with batch size of 50, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.082us (12.2 Mqps) with batch size of 100, # of found keys 104857600 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.154us (6.5 Mqps) with batch size of 0, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.077us (13.0 Mqps) with batch size of 10, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.077us (12.9 Mqps) with batch size of 25, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.078us (12.8 Mqps) with batch size of 50, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.079us (12.6 Mqps) with batch size of 100, # of found keys 83886080 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.218us (4.6 Mqps) with batch size of 0, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.083us (12.0 Mqps) with batch size of 10, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.085us (11.7 Mqps) with batch size of 25, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.086us (11.6 Mqps) with batch size of 50, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.078us (12.8 Mqps) with batch size of 100, # of found keys 73400320 ``` Reviewers: sdong, igor, yhchiang Reviewed By: igor Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D23451
10 years ago
const std::string CuckooTablePropertyNames::kIdentityAsFirstHash =
"rocksdb.cuckoo.hash.identityfirst";
const std::string CuckooTablePropertyNames::kUseModuleHash =
"rocksdb.cuckoo.hash.usemodule";
const std::string CuckooTablePropertyNames::kUserKeyLength =
"rocksdb.cuckoo.hash.userkeylength";
// Obtained by running echo rocksdb.table.cuckoo | sha1sum
extern const uint64_t kCuckooTableMagicNumber = 0x926789d0c5f17873ull;
CuckooTableBuilder::CuckooTableBuilder(
WritableFileWriter* file, double max_hash_table_ratio,
uint32_t max_num_hash_table, uint32_t max_search_depth,
const Comparator* user_comparator, uint32_t cuckoo_block_size,
bool use_module_hash, bool identity_as_first_hash,
uint64_t (*get_slice_hash)(const Slice&, uint32_t, uint64_t),
uint32_t column_family_id, const std::string& column_family_name)
: num_hash_func_(2),
file_(file),
Improve Cuckoo Table Reader performance. Inlined hash function and number of buckets a power of two. Summary: Use inlined hash functions instead of function pointer. Make number of buckets a power of two and use bitwise and instead of mod. After these changes, we get almost 50% improvement in performance. Results: With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.231us (4.3 Mqps) with batch size of 0 Time taken per op is 0.229us (4.4 Mqps) with batch size of 0 Time taken per op is 0.185us (5.4 Mqps) with batch size of 0 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.108us (9.3 Mqps) with batch size of 10 Time taken per op is 0.100us (10.0 Mqps) with batch size of 10 Time taken per op is 0.103us (9.7 Mqps) with batch size of 10 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.101us (9.9 Mqps) with batch size of 25 Time taken per op is 0.098us (10.2 Mqps) with batch size of 25 Time taken per op is 0.097us (10.3 Mqps) with batch size of 25 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.100us (10.0 Mqps) with batch size of 50 Time taken per op is 0.097us (10.3 Mqps) with batch size of 50 Time taken per op is 0.097us (10.3 Mqps) with batch size of 50 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.102us (9.8 Mqps) with batch size of 100 Time taken per op is 0.098us (10.2 Mqps) with batch size of 100 Time taken per op is 0.115us (8.7 Mqps) with batch size of 100 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.201us (5.0 Mqps) with batch size of 0 Time taken per op is 0.155us (6.5 Mqps) with batch size of 0 Time taken per op is 0.152us (6.6 Mqps) with batch size of 0 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.089us (11.3 Mqps) with batch size of 10 Time taken per op is 0.084us (11.9 Mqps) with batch size of 10 Time taken per op is 0.086us (11.6 Mqps) with batch size of 10 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.087us (11.5 Mqps) with batch size of 25 Time taken per op is 0.085us (11.7 Mqps) with batch size of 25 Time taken per op is 0.093us (10.8 Mqps) with batch size of 25 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.094us (10.6 Mqps) with batch size of 50 Time taken per op is 0.094us (10.7 Mqps) with batch size of 50 Time taken per op is 0.093us (10.8 Mqps) with batch size of 50 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.092us (10.9 Mqps) with batch size of 100 Time taken per op is 0.089us (11.2 Mqps) with batch size of 100 Time taken per op is 0.088us (11.3 Mqps) with batch size of 100 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.154us (6.5 Mqps) with batch size of 0 Time taken per op is 0.168us (6.0 Mqps) with batch size of 0 Time taken per op is 0.190us (5.3 Mqps) with batch size of 0 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.081us (12.4 Mqps) with batch size of 10 Time taken per op is 0.077us (13.0 Mqps) with batch size of 10 Time taken per op is 0.083us (12.1 Mqps) with batch size of 10 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.077us (13.0 Mqps) with batch size of 25 Time taken per op is 0.073us (13.7 Mqps) with batch size of 25 Time taken per op is 0.073us (13.7 Mqps) with batch size of 25 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.076us (13.1 Mqps) with batch size of 50 Time taken per op is 0.072us (13.8 Mqps) with batch size of 50 Time taken per op is 0.072us (13.8 Mqps) with batch size of 50 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.077us (13.0 Mqps) with batch size of 100 Time taken per op is 0.074us (13.6 Mqps) with batch size of 100 Time taken per op is 0.073us (13.6 Mqps) with batch size of 100 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.190us (5.3 Mqps) with batch size of 0 Time taken per op is 0.186us (5.4 Mqps) with batch size of 0 Time taken per op is 0.184us (5.4 Mqps) with batch size of 0 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.079us (12.7 Mqps) with batch size of 10 Time taken per op is 0.070us (14.2 Mqps) with batch size of 10 Time taken per op is 0.072us (14.0 Mqps) with batch size of 10 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.080us (12.5 Mqps) with batch size of 25 Time taken per op is 0.072us (14.0 Mqps) with batch size of 25 Time taken per op is 0.071us (14.1 Mqps) with batch size of 25 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.082us (12.1 Mqps) with batch size of 50 Time taken per op is 0.071us (14.1 Mqps) with batch size of 50 Time taken per op is 0.073us (13.6 Mqps) with batch size of 50 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.080us (12.5 Mqps) with batch size of 100 Time taken per op is 0.077us (13.0 Mqps) with batch size of 100 Time taken per op is 0.078us (12.8 Mqps) with batch size of 100 Test Plan: make check all make valgrind_check make asan_check Reviewers: sdong, ljin Reviewed By: ljin Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D22539
10 years ago
max_hash_table_ratio_(max_hash_table_ratio),
max_num_hash_func_(max_num_hash_table),
max_search_depth_(max_search_depth),
cuckoo_block_size_(std::max(1U, cuckoo_block_size)),
hash_table_size_(use_module_hash ? 0 : 2),
is_last_level_file_(false),
has_seen_first_key_(false),
has_seen_first_value_(false),
key_size_(0),
value_size_(0),
num_entries_(0),
num_values_(0),
ucomp_(user_comparator),
use_module_hash_(use_module_hash),
CuckooTable: add one option to allow identity function for the first hash function Summary: MurmurHash becomes expensive when we do millions Get() a second in one thread. Add this option to allow the first hash function to use identity function as hash function. It results in QPS increase from 3.7M/s to ~4.3M/s. I did not observe improvement for end to end RocksDB performance. This may be caused by other bottlenecks that I will address in a separate diff. Test Plan: ``` [ljin@dev1964 rocksdb] ./cuckoo_table_reader_test --enable_perf --file_dir=/dev/shm --write --identity_as_first_hash=0 ==== Test CuckooReaderTest.WhenKeyExists ==== Test CuckooReaderTest.WhenKeyExistsWithUint64Comparator ==== Test CuckooReaderTest.CheckIterator ==== Test CuckooReaderTest.CheckIteratorUint64 ==== Test CuckooReaderTest.WhenKeyNotFound ==== Test CuckooReaderTest.TestReadPerformance With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.272us (3.7 Mqps) with batch size of 0, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.138us (7.2 Mqps) with batch size of 10, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.142us (7.1 Mqps) with batch size of 25, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.142us (7.0 Mqps) with batch size of 50, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.144us (6.9 Mqps) with batch size of 100, # of found keys 125829120 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.201us (5.0 Mqps) with batch size of 0, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.121us (8.3 Mqps) with batch size of 10, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.123us (8.1 Mqps) with batch size of 25, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.121us (8.3 Mqps) with batch size of 50, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.112us (8.9 Mqps) with batch size of 100, # of found keys 104857600 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.251us (4.0 Mqps) with batch size of 0, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.107us (9.4 Mqps) with batch size of 10, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.099us (10.1 Mqps) with batch size of 25, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.100us (10.0 Mqps) with batch size of 50, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.116us (8.6 Mqps) with batch size of 100, # of found keys 83886080 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.189us (5.3 Mqps) with batch size of 0, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.095us (10.5 Mqps) with batch size of 10, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.096us (10.4 Mqps) with batch size of 25, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.098us (10.2 Mqps) with batch size of 50, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.105us (9.5 Mqps) with batch size of 100, # of found keys 73400320 [ljin@dev1964 rocksdb] ./cuckoo_table_reader_test --enable_perf --file_dir=/dev/shm --write --identity_as_first_hash=1 ==== Test CuckooReaderTest.WhenKeyExists ==== Test CuckooReaderTest.WhenKeyExistsWithUint64Comparator ==== Test CuckooReaderTest.CheckIterator ==== Test CuckooReaderTest.CheckIteratorUint64 ==== Test CuckooReaderTest.WhenKeyNotFound ==== Test CuckooReaderTest.TestReadPerformance With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.230us (4.3 Mqps) with batch size of 0, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.086us (11.7 Mqps) with batch size of 10, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.088us (11.3 Mqps) with batch size of 25, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.083us (12.1 Mqps) with batch size of 50, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.083us (12.1 Mqps) with batch size of 100, # of found keys 125829120 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.159us (6.3 Mqps) with batch size of 0, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.078us (12.8 Mqps) with batch size of 10, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.080us (12.6 Mqps) with batch size of 25, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.080us (12.5 Mqps) with batch size of 50, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.082us (12.2 Mqps) with batch size of 100, # of found keys 104857600 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.154us (6.5 Mqps) with batch size of 0, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.077us (13.0 Mqps) with batch size of 10, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.077us (12.9 Mqps) with batch size of 25, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.078us (12.8 Mqps) with batch size of 50, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.079us (12.6 Mqps) with batch size of 100, # of found keys 83886080 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.218us (4.6 Mqps) with batch size of 0, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.083us (12.0 Mqps) with batch size of 10, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.085us (11.7 Mqps) with batch size of 25, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.086us (11.6 Mqps) with batch size of 50, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.078us (12.8 Mqps) with batch size of 100, # of found keys 73400320 ``` Reviewers: sdong, igor, yhchiang Reviewed By: igor Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D23451
10 years ago
identity_as_first_hash_(identity_as_first_hash),
get_slice_hash_(get_slice_hash),
closed_(false) {
// Data is in a huge block.
properties_.num_data_blocks = 1;
properties_.index_size = 0;
properties_.filter_size = 0;
properties_.column_family_id = column_family_id;
properties_.column_family_name = column_family_name;
}
void CuckooTableBuilder::Add(const Slice& key, const Slice& value) {
if (num_entries_ >= kMaxVectorIdx - 1) {
status_ = Status::NotSupported("Number of keys in a file must be < 2^32-1");
return;
}
ParsedInternalKey ikey;
if (!ParseInternalKey(key, &ikey)) {
status_ = Status::Corruption("Unable to parse key into inernal key.");
return;
}
if (ikey.type != kTypeDeletion && ikey.type != kTypeValue) {
status_ = Status::NotSupported("Unsupported key type " +
ToString(ikey.type));
return;
}
// Determine if we can ignore the sequence number and value type from
// internal keys by looking at sequence number from first key. We assume
// that if first key has a zero sequence number, then all the remaining
// keys will have zero seq. no.
if (!has_seen_first_key_) {
is_last_level_file_ = ikey.sequence == 0;
has_seen_first_key_ = true;
smallest_user_key_.assign(ikey.user_key.data(), ikey.user_key.size());
largest_user_key_.assign(ikey.user_key.data(), ikey.user_key.size());
key_size_ = is_last_level_file_ ? ikey.user_key.size() : key.size();
}
if (key_size_ != (is_last_level_file_ ? ikey.user_key.size() : key.size())) {
status_ = Status::NotSupported("all keys have to be the same size");
return;
}
if (ikey.type == kTypeValue) {
if (!has_seen_first_value_) {
has_seen_first_value_ = true;
value_size_ = value.size();
}
if (value_size_ != value.size()) {
status_ = Status::NotSupported("all values have to be the same size");
return;
}
if (is_last_level_file_) {
kvs_.append(ikey.user_key.data(), ikey.user_key.size());
} else {
kvs_.append(key.data(), key.size());
}
kvs_.append(value.data(), value.size());
++num_values_;
} else {
if (is_last_level_file_) {
deleted_keys_.append(ikey.user_key.data(), ikey.user_key.size());
} else {
deleted_keys_.append(key.data(), key.size());
}
}
++num_entries_;
// In order to fill the empty buckets in the hash table, we identify a
// key which is not used so far (unused_user_key). We determine this by
// maintaining smallest and largest keys inserted so far in bytewise order
// and use them to find a key outside this range in Finish() operation.
// Note that this strategy is independent of user comparator used here.
if (ikey.user_key.compare(smallest_user_key_) < 0) {
smallest_user_key_.assign(ikey.user_key.data(), ikey.user_key.size());
} else if (ikey.user_key.compare(largest_user_key_) > 0) {
largest_user_key_.assign(ikey.user_key.data(), ikey.user_key.size());
}
if (!use_module_hash_) {
if (hash_table_size_ < num_entries_ / max_hash_table_ratio_) {
hash_table_size_ *= 2;
}
Improve Cuckoo Table Reader performance. Inlined hash function and number of buckets a power of two. Summary: Use inlined hash functions instead of function pointer. Make number of buckets a power of two and use bitwise and instead of mod. After these changes, we get almost 50% improvement in performance. Results: With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.231us (4.3 Mqps) with batch size of 0 Time taken per op is 0.229us (4.4 Mqps) with batch size of 0 Time taken per op is 0.185us (5.4 Mqps) with batch size of 0 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.108us (9.3 Mqps) with batch size of 10 Time taken per op is 0.100us (10.0 Mqps) with batch size of 10 Time taken per op is 0.103us (9.7 Mqps) with batch size of 10 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.101us (9.9 Mqps) with batch size of 25 Time taken per op is 0.098us (10.2 Mqps) with batch size of 25 Time taken per op is 0.097us (10.3 Mqps) with batch size of 25 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.100us (10.0 Mqps) with batch size of 50 Time taken per op is 0.097us (10.3 Mqps) with batch size of 50 Time taken per op is 0.097us (10.3 Mqps) with batch size of 50 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.102us (9.8 Mqps) with batch size of 100 Time taken per op is 0.098us (10.2 Mqps) with batch size of 100 Time taken per op is 0.115us (8.7 Mqps) with batch size of 100 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.201us (5.0 Mqps) with batch size of 0 Time taken per op is 0.155us (6.5 Mqps) with batch size of 0 Time taken per op is 0.152us (6.6 Mqps) with batch size of 0 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.089us (11.3 Mqps) with batch size of 10 Time taken per op is 0.084us (11.9 Mqps) with batch size of 10 Time taken per op is 0.086us (11.6 Mqps) with batch size of 10 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.087us (11.5 Mqps) with batch size of 25 Time taken per op is 0.085us (11.7 Mqps) with batch size of 25 Time taken per op is 0.093us (10.8 Mqps) with batch size of 25 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.094us (10.6 Mqps) with batch size of 50 Time taken per op is 0.094us (10.7 Mqps) with batch size of 50 Time taken per op is 0.093us (10.8 Mqps) with batch size of 50 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.092us (10.9 Mqps) with batch size of 100 Time taken per op is 0.089us (11.2 Mqps) with batch size of 100 Time taken per op is 0.088us (11.3 Mqps) with batch size of 100 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.154us (6.5 Mqps) with batch size of 0 Time taken per op is 0.168us (6.0 Mqps) with batch size of 0 Time taken per op is 0.190us (5.3 Mqps) with batch size of 0 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.081us (12.4 Mqps) with batch size of 10 Time taken per op is 0.077us (13.0 Mqps) with batch size of 10 Time taken per op is 0.083us (12.1 Mqps) with batch size of 10 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.077us (13.0 Mqps) with batch size of 25 Time taken per op is 0.073us (13.7 Mqps) with batch size of 25 Time taken per op is 0.073us (13.7 Mqps) with batch size of 25 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.076us (13.1 Mqps) with batch size of 50 Time taken per op is 0.072us (13.8 Mqps) with batch size of 50 Time taken per op is 0.072us (13.8 Mqps) with batch size of 50 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.077us (13.0 Mqps) with batch size of 100 Time taken per op is 0.074us (13.6 Mqps) with batch size of 100 Time taken per op is 0.073us (13.6 Mqps) with batch size of 100 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.190us (5.3 Mqps) with batch size of 0 Time taken per op is 0.186us (5.4 Mqps) with batch size of 0 Time taken per op is 0.184us (5.4 Mqps) with batch size of 0 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.079us (12.7 Mqps) with batch size of 10 Time taken per op is 0.070us (14.2 Mqps) with batch size of 10 Time taken per op is 0.072us (14.0 Mqps) with batch size of 10 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.080us (12.5 Mqps) with batch size of 25 Time taken per op is 0.072us (14.0 Mqps) with batch size of 25 Time taken per op is 0.071us (14.1 Mqps) with batch size of 25 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.082us (12.1 Mqps) with batch size of 50 Time taken per op is 0.071us (14.1 Mqps) with batch size of 50 Time taken per op is 0.073us (13.6 Mqps) with batch size of 50 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.080us (12.5 Mqps) with batch size of 100 Time taken per op is 0.077us (13.0 Mqps) with batch size of 100 Time taken per op is 0.078us (12.8 Mqps) with batch size of 100 Test Plan: make check all make valgrind_check make asan_check Reviewers: sdong, ljin Reviewed By: ljin Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D22539
10 years ago
}
}
bool CuckooTableBuilder::IsDeletedKey(uint64_t idx) const {
assert(closed_);
return idx >= num_values_;
}
Slice CuckooTableBuilder::GetKey(uint64_t idx) const {
assert(closed_);
if (IsDeletedKey(idx)) {
return Slice(&deleted_keys_[static_cast<size_t>((idx - num_values_) * key_size_)], static_cast<size_t>(key_size_));
}
return Slice(&kvs_[static_cast<size_t>(idx * (key_size_ + value_size_))], static_cast<size_t>(key_size_));
}
Slice CuckooTableBuilder::GetUserKey(uint64_t idx) const {
assert(closed_);
return is_last_level_file_ ? GetKey(idx) : ExtractUserKey(GetKey(idx));
}
Slice CuckooTableBuilder::GetValue(uint64_t idx) const {
assert(closed_);
if (IsDeletedKey(idx)) {
static std::string empty_value(static_cast<unsigned int>(value_size_), 'a');
return Slice(empty_value);
}
return Slice(&kvs_[static_cast<size_t>(idx * (key_size_ + value_size_) + key_size_)], static_cast<size_t>(value_size_));
}
Status CuckooTableBuilder::MakeHashTable(std::vector<CuckooBucket>* buckets) {
buckets->resize(static_cast<size_t>(hash_table_size_ + cuckoo_block_size_ - 1));
uint32_t make_space_for_key_call_id = 0;
for (uint32_t vector_idx = 0; vector_idx < num_entries_; vector_idx++) {
uint64_t bucket_id = 0;
bool bucket_found = false;
autovector<uint64_t> hash_vals;
Slice user_key = GetUserKey(vector_idx);
for (uint32_t hash_cnt = 0; hash_cnt < num_hash_func_ && !bucket_found;
++hash_cnt) {
uint64_t hash_val = CuckooHash(user_key, hash_cnt, use_module_hash_,
hash_table_size_, identity_as_first_hash_, get_slice_hash_);
// If there is a collision, check next cuckoo_block_size_ locations for
// empty locations. While checking, if we reach end of the hash table,
// stop searching and proceed for next hash function.
for (uint32_t block_idx = 0; block_idx < cuckoo_block_size_;
++block_idx, ++hash_val) {
if ((*buckets)[static_cast<size_t>(hash_val)].vector_idx == kMaxVectorIdx) {
bucket_id = hash_val;
bucket_found = true;
break;
} else {
if (ucomp_->Compare(user_key,
GetUserKey((*buckets)[static_cast<size_t>(hash_val)].vector_idx)) == 0) {
return Status::NotSupported("Same key is being inserted again.");
}
hash_vals.push_back(hash_val);
}
}
}
while (!bucket_found && !MakeSpaceForKey(hash_vals,
Improve Cuckoo Table Reader performance. Inlined hash function and number of buckets a power of two. Summary: Use inlined hash functions instead of function pointer. Make number of buckets a power of two and use bitwise and instead of mod. After these changes, we get almost 50% improvement in performance. Results: With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.231us (4.3 Mqps) with batch size of 0 Time taken per op is 0.229us (4.4 Mqps) with batch size of 0 Time taken per op is 0.185us (5.4 Mqps) with batch size of 0 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.108us (9.3 Mqps) with batch size of 10 Time taken per op is 0.100us (10.0 Mqps) with batch size of 10 Time taken per op is 0.103us (9.7 Mqps) with batch size of 10 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.101us (9.9 Mqps) with batch size of 25 Time taken per op is 0.098us (10.2 Mqps) with batch size of 25 Time taken per op is 0.097us (10.3 Mqps) with batch size of 25 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.100us (10.0 Mqps) with batch size of 50 Time taken per op is 0.097us (10.3 Mqps) with batch size of 50 Time taken per op is 0.097us (10.3 Mqps) with batch size of 50 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.102us (9.8 Mqps) with batch size of 100 Time taken per op is 0.098us (10.2 Mqps) with batch size of 100 Time taken per op is 0.115us (8.7 Mqps) with batch size of 100 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.201us (5.0 Mqps) with batch size of 0 Time taken per op is 0.155us (6.5 Mqps) with batch size of 0 Time taken per op is 0.152us (6.6 Mqps) with batch size of 0 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.089us (11.3 Mqps) with batch size of 10 Time taken per op is 0.084us (11.9 Mqps) with batch size of 10 Time taken per op is 0.086us (11.6 Mqps) with batch size of 10 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.087us (11.5 Mqps) with batch size of 25 Time taken per op is 0.085us (11.7 Mqps) with batch size of 25 Time taken per op is 0.093us (10.8 Mqps) with batch size of 25 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.094us (10.6 Mqps) with batch size of 50 Time taken per op is 0.094us (10.7 Mqps) with batch size of 50 Time taken per op is 0.093us (10.8 Mqps) with batch size of 50 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.092us (10.9 Mqps) with batch size of 100 Time taken per op is 0.089us (11.2 Mqps) with batch size of 100 Time taken per op is 0.088us (11.3 Mqps) with batch size of 100 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.154us (6.5 Mqps) with batch size of 0 Time taken per op is 0.168us (6.0 Mqps) with batch size of 0 Time taken per op is 0.190us (5.3 Mqps) with batch size of 0 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.081us (12.4 Mqps) with batch size of 10 Time taken per op is 0.077us (13.0 Mqps) with batch size of 10 Time taken per op is 0.083us (12.1 Mqps) with batch size of 10 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.077us (13.0 Mqps) with batch size of 25 Time taken per op is 0.073us (13.7 Mqps) with batch size of 25 Time taken per op is 0.073us (13.7 Mqps) with batch size of 25 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.076us (13.1 Mqps) with batch size of 50 Time taken per op is 0.072us (13.8 Mqps) with batch size of 50 Time taken per op is 0.072us (13.8 Mqps) with batch size of 50 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.077us (13.0 Mqps) with batch size of 100 Time taken per op is 0.074us (13.6 Mqps) with batch size of 100 Time taken per op is 0.073us (13.6 Mqps) with batch size of 100 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.190us (5.3 Mqps) with batch size of 0 Time taken per op is 0.186us (5.4 Mqps) with batch size of 0 Time taken per op is 0.184us (5.4 Mqps) with batch size of 0 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.079us (12.7 Mqps) with batch size of 10 Time taken per op is 0.070us (14.2 Mqps) with batch size of 10 Time taken per op is 0.072us (14.0 Mqps) with batch size of 10 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.080us (12.5 Mqps) with batch size of 25 Time taken per op is 0.072us (14.0 Mqps) with batch size of 25 Time taken per op is 0.071us (14.1 Mqps) with batch size of 25 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.082us (12.1 Mqps) with batch size of 50 Time taken per op is 0.071us (14.1 Mqps) with batch size of 50 Time taken per op is 0.073us (13.6 Mqps) with batch size of 50 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.080us (12.5 Mqps) with batch size of 100 Time taken per op is 0.077us (13.0 Mqps) with batch size of 100 Time taken per op is 0.078us (12.8 Mqps) with batch size of 100 Test Plan: make check all make valgrind_check make asan_check Reviewers: sdong, ljin Reviewed By: ljin Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D22539
10 years ago
++make_space_for_key_call_id, buckets, &bucket_id)) {
// Rehash by increashing number of hash tables.
if (num_hash_func_ >= max_num_hash_func_) {
return Status::NotSupported("Too many collisions. Unable to hash.");
}
// We don't really need to rehash the entire table because old hashes are
// still valid and we only increased the number of hash functions.
uint64_t hash_val = CuckooHash(user_key, num_hash_func_, use_module_hash_,
hash_table_size_, identity_as_first_hash_, get_slice_hash_);
++num_hash_func_;
for (uint32_t block_idx = 0; block_idx < cuckoo_block_size_;
++block_idx, ++hash_val) {
if ((*buckets)[static_cast<size_t>(hash_val)].vector_idx == kMaxVectorIdx) {
bucket_found = true;
bucket_id = hash_val;
break;
} else {
hash_vals.push_back(hash_val);
}
}
}
(*buckets)[static_cast<size_t>(bucket_id)].vector_idx = vector_idx;
}
return Status::OK();
}
Status CuckooTableBuilder::Finish() {
assert(!closed_);
closed_ = true;
std::vector<CuckooBucket> buckets;
Improve Cuckoo Table Reader performance. Inlined hash function and number of buckets a power of two. Summary: Use inlined hash functions instead of function pointer. Make number of buckets a power of two and use bitwise and instead of mod. After these changes, we get almost 50% improvement in performance. Results: With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.231us (4.3 Mqps) with batch size of 0 Time taken per op is 0.229us (4.4 Mqps) with batch size of 0 Time taken per op is 0.185us (5.4 Mqps) with batch size of 0 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.108us (9.3 Mqps) with batch size of 10 Time taken per op is 0.100us (10.0 Mqps) with batch size of 10 Time taken per op is 0.103us (9.7 Mqps) with batch size of 10 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.101us (9.9 Mqps) with batch size of 25 Time taken per op is 0.098us (10.2 Mqps) with batch size of 25 Time taken per op is 0.097us (10.3 Mqps) with batch size of 25 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.100us (10.0 Mqps) with batch size of 50 Time taken per op is 0.097us (10.3 Mqps) with batch size of 50 Time taken per op is 0.097us (10.3 Mqps) with batch size of 50 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.102us (9.8 Mqps) with batch size of 100 Time taken per op is 0.098us (10.2 Mqps) with batch size of 100 Time taken per op is 0.115us (8.7 Mqps) with batch size of 100 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.201us (5.0 Mqps) with batch size of 0 Time taken per op is 0.155us (6.5 Mqps) with batch size of 0 Time taken per op is 0.152us (6.6 Mqps) with batch size of 0 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.089us (11.3 Mqps) with batch size of 10 Time taken per op is 0.084us (11.9 Mqps) with batch size of 10 Time taken per op is 0.086us (11.6 Mqps) with batch size of 10 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.087us (11.5 Mqps) with batch size of 25 Time taken per op is 0.085us (11.7 Mqps) with batch size of 25 Time taken per op is 0.093us (10.8 Mqps) with batch size of 25 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.094us (10.6 Mqps) with batch size of 50 Time taken per op is 0.094us (10.7 Mqps) with batch size of 50 Time taken per op is 0.093us (10.8 Mqps) with batch size of 50 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.092us (10.9 Mqps) with batch size of 100 Time taken per op is 0.089us (11.2 Mqps) with batch size of 100 Time taken per op is 0.088us (11.3 Mqps) with batch size of 100 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.154us (6.5 Mqps) with batch size of 0 Time taken per op is 0.168us (6.0 Mqps) with batch size of 0 Time taken per op is 0.190us (5.3 Mqps) with batch size of 0 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.081us (12.4 Mqps) with batch size of 10 Time taken per op is 0.077us (13.0 Mqps) with batch size of 10 Time taken per op is 0.083us (12.1 Mqps) with batch size of 10 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.077us (13.0 Mqps) with batch size of 25 Time taken per op is 0.073us (13.7 Mqps) with batch size of 25 Time taken per op is 0.073us (13.7 Mqps) with batch size of 25 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.076us (13.1 Mqps) with batch size of 50 Time taken per op is 0.072us (13.8 Mqps) with batch size of 50 Time taken per op is 0.072us (13.8 Mqps) with batch size of 50 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.077us (13.0 Mqps) with batch size of 100 Time taken per op is 0.074us (13.6 Mqps) with batch size of 100 Time taken per op is 0.073us (13.6 Mqps) with batch size of 100 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.190us (5.3 Mqps) with batch size of 0 Time taken per op is 0.186us (5.4 Mqps) with batch size of 0 Time taken per op is 0.184us (5.4 Mqps) with batch size of 0 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.079us (12.7 Mqps) with batch size of 10 Time taken per op is 0.070us (14.2 Mqps) with batch size of 10 Time taken per op is 0.072us (14.0 Mqps) with batch size of 10 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.080us (12.5 Mqps) with batch size of 25 Time taken per op is 0.072us (14.0 Mqps) with batch size of 25 Time taken per op is 0.071us (14.1 Mqps) with batch size of 25 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.082us (12.1 Mqps) with batch size of 50 Time taken per op is 0.071us (14.1 Mqps) with batch size of 50 Time taken per op is 0.073us (13.6 Mqps) with batch size of 50 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.080us (12.5 Mqps) with batch size of 100 Time taken per op is 0.077us (13.0 Mqps) with batch size of 100 Time taken per op is 0.078us (12.8 Mqps) with batch size of 100 Test Plan: make check all make valgrind_check make asan_check Reviewers: sdong, ljin Reviewed By: ljin Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D22539
10 years ago
Status s;
std::string unused_bucket;
if (num_entries_ > 0) {
// Calculate the real hash size if module hash is enabled.
if (use_module_hash_) {
hash_table_size_ =
static_cast<uint64_t>(num_entries_ / max_hash_table_ratio_);
}
Improve Cuckoo Table Reader performance. Inlined hash function and number of buckets a power of two. Summary: Use inlined hash functions instead of function pointer. Make number of buckets a power of two and use bitwise and instead of mod. After these changes, we get almost 50% improvement in performance. Results: With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.231us (4.3 Mqps) with batch size of 0 Time taken per op is 0.229us (4.4 Mqps) with batch size of 0 Time taken per op is 0.185us (5.4 Mqps) with batch size of 0 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.108us (9.3 Mqps) with batch size of 10 Time taken per op is 0.100us (10.0 Mqps) with batch size of 10 Time taken per op is 0.103us (9.7 Mqps) with batch size of 10 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.101us (9.9 Mqps) with batch size of 25 Time taken per op is 0.098us (10.2 Mqps) with batch size of 25 Time taken per op is 0.097us (10.3 Mqps) with batch size of 25 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.100us (10.0 Mqps) with batch size of 50 Time taken per op is 0.097us (10.3 Mqps) with batch size of 50 Time taken per op is 0.097us (10.3 Mqps) with batch size of 50 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.102us (9.8 Mqps) with batch size of 100 Time taken per op is 0.098us (10.2 Mqps) with batch size of 100 Time taken per op is 0.115us (8.7 Mqps) with batch size of 100 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.201us (5.0 Mqps) with batch size of 0 Time taken per op is 0.155us (6.5 Mqps) with batch size of 0 Time taken per op is 0.152us (6.6 Mqps) with batch size of 0 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.089us (11.3 Mqps) with batch size of 10 Time taken per op is 0.084us (11.9 Mqps) with batch size of 10 Time taken per op is 0.086us (11.6 Mqps) with batch size of 10 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.087us (11.5 Mqps) with batch size of 25 Time taken per op is 0.085us (11.7 Mqps) with batch size of 25 Time taken per op is 0.093us (10.8 Mqps) with batch size of 25 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.094us (10.6 Mqps) with batch size of 50 Time taken per op is 0.094us (10.7 Mqps) with batch size of 50 Time taken per op is 0.093us (10.8 Mqps) with batch size of 50 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.092us (10.9 Mqps) with batch size of 100 Time taken per op is 0.089us (11.2 Mqps) with batch size of 100 Time taken per op is 0.088us (11.3 Mqps) with batch size of 100 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.154us (6.5 Mqps) with batch size of 0 Time taken per op is 0.168us (6.0 Mqps) with batch size of 0 Time taken per op is 0.190us (5.3 Mqps) with batch size of 0 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.081us (12.4 Mqps) with batch size of 10 Time taken per op is 0.077us (13.0 Mqps) with batch size of 10 Time taken per op is 0.083us (12.1 Mqps) with batch size of 10 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.077us (13.0 Mqps) with batch size of 25 Time taken per op is 0.073us (13.7 Mqps) with batch size of 25 Time taken per op is 0.073us (13.7 Mqps) with batch size of 25 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.076us (13.1 Mqps) with batch size of 50 Time taken per op is 0.072us (13.8 Mqps) with batch size of 50 Time taken per op is 0.072us (13.8 Mqps) with batch size of 50 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.077us (13.0 Mqps) with batch size of 100 Time taken per op is 0.074us (13.6 Mqps) with batch size of 100 Time taken per op is 0.073us (13.6 Mqps) with batch size of 100 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.190us (5.3 Mqps) with batch size of 0 Time taken per op is 0.186us (5.4 Mqps) with batch size of 0 Time taken per op is 0.184us (5.4 Mqps) with batch size of 0 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.079us (12.7 Mqps) with batch size of 10 Time taken per op is 0.070us (14.2 Mqps) with batch size of 10 Time taken per op is 0.072us (14.0 Mqps) with batch size of 10 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.080us (12.5 Mqps) with batch size of 25 Time taken per op is 0.072us (14.0 Mqps) with batch size of 25 Time taken per op is 0.071us (14.1 Mqps) with batch size of 25 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.082us (12.1 Mqps) with batch size of 50 Time taken per op is 0.071us (14.1 Mqps) with batch size of 50 Time taken per op is 0.073us (13.6 Mqps) with batch size of 50 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.080us (12.5 Mqps) with batch size of 100 Time taken per op is 0.077us (13.0 Mqps) with batch size of 100 Time taken per op is 0.078us (12.8 Mqps) with batch size of 100 Test Plan: make check all make valgrind_check make asan_check Reviewers: sdong, ljin Reviewed By: ljin Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D22539
10 years ago
s = MakeHashTable(&buckets);
if (!s.ok()) {
return s;
}
// Determine unused_user_key to fill empty buckets.
std::string unused_user_key = smallest_user_key_;
int curr_pos = static_cast<int>(unused_user_key.size()) - 1;
while (curr_pos >= 0) {
--unused_user_key[curr_pos];
if (Slice(unused_user_key).compare(smallest_user_key_) < 0) {
break;
}
--curr_pos;
}
if (curr_pos < 0) {
// Try using the largest key to identify an unused key.
unused_user_key = largest_user_key_;
curr_pos = static_cast<int>(unused_user_key.size()) - 1;
while (curr_pos >= 0) {
++unused_user_key[curr_pos];
if (Slice(unused_user_key).compare(largest_user_key_) > 0) {
break;
}
--curr_pos;
}
}
if (curr_pos < 0) {
return Status::Corruption("Unable to find unused key");
}
if (is_last_level_file_) {
unused_bucket = unused_user_key;
} else {
ParsedInternalKey ikey(unused_user_key, 0, kTypeValue);
AppendInternalKey(&unused_bucket, ikey);
}
}
properties_.num_entries = num_entries_;
properties_.num_deletions = num_entries_ - num_values_;
properties_.fixed_key_len = key_size_;
properties_.user_collected_properties[
CuckooTablePropertyNames::kValueLength].assign(
reinterpret_cast<const char*>(&value_size_), sizeof(value_size_));
uint64_t bucket_size = key_size_ + value_size_;
unused_bucket.resize(static_cast<size_t>(bucket_size), 'a');
// Write the table.
uint32_t num_added = 0;
for (auto& bucket : buckets) {
if (bucket.vector_idx == kMaxVectorIdx) {
s = file_->Append(Slice(unused_bucket));
} else {
++num_added;
s = file_->Append(GetKey(bucket.vector_idx));
if (s.ok()) {
if (value_size_ > 0) {
s = file_->Append(GetValue(bucket.vector_idx));
}
}
}
if (!s.ok()) {
return s;
}
}
assert(num_added == NumEntries());
properties_.raw_key_size = num_added * properties_.fixed_key_len;
properties_.raw_value_size = num_added * value_size_;
uint64_t offset = buckets.size() * bucket_size;
properties_.data_size = offset;
unused_bucket.resize(static_cast<size_t>(properties_.fixed_key_len));
properties_.user_collected_properties[
CuckooTablePropertyNames::kEmptyKey] = unused_bucket;
properties_.user_collected_properties[
CuckooTablePropertyNames::kNumHashFunc].assign(
reinterpret_cast<char*>(&num_hash_func_), sizeof(num_hash_func_));
properties_.user_collected_properties[
CuckooTablePropertyNames::kHashTableSize].assign(
reinterpret_cast<const char*>(&hash_table_size_),
sizeof(hash_table_size_));
properties_.user_collected_properties[
CuckooTablePropertyNames::kIsLastLevel].assign(
reinterpret_cast<const char*>(&is_last_level_file_),
sizeof(is_last_level_file_));
properties_.user_collected_properties[
CuckooTablePropertyNames::kCuckooBlockSize].assign(
reinterpret_cast<const char*>(&cuckoo_block_size_),
sizeof(cuckoo_block_size_));
CuckooTable: add one option to allow identity function for the first hash function Summary: MurmurHash becomes expensive when we do millions Get() a second in one thread. Add this option to allow the first hash function to use identity function as hash function. It results in QPS increase from 3.7M/s to ~4.3M/s. I did not observe improvement for end to end RocksDB performance. This may be caused by other bottlenecks that I will address in a separate diff. Test Plan: ``` [ljin@dev1964 rocksdb] ./cuckoo_table_reader_test --enable_perf --file_dir=/dev/shm --write --identity_as_first_hash=0 ==== Test CuckooReaderTest.WhenKeyExists ==== Test CuckooReaderTest.WhenKeyExistsWithUint64Comparator ==== Test CuckooReaderTest.CheckIterator ==== Test CuckooReaderTest.CheckIteratorUint64 ==== Test CuckooReaderTest.WhenKeyNotFound ==== Test CuckooReaderTest.TestReadPerformance With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.272us (3.7 Mqps) with batch size of 0, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.138us (7.2 Mqps) with batch size of 10, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.142us (7.1 Mqps) with batch size of 25, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.142us (7.0 Mqps) with batch size of 50, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.144us (6.9 Mqps) with batch size of 100, # of found keys 125829120 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.201us (5.0 Mqps) with batch size of 0, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.121us (8.3 Mqps) with batch size of 10, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.123us (8.1 Mqps) with batch size of 25, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.121us (8.3 Mqps) with batch size of 50, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.112us (8.9 Mqps) with batch size of 100, # of found keys 104857600 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.251us (4.0 Mqps) with batch size of 0, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.107us (9.4 Mqps) with batch size of 10, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.099us (10.1 Mqps) with batch size of 25, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.100us (10.0 Mqps) with batch size of 50, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.116us (8.6 Mqps) with batch size of 100, # of found keys 83886080 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.189us (5.3 Mqps) with batch size of 0, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.095us (10.5 Mqps) with batch size of 10, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.096us (10.4 Mqps) with batch size of 25, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.098us (10.2 Mqps) with batch size of 50, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.105us (9.5 Mqps) with batch size of 100, # of found keys 73400320 [ljin@dev1964 rocksdb] ./cuckoo_table_reader_test --enable_perf --file_dir=/dev/shm --write --identity_as_first_hash=1 ==== Test CuckooReaderTest.WhenKeyExists ==== Test CuckooReaderTest.WhenKeyExistsWithUint64Comparator ==== Test CuckooReaderTest.CheckIterator ==== Test CuckooReaderTest.CheckIteratorUint64 ==== Test CuckooReaderTest.WhenKeyNotFound ==== Test CuckooReaderTest.TestReadPerformance With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.230us (4.3 Mqps) with batch size of 0, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.086us (11.7 Mqps) with batch size of 10, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.088us (11.3 Mqps) with batch size of 25, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.083us (12.1 Mqps) with batch size of 50, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.083us (12.1 Mqps) with batch size of 100, # of found keys 125829120 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.159us (6.3 Mqps) with batch size of 0, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.078us (12.8 Mqps) with batch size of 10, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.080us (12.6 Mqps) with batch size of 25, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.080us (12.5 Mqps) with batch size of 50, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.082us (12.2 Mqps) with batch size of 100, # of found keys 104857600 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.154us (6.5 Mqps) with batch size of 0, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.077us (13.0 Mqps) with batch size of 10, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.077us (12.9 Mqps) with batch size of 25, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.078us (12.8 Mqps) with batch size of 50, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.079us (12.6 Mqps) with batch size of 100, # of found keys 83886080 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.218us (4.6 Mqps) with batch size of 0, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.083us (12.0 Mqps) with batch size of 10, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.085us (11.7 Mqps) with batch size of 25, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.086us (11.6 Mqps) with batch size of 50, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.078us (12.8 Mqps) with batch size of 100, # of found keys 73400320 ``` Reviewers: sdong, igor, yhchiang Reviewed By: igor Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D23451
10 years ago
properties_.user_collected_properties[
CuckooTablePropertyNames::kIdentityAsFirstHash].assign(
reinterpret_cast<const char*>(&identity_as_first_hash_),
sizeof(identity_as_first_hash_));
properties_.user_collected_properties[
CuckooTablePropertyNames::kUseModuleHash].assign(
reinterpret_cast<const char*>(&use_module_hash_),
sizeof(use_module_hash_));
uint32_t user_key_len = static_cast<uint32_t>(smallest_user_key_.size());
properties_.user_collected_properties[
CuckooTablePropertyNames::kUserKeyLength].assign(
reinterpret_cast<const char*>(&user_key_len),
sizeof(user_key_len));
// Write meta blocks.
MetaIndexBuilder meta_index_builder;
PropertyBlockBuilder property_block_builder;
property_block_builder.AddTableProperty(properties_);
property_block_builder.Add(properties_.user_collected_properties);
Slice property_block = property_block_builder.Finish();
BlockHandle property_block_handle;
property_block_handle.set_offset(offset);
property_block_handle.set_size(property_block.size());
s = file_->Append(property_block);
offset += property_block.size();
if (!s.ok()) {
return s;
}
meta_index_builder.Add(kPropertiesBlock, property_block_handle);
Slice meta_index_block = meta_index_builder.Finish();
BlockHandle meta_index_block_handle;
meta_index_block_handle.set_offset(offset);
meta_index_block_handle.set_size(meta_index_block.size());
s = file_->Append(meta_index_block);
if (!s.ok()) {
return s;
}
Footer footer(kCuckooTableMagicNumber, 1);
footer.set_metaindex_handle(meta_index_block_handle);
footer.set_index_handle(BlockHandle::NullBlockHandle());
std::string footer_encoding;
footer.EncodeTo(&footer_encoding);
s = file_->Append(footer_encoding);
return s;
}
void CuckooTableBuilder::Abandon() {
assert(!closed_);
closed_ = true;
}
uint64_t CuckooTableBuilder::NumEntries() const {
return num_entries_;
}
uint64_t CuckooTableBuilder::FileSize() const {
if (closed_) {
return file_->GetFileSize();
} else if (num_entries_ == 0) {
return 0;
}
Improve Cuckoo Table Reader performance. Inlined hash function and number of buckets a power of two. Summary: Use inlined hash functions instead of function pointer. Make number of buckets a power of two and use bitwise and instead of mod. After these changes, we get almost 50% improvement in performance. Results: With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.231us (4.3 Mqps) with batch size of 0 Time taken per op is 0.229us (4.4 Mqps) with batch size of 0 Time taken per op is 0.185us (5.4 Mqps) with batch size of 0 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.108us (9.3 Mqps) with batch size of 10 Time taken per op is 0.100us (10.0 Mqps) with batch size of 10 Time taken per op is 0.103us (9.7 Mqps) with batch size of 10 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.101us (9.9 Mqps) with batch size of 25 Time taken per op is 0.098us (10.2 Mqps) with batch size of 25 Time taken per op is 0.097us (10.3 Mqps) with batch size of 25 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.100us (10.0 Mqps) with batch size of 50 Time taken per op is 0.097us (10.3 Mqps) with batch size of 50 Time taken per op is 0.097us (10.3 Mqps) with batch size of 50 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.102us (9.8 Mqps) with batch size of 100 Time taken per op is 0.098us (10.2 Mqps) with batch size of 100 Time taken per op is 0.115us (8.7 Mqps) with batch size of 100 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.201us (5.0 Mqps) with batch size of 0 Time taken per op is 0.155us (6.5 Mqps) with batch size of 0 Time taken per op is 0.152us (6.6 Mqps) with batch size of 0 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.089us (11.3 Mqps) with batch size of 10 Time taken per op is 0.084us (11.9 Mqps) with batch size of 10 Time taken per op is 0.086us (11.6 Mqps) with batch size of 10 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.087us (11.5 Mqps) with batch size of 25 Time taken per op is 0.085us (11.7 Mqps) with batch size of 25 Time taken per op is 0.093us (10.8 Mqps) with batch size of 25 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.094us (10.6 Mqps) with batch size of 50 Time taken per op is 0.094us (10.7 Mqps) with batch size of 50 Time taken per op is 0.093us (10.8 Mqps) with batch size of 50 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.092us (10.9 Mqps) with batch size of 100 Time taken per op is 0.089us (11.2 Mqps) with batch size of 100 Time taken per op is 0.088us (11.3 Mqps) with batch size of 100 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.154us (6.5 Mqps) with batch size of 0 Time taken per op is 0.168us (6.0 Mqps) with batch size of 0 Time taken per op is 0.190us (5.3 Mqps) with batch size of 0 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.081us (12.4 Mqps) with batch size of 10 Time taken per op is 0.077us (13.0 Mqps) with batch size of 10 Time taken per op is 0.083us (12.1 Mqps) with batch size of 10 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.077us (13.0 Mqps) with batch size of 25 Time taken per op is 0.073us (13.7 Mqps) with batch size of 25 Time taken per op is 0.073us (13.7 Mqps) with batch size of 25 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.076us (13.1 Mqps) with batch size of 50 Time taken per op is 0.072us (13.8 Mqps) with batch size of 50 Time taken per op is 0.072us (13.8 Mqps) with batch size of 50 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.077us (13.0 Mqps) with batch size of 100 Time taken per op is 0.074us (13.6 Mqps) with batch size of 100 Time taken per op is 0.073us (13.6 Mqps) with batch size of 100 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.190us (5.3 Mqps) with batch size of 0 Time taken per op is 0.186us (5.4 Mqps) with batch size of 0 Time taken per op is 0.184us (5.4 Mqps) with batch size of 0 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.079us (12.7 Mqps) with batch size of 10 Time taken per op is 0.070us (14.2 Mqps) with batch size of 10 Time taken per op is 0.072us (14.0 Mqps) with batch size of 10 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.080us (12.5 Mqps) with batch size of 25 Time taken per op is 0.072us (14.0 Mqps) with batch size of 25 Time taken per op is 0.071us (14.1 Mqps) with batch size of 25 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.082us (12.1 Mqps) with batch size of 50 Time taken per op is 0.071us (14.1 Mqps) with batch size of 50 Time taken per op is 0.073us (13.6 Mqps) with batch size of 50 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.080us (12.5 Mqps) with batch size of 100 Time taken per op is 0.077us (13.0 Mqps) with batch size of 100 Time taken per op is 0.078us (12.8 Mqps) with batch size of 100 Test Plan: make check all make valgrind_check make asan_check Reviewers: sdong, ljin Reviewed By: ljin Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D22539
10 years ago
if (use_module_hash_) {
return static_cast<uint64_t>((key_size_ + value_size_) *
num_entries_ / max_hash_table_ratio_);
} else {
// Account for buckets being a power of two.
// As elements are added, file size remains constant for a while and
// doubles its size. Since compaction algorithm stops adding elements
// only after it exceeds the file limit, we account for the extra element
// being added here.
uint64_t expected_hash_table_size = hash_table_size_;
if (expected_hash_table_size < (num_entries_ + 1) / max_hash_table_ratio_) {
expected_hash_table_size *= 2;
}
return (key_size_ + value_size_) * expected_hash_table_size - 1;
Improve Cuckoo Table Reader performance. Inlined hash function and number of buckets a power of two. Summary: Use inlined hash functions instead of function pointer. Make number of buckets a power of two and use bitwise and instead of mod. After these changes, we get almost 50% improvement in performance. Results: With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.231us (4.3 Mqps) with batch size of 0 Time taken per op is 0.229us (4.4 Mqps) with batch size of 0 Time taken per op is 0.185us (5.4 Mqps) with batch size of 0 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.108us (9.3 Mqps) with batch size of 10 Time taken per op is 0.100us (10.0 Mqps) with batch size of 10 Time taken per op is 0.103us (9.7 Mqps) with batch size of 10 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.101us (9.9 Mqps) with batch size of 25 Time taken per op is 0.098us (10.2 Mqps) with batch size of 25 Time taken per op is 0.097us (10.3 Mqps) with batch size of 25 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.100us (10.0 Mqps) with batch size of 50 Time taken per op is 0.097us (10.3 Mqps) with batch size of 50 Time taken per op is 0.097us (10.3 Mqps) with batch size of 50 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.102us (9.8 Mqps) with batch size of 100 Time taken per op is 0.098us (10.2 Mqps) with batch size of 100 Time taken per op is 0.115us (8.7 Mqps) with batch size of 100 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.201us (5.0 Mqps) with batch size of 0 Time taken per op is 0.155us (6.5 Mqps) with batch size of 0 Time taken per op is 0.152us (6.6 Mqps) with batch size of 0 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.089us (11.3 Mqps) with batch size of 10 Time taken per op is 0.084us (11.9 Mqps) with batch size of 10 Time taken per op is 0.086us (11.6 Mqps) with batch size of 10 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.087us (11.5 Mqps) with batch size of 25 Time taken per op is 0.085us (11.7 Mqps) with batch size of 25 Time taken per op is 0.093us (10.8 Mqps) with batch size of 25 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.094us (10.6 Mqps) with batch size of 50 Time taken per op is 0.094us (10.7 Mqps) with batch size of 50 Time taken per op is 0.093us (10.8 Mqps) with batch size of 50 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.092us (10.9 Mqps) with batch size of 100 Time taken per op is 0.089us (11.2 Mqps) with batch size of 100 Time taken per op is 0.088us (11.3 Mqps) with batch size of 100 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.154us (6.5 Mqps) with batch size of 0 Time taken per op is 0.168us (6.0 Mqps) with batch size of 0 Time taken per op is 0.190us (5.3 Mqps) with batch size of 0 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.081us (12.4 Mqps) with batch size of 10 Time taken per op is 0.077us (13.0 Mqps) with batch size of 10 Time taken per op is 0.083us (12.1 Mqps) with batch size of 10 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.077us (13.0 Mqps) with batch size of 25 Time taken per op is 0.073us (13.7 Mqps) with batch size of 25 Time taken per op is 0.073us (13.7 Mqps) with batch size of 25 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.076us (13.1 Mqps) with batch size of 50 Time taken per op is 0.072us (13.8 Mqps) with batch size of 50 Time taken per op is 0.072us (13.8 Mqps) with batch size of 50 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.077us (13.0 Mqps) with batch size of 100 Time taken per op is 0.074us (13.6 Mqps) with batch size of 100 Time taken per op is 0.073us (13.6 Mqps) with batch size of 100 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.190us (5.3 Mqps) with batch size of 0 Time taken per op is 0.186us (5.4 Mqps) with batch size of 0 Time taken per op is 0.184us (5.4 Mqps) with batch size of 0 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.079us (12.7 Mqps) with batch size of 10 Time taken per op is 0.070us (14.2 Mqps) with batch size of 10 Time taken per op is 0.072us (14.0 Mqps) with batch size of 10 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.080us (12.5 Mqps) with batch size of 25 Time taken per op is 0.072us (14.0 Mqps) with batch size of 25 Time taken per op is 0.071us (14.1 Mqps) with batch size of 25 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.082us (12.1 Mqps) with batch size of 50 Time taken per op is 0.071us (14.1 Mqps) with batch size of 50 Time taken per op is 0.073us (13.6 Mqps) with batch size of 50 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.080us (12.5 Mqps) with batch size of 100 Time taken per op is 0.077us (13.0 Mqps) with batch size of 100 Time taken per op is 0.078us (12.8 Mqps) with batch size of 100 Test Plan: make check all make valgrind_check make asan_check Reviewers: sdong, ljin Reviewed By: ljin Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D22539
10 years ago
}
}
// This method is invoked when there is no place to insert the target key.
// It searches for a set of elements that can be moved to accommodate target
// key. The search is a BFS graph traversal with first level (hash_vals)
// being all the buckets target key could go to.
// Then, from each node (curr_node), we find all the buckets that curr_node
// could go to. They form the children of curr_node in the tree.
// We continue the traversal until we find an empty bucket, in which case, we
// move all elements along the path from first level to this empty bucket, to
// make space for target key which is inserted at first level (*bucket_id).
// If tree depth exceedes max depth, we return false indicating failure.
bool CuckooTableBuilder::MakeSpaceForKey(
const autovector<uint64_t>& hash_vals,
const uint32_t make_space_for_key_call_id,
std::vector<CuckooBucket>* buckets, uint64_t* bucket_id) {
struct CuckooNode {
uint64_t bucket_id;
uint32_t depth;
uint32_t parent_pos;
CuckooNode(uint64_t _bucket_id, uint32_t _depth, int _parent_pos)
: bucket_id(_bucket_id), depth(_depth), parent_pos(_parent_pos) {}
};
// This is BFS search tree that is stored simply as a vector.
// Each node stores the index of parent node in the vector.
std::vector<CuckooNode> tree;
// We want to identify already visited buckets in the current method call so
// that we don't add same buckets again for exploration in the tree.
// We do this by maintaining a count of current method call in
// make_space_for_key_call_id, which acts as a unique id for this invocation
// of the method. We store this number into the nodes that we explore in
// current method call.
// It is unlikely for the increment operation to overflow because the maximum
// no. of times this will be called is <= max_num_hash_func_ + num_entries_.
for (uint32_t hash_cnt = 0; hash_cnt < num_hash_func_; ++hash_cnt) {
uint64_t bid = hash_vals[hash_cnt];
(*buckets)[static_cast<size_t>(bid)].make_space_for_key_call_id = make_space_for_key_call_id;
tree.push_back(CuckooNode(bid, 0, 0));
}
bool null_found = false;
uint32_t curr_pos = 0;
while (!null_found && curr_pos < tree.size()) {
CuckooNode& curr_node = tree[curr_pos];
uint32_t curr_depth = curr_node.depth;
if (curr_depth >= max_search_depth_) {
break;
}
CuckooBucket& curr_bucket = (*buckets)[static_cast<size_t>(curr_node.bucket_id)];
for (uint32_t hash_cnt = 0;
hash_cnt < num_hash_func_ && !null_found; ++hash_cnt) {
uint64_t child_bucket_id = CuckooHash(GetUserKey(curr_bucket.vector_idx),
hash_cnt, use_module_hash_, hash_table_size_, identity_as_first_hash_,
CuckooTable: add one option to allow identity function for the first hash function Summary: MurmurHash becomes expensive when we do millions Get() a second in one thread. Add this option to allow the first hash function to use identity function as hash function. It results in QPS increase from 3.7M/s to ~4.3M/s. I did not observe improvement for end to end RocksDB performance. This may be caused by other bottlenecks that I will address in a separate diff. Test Plan: ``` [ljin@dev1964 rocksdb] ./cuckoo_table_reader_test --enable_perf --file_dir=/dev/shm --write --identity_as_first_hash=0 ==== Test CuckooReaderTest.WhenKeyExists ==== Test CuckooReaderTest.WhenKeyExistsWithUint64Comparator ==== Test CuckooReaderTest.CheckIterator ==== Test CuckooReaderTest.CheckIteratorUint64 ==== Test CuckooReaderTest.WhenKeyNotFound ==== Test CuckooReaderTest.TestReadPerformance With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.272us (3.7 Mqps) with batch size of 0, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.138us (7.2 Mqps) with batch size of 10, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.142us (7.1 Mqps) with batch size of 25, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.142us (7.0 Mqps) with batch size of 50, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.144us (6.9 Mqps) with batch size of 100, # of found keys 125829120 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.201us (5.0 Mqps) with batch size of 0, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.121us (8.3 Mqps) with batch size of 10, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.123us (8.1 Mqps) with batch size of 25, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.121us (8.3 Mqps) with batch size of 50, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.112us (8.9 Mqps) with batch size of 100, # of found keys 104857600 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.251us (4.0 Mqps) with batch size of 0, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.107us (9.4 Mqps) with batch size of 10, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.099us (10.1 Mqps) with batch size of 25, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.100us (10.0 Mqps) with batch size of 50, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.116us (8.6 Mqps) with batch size of 100, # of found keys 83886080 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.189us (5.3 Mqps) with batch size of 0, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.095us (10.5 Mqps) with batch size of 10, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.096us (10.4 Mqps) with batch size of 25, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.098us (10.2 Mqps) with batch size of 50, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.105us (9.5 Mqps) with batch size of 100, # of found keys 73400320 [ljin@dev1964 rocksdb] ./cuckoo_table_reader_test --enable_perf --file_dir=/dev/shm --write --identity_as_first_hash=1 ==== Test CuckooReaderTest.WhenKeyExists ==== Test CuckooReaderTest.WhenKeyExistsWithUint64Comparator ==== Test CuckooReaderTest.CheckIterator ==== Test CuckooReaderTest.CheckIteratorUint64 ==== Test CuckooReaderTest.WhenKeyNotFound ==== Test CuckooReaderTest.TestReadPerformance With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.230us (4.3 Mqps) with batch size of 0, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.086us (11.7 Mqps) with batch size of 10, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.088us (11.3 Mqps) with batch size of 25, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.083us (12.1 Mqps) with batch size of 50, # of found keys 125829120 With 125829120 items, utilization is 93.75%, number of hash functions: 2. Time taken per op is 0.083us (12.1 Mqps) with batch size of 100, # of found keys 125829120 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.159us (6.3 Mqps) with batch size of 0, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.078us (12.8 Mqps) with batch size of 10, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.080us (12.6 Mqps) with batch size of 25, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.080us (12.5 Mqps) with batch size of 50, # of found keys 104857600 With 104857600 items, utilization is 78.12%, number of hash functions: 2. Time taken per op is 0.082us (12.2 Mqps) with batch size of 100, # of found keys 104857600 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.154us (6.5 Mqps) with batch size of 0, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.077us (13.0 Mqps) with batch size of 10, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.077us (12.9 Mqps) with batch size of 25, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.078us (12.8 Mqps) with batch size of 50, # of found keys 83886080 With 83886080 items, utilization is 62.50%, number of hash functions: 2. Time taken per op is 0.079us (12.6 Mqps) with batch size of 100, # of found keys 83886080 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.218us (4.6 Mqps) with batch size of 0, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.083us (12.0 Mqps) with batch size of 10, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.085us (11.7 Mqps) with batch size of 25, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.086us (11.6 Mqps) with batch size of 50, # of found keys 73400320 With 73400320 items, utilization is 54.69%, number of hash functions: 2. Time taken per op is 0.078us (12.8 Mqps) with batch size of 100, # of found keys 73400320 ``` Reviewers: sdong, igor, yhchiang Reviewed By: igor Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D23451
10 years ago
get_slice_hash_);
Improve Cuckoo Table Reader performance. Inlined hash function and number of buckets a power of two. Summary: Use inlined hash functions instead of function pointer. Make number of buckets a power of two and use bitwise and instead of mod. After these changes, we get almost 50% improvement in performance. Results: With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.231us (4.3 Mqps) with batch size of 0 Time taken per op is 0.229us (4.4 Mqps) with batch size of 0 Time taken per op is 0.185us (5.4 Mqps) with batch size of 0 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.108us (9.3 Mqps) with batch size of 10 Time taken per op is 0.100us (10.0 Mqps) with batch size of 10 Time taken per op is 0.103us (9.7 Mqps) with batch size of 10 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.101us (9.9 Mqps) with batch size of 25 Time taken per op is 0.098us (10.2 Mqps) with batch size of 25 Time taken per op is 0.097us (10.3 Mqps) with batch size of 25 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.100us (10.0 Mqps) with batch size of 50 Time taken per op is 0.097us (10.3 Mqps) with batch size of 50 Time taken per op is 0.097us (10.3 Mqps) with batch size of 50 With 120000000 items, utilization is 89.41%, number of hash functions: 2. Time taken per op is 0.102us (9.8 Mqps) with batch size of 100 Time taken per op is 0.098us (10.2 Mqps) with batch size of 100 Time taken per op is 0.115us (8.7 Mqps) with batch size of 100 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.201us (5.0 Mqps) with batch size of 0 Time taken per op is 0.155us (6.5 Mqps) with batch size of 0 Time taken per op is 0.152us (6.6 Mqps) with batch size of 0 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.089us (11.3 Mqps) with batch size of 10 Time taken per op is 0.084us (11.9 Mqps) with batch size of 10 Time taken per op is 0.086us (11.6 Mqps) with batch size of 10 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.087us (11.5 Mqps) with batch size of 25 Time taken per op is 0.085us (11.7 Mqps) with batch size of 25 Time taken per op is 0.093us (10.8 Mqps) with batch size of 25 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.094us (10.6 Mqps) with batch size of 50 Time taken per op is 0.094us (10.7 Mqps) with batch size of 50 Time taken per op is 0.093us (10.8 Mqps) with batch size of 50 With 100000000 items, utilization is 74.51%, number of hash functions: 2. Time taken per op is 0.092us (10.9 Mqps) with batch size of 100 Time taken per op is 0.089us (11.2 Mqps) with batch size of 100 Time taken per op is 0.088us (11.3 Mqps) with batch size of 100 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.154us (6.5 Mqps) with batch size of 0 Time taken per op is 0.168us (6.0 Mqps) with batch size of 0 Time taken per op is 0.190us (5.3 Mqps) with batch size of 0 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.081us (12.4 Mqps) with batch size of 10 Time taken per op is 0.077us (13.0 Mqps) with batch size of 10 Time taken per op is 0.083us (12.1 Mqps) with batch size of 10 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.077us (13.0 Mqps) with batch size of 25 Time taken per op is 0.073us (13.7 Mqps) with batch size of 25 Time taken per op is 0.073us (13.7 Mqps) with batch size of 25 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.076us (13.1 Mqps) with batch size of 50 Time taken per op is 0.072us (13.8 Mqps) with batch size of 50 Time taken per op is 0.072us (13.8 Mqps) with batch size of 50 With 80000000 items, utilization is 59.60%, number of hash functions: 2. Time taken per op is 0.077us (13.0 Mqps) with batch size of 100 Time taken per op is 0.074us (13.6 Mqps) with batch size of 100 Time taken per op is 0.073us (13.6 Mqps) with batch size of 100 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.190us (5.3 Mqps) with batch size of 0 Time taken per op is 0.186us (5.4 Mqps) with batch size of 0 Time taken per op is 0.184us (5.4 Mqps) with batch size of 0 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.079us (12.7 Mqps) with batch size of 10 Time taken per op is 0.070us (14.2 Mqps) with batch size of 10 Time taken per op is 0.072us (14.0 Mqps) with batch size of 10 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.080us (12.5 Mqps) with batch size of 25 Time taken per op is 0.072us (14.0 Mqps) with batch size of 25 Time taken per op is 0.071us (14.1 Mqps) with batch size of 25 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.082us (12.1 Mqps) with batch size of 50 Time taken per op is 0.071us (14.1 Mqps) with batch size of 50 Time taken per op is 0.073us (13.6 Mqps) with batch size of 50 With 70000000 items, utilization is 52.15%, number of hash functions: 2. Time taken per op is 0.080us (12.5 Mqps) with batch size of 100 Time taken per op is 0.077us (13.0 Mqps) with batch size of 100 Time taken per op is 0.078us (12.8 Mqps) with batch size of 100 Test Plan: make check all make valgrind_check make asan_check Reviewers: sdong, ljin Reviewed By: ljin Subscribers: leveldb Differential Revision: https://reviews.facebook.net/D22539
10 years ago
// Iterate inside Cuckoo Block.
for (uint32_t block_idx = 0; block_idx < cuckoo_block_size_;
++block_idx, ++child_bucket_id) {
if ((*buckets)[static_cast<size_t>(child_bucket_id)].make_space_for_key_call_id ==
make_space_for_key_call_id) {
continue;
}
(*buckets)[static_cast<size_t>(child_bucket_id)].make_space_for_key_call_id =
make_space_for_key_call_id;
tree.push_back(CuckooNode(child_bucket_id, curr_depth + 1,
curr_pos));
if ((*buckets)[static_cast<size_t>(child_bucket_id)].vector_idx == kMaxVectorIdx) {
null_found = true;
break;
}
}
}
++curr_pos;
}
if (null_found) {
// There is an empty node in tree.back(). Now, traverse the path from this
// empty node to top of the tree and at every node in the path, replace
// child with the parent. Stop when first level is reached in the tree
// (happens when 0 <= bucket_to_replace_pos < num_hash_func_) and return
// this location in first level for target key to be inserted.
uint32_t bucket_to_replace_pos = static_cast<uint32_t>(tree.size()) - 1;
while (bucket_to_replace_pos >= num_hash_func_) {
CuckooNode& curr_node = tree[bucket_to_replace_pos];
(*buckets)[static_cast<size_t>(curr_node.bucket_id)] =
(*buckets)[static_cast<size_t>(tree[curr_node.parent_pos].bucket_id)];
bucket_to_replace_pos = curr_node.parent_pos;
}
*bucket_id = tree[bucket_to_replace_pos].bucket_id;
}
return null_found;
}
} // namespace rocksdb
#endif // ROCKSDB_LITE