1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
|
#include "benchmark.hpp"
#include "execution.hpp"
#include "config.hpp"
#include "hash_maps.hpp"
#include <iostream>
#include <vector>
using config_t = tao::config::value;
template <class T>
struct hash_map_benchmark;
template <class T>
struct benchmark_thread : execution_thread {
benchmark_thread(hash_map_benchmark<T>& benchmark, std::uint32_t id, const execution& exec) :
execution_thread(id, exec),
_benchmark(benchmark)
{}
virtual void setup(const config_t& config) override {
execution_thread::setup(config);
_key_range = config.optional<std::uint64_t>("key_range").value_or(_benchmark.key_range);
_key_offset = config.optional<std::uint64_t>("key_offset").value_or(_benchmark.key_offset);
auto remove_ratio = config.optional<double>("remove_ratio").value_or(0.2);
if (remove_ratio < 0.0 || remove_ratio > 1.0)
throw std::runtime_error("remove_ratio must be >= 0.0 and <= 1.0");
auto insert_ratio = config.optional<double>("insert_ratio").value_or(0.2);
if (insert_ratio < 0.0 || insert_ratio > 1.0)
throw std::runtime_error("insert_ratio must be >= 0.0 and <= 1.0");
auto update_ratio = remove_ratio + insert_ratio;
if (update_ratio > 1.0)
throw std::runtime_error("The sum of remove_ratio and insert_ratio must be <= 1.0");
constexpr auto rand_range = std::numeric_limits<std::uint64_t>::max();
_scale_insert = static_cast<std::uint64_t>(insert_ratio * rand_range);
_scale_remove = static_cast<std::uint64_t>(update_ratio * rand_range);
}
virtual void initialize(std::uint32_t num_threads) override;
virtual void run() override;
virtual thread_report report() const override {
tao::json::value data {
{"runtime", _runtime.count()},
{"insert", insert_operations},
{"remove", remove_operations},
{"get", get_operations},
};
return { data, insert_operations + remove_operations + get_operations };
}
protected:
std::uint64_t insert_operations = 0;
std::uint64_t remove_operations = 0;
std::uint64_t get_operations = 0;
private:
hash_map_benchmark<T>& _benchmark;
std::uint64_t _key_range = 0;
std::uint64_t _key_offset = 0;
std::uint64_t _scale_remove = 0;
std::uint64_t _scale_insert = 0;
};
template <class T>
struct hash_map_benchmark : benchmark {
virtual void setup(const config_t& config) override;
virtual std::unique_ptr<execution_thread> create_thread(
std::uint32_t id,
const execution& exec,
const std::string& type) override
{
if (type == "mixed")
return std::make_unique<benchmark_thread<T>>(*this, id, exec);
throw std::runtime_error("Invalid thread type: " + type);
}
std::unique_ptr<T> hash_map;
std::uint32_t batch_size = 0;
std::uint64_t key_range = 0;
std::uint64_t key_offset = 0;
config::prefill prefill{};
};
template <class T>
void hash_map_benchmark<T>::setup(const tao::config::value& config) {
hash_map = hash_map_builder<T>::create(config.at("ds"));
batch_size = config.optional<std::uint32_t>("batch_size").value_or(100);
key_range = config.optional<std::uint64_t>("key_range").value_or(2048);
key_offset = config.optional<std::uint64_t>("key_offset").value_or(0);
// by default we prefill 10% of the configured key-range
prefill.setup(config, key_range / 10);
if (this->prefill.count > key_range)
throw std::runtime_error("prefill.count must be less or equal key_range");
}
template <class T>
void benchmark_thread<T>::initialize(std::uint32_t num_threads) {
auto id = this->id() & execution::thread_id_mask;
std::uint64_t cnt = _benchmark.prefill.get_thread_quota(id, num_threads);
[[maybe_unused]] region_guard_t<T> guard{};
auto step_size = _benchmark.key_range / _benchmark.prefill.count;
std::uint64_t key = id * step_size + _benchmark.key_offset;
step_size *= num_threads;
for (std::uint64_t i = 0 ; i < cnt; ++i, key += step_size) {
if (!try_emplace(*_benchmark.hash_map, static_cast<unsigned>(key))) {
throw initialization_failure();
}
}
}
template <class T>
void benchmark_thread<T>::run() {
T& hash_map = *_benchmark.hash_map;
const std::uint32_t n = _benchmark.batch_size;
std::uint32_t insert = 0;
std::uint32_t remove = 0;
std::uint32_t get = 0;
[[maybe_unused]] region_guard_t<T> guard{};
for (std::uint32_t i = 0; i < n; ++i) {
auto r = _randomizer();
auto key = static_cast<unsigned>((r % _key_range) + _key_offset);
if (r < _scale_insert) {
if (try_emplace(hash_map, key))
++insert;
} else if (r < _scale_remove) {
if (try_remove(hash_map, key))
++remove;
} else {
if (try_get(hash_map, key))
++get;
}
simulate_workload();
}
insert_operations += insert;
remove_operations += remove;
get_operations += get;
}
namespace {
template <class T>
inline std::shared_ptr<benchmark_builder> make_benchmark_builder() {
return std::make_shared<typed_benchmark_builder<T, hash_map_benchmark>>();
}
auto benchmark_variations()
{
using namespace xenium;
return benchmark_builders
{
#ifdef WITH_VYUKOV_HASH_MAP
#ifdef WITH_GENERIC_EPOCH_BASED
make_benchmark_builder<vyukov_hash_map<QUEUE_ITEM, QUEUE_ITEM, policy::reclaimer<reclamation::epoch_based<>>>>(),
make_benchmark_builder<vyukov_hash_map<QUEUE_ITEM, QUEUE_ITEM, policy::reclaimer<reclamation::new_epoch_based<>>>>(),
make_benchmark_builder<vyukov_hash_map<QUEUE_ITEM, QUEUE_ITEM, policy::reclaimer<reclamation::debra<>>>>(),
#endif
#ifdef WITH_QUIESCENT_STATE_BASED
make_benchmark_builder<vyukov_hash_map<QUEUE_ITEM, QUEUE_ITEM, policy::reclaimer<reclamation::quiescent_state_based>>>(),
#endif
#ifdef WITH_HAZARD_POINTER
make_benchmark_builder<
vyukov_hash_map<QUEUE_ITEM, QUEUE_ITEM, policy::reclaimer<
reclamation::hazard_pointer<>::with<policy::allocation_strategy<reclamation::hp_allocation::static_strategy<3>>>>>>(),
make_benchmark_builder<
vyukov_hash_map<QUEUE_ITEM, QUEUE_ITEM, policy::reclaimer<
reclamation::hazard_pointer<>::with<policy::allocation_strategy<reclamation::hp_allocation::dynamic_strategy<3>>>>>>(),
#endif
#endif
#ifdef WITH_HARRIS_MICHAEL_HASH_MAP
#ifdef WITH_GENERIC_EPOCH_BASED
make_benchmark_builder<harris_michael_hash_map<QUEUE_ITEM, QUEUE_ITEM, policy::reclaimer<reclamation::epoch_based<>>>>(),
make_benchmark_builder<harris_michael_hash_map<QUEUE_ITEM, QUEUE_ITEM, policy::reclaimer<reclamation::new_epoch_based<>>>>(),
make_benchmark_builder<harris_michael_hash_map<QUEUE_ITEM, QUEUE_ITEM, policy::reclaimer<reclamation::debra<>>>>(),
#endif
#ifdef WITH_QUIESCENT_STATE_BASED
make_benchmark_builder<harris_michael_hash_map<QUEUE_ITEM, QUEUE_ITEM, policy::reclaimer<reclamation::quiescent_state_based>>>(),
#endif
#ifdef WITH_HAZARD_POINTER
make_benchmark_builder<
harris_michael_hash_map<QUEUE_ITEM, QUEUE_ITEM, policy::reclaimer<
reclamation::hazard_pointer<>::with<
policy::allocation_strategy<reclamation::hp_allocation::static_strategy<3>>>>>>(),
make_benchmark_builder<
harris_michael_hash_map<QUEUE_ITEM, QUEUE_ITEM, policy::reclaimer<
reclamation::hazard_pointer<>::with<policy::allocation_strategy<reclamation::hp_allocation::dynamic_strategy<3>>>>>>(),
#endif
#endif
#ifdef WITH_CDS_MICHAEL_HASHMAP
make_benchmark_builder<
cds::container::MichaelHashMap<cds::gc::HP,
cds::container::MichaelKVList< cds::gc::HP, QUEUE_ITEM, QUEUE_ITEM>>>(),
#endif
#ifdef WITH_CDS_FELDMAN_HASHMAP
make_benchmark_builder<cds::container::FeldmanHashMap<cds::gc::HP, QUEUE_ITEM, QUEUE_ITEM>>(),
#endif
};
}
}
void register_hash_map_benchmark(registered_benchmarks& benchmarks) {
benchmarks.emplace("hash_map", benchmark_variations());
}
|
