optmized the emplace_back of rocksdb::autovector

autovcector_ben.cpp

//g++ -o2 autovcector_ben.cpp -lbenchmark -lpthread -o autovector_ben

#include <benchmark/benchmark.h>
#include "autovector.h"

std::vector<std::string> GetTestKeys(size_t size) {
  std::vector<std::string> keys;
  keys.resize(size);

  int index = 0;
  for (auto& key : keys) {
      key = "item-" + std::to_string(index++);
  }
  return keys;
}

std::vector<std::string> string_keys = GetTestKeys(100000*2*8);
// Define another benchmark
static void BM_AutovectorPushback(benchmark::State& state) {
    for (auto _ : state) {
            for (int i = 0; i< 100000; ++i) {
                autovector<std::string> as;
                for (size_t j = 0; j < state.range(0); ++j) {
                    as.push_back(string_keys[i]);
                }
            }
    }
}

BENCHMARK(BM_AutovectorPushback)->Arg(0)->Arg(1)->Arg(4)->Arg(8)->Arg(16);

static void BM_AutovectorEmplaceback(benchmark::State& state) {
    for (auto _ : state) {
            for (int i = 0; i< 100000; ++i) {
                autovector<std::string, 8> as;
                for (size_t j = 0; j < state.range(0); ++j) {
                    as.emplace_back(string_keys[i].c_str());
                }
            }
    }
}

BENCHMARK(BM_AutovectorEmplaceback)->Arg(0)->Arg(1)->Arg(4)->Arg(8)->Arg(16);

static void BM_AutovectorEmplaceback_optimize(benchmark::State& state) {
    for (auto _ : state) {
            for (int i = 0; i< 100000; ++i) {
                autovector<std::string, 8> as;
                for (size_t j = 0; j < state.range(0); ++j) {
                    as.emplace_back_optimize(string_keys[i].c_str());
                }
            }
    }
}

BENCHMARK(BM_AutovectorEmplaceback_optimize)->Arg(0)->Arg(1)->Arg(4)->Arg(8)->Arg(16);


BENCHMARK_MAIN();

rocksdb_autovector_optimize

#benchmark the rocksdb::autovector, the element in vector/autovecor was std::string
2018-10-30 09:49:40
Running ./autovector_ben
Run on (4 X 3400 MHz CPU s)
CPU Caches:
  L1 Data 32K (x2)
  L1 Instruction 32K (x2)
  L2 Unified 256K (x2)
  L3 Unified 4096K (x1)
----------------------------------------------------------------------------
Benchmark                                     Time           CPU Iterations
----------------------------------------------------------------------------
BM_AutovectorPushback/0                 7471002 ns    7471133 ns         90
BM_AutovectorPushback/1                 9029570 ns    9029851 ns         73
BM_AutovectorPushback/4                13361825 ns   13362373 ns         51
BM_AutovectorPushback/8                20416884 ns   20417729 ns         34
BM_AutovectorPushback/16              151630310 ns  151636696 ns          4
BM_AutovectorEmplaceback/0              7595114 ns    7595075 ns         88
BM_AutovectorEmplaceback/1             11405845 ns   11406004 ns         59
BM_AutovectorEmplaceback/4             23747714 ns   23748745 ns         29
BM_AutovectorEmplaceback/8             38301110 ns   38302593 ns         18
BM_AutovectorEmplaceback/16           184395206 ns  184402547 ns          4
BM_AutovectorEmplaceback_optimize/0     7551782 ns    7552118 ns         90
BM_AutovectorEmplaceback_optimize/1    11163601 ns   11163718 ns         60
BM_AutovectorEmplaceback_optimize/4    23270649 ns   23270982 ns         30
BM_AutovectorEmplaceback_optimize/8    36659024 ns   36660693 ns         19
BM_AutovectorEmplaceback_optimize/16  166090771 ns  166092358 ns          4