See this : https://quick-bench.com/q/HCX5vCh3XSrwH6DpWNRFIDjbkBs

The code is below(gcc -O3 build):

static void StringCreation(benchmark::State& state) {
  // Code inside this loop is measured repeatedly
  for (auto _ : state) {
    std::string created_string("hello");
    // Make sure the variable is not optimized away by compiler
    benchmark::DoNotOptimize(created_string);
  }
}
// Register the function as a benchmark
BENCHMARK(StringCreation);

static void StringCopy(benchmark::State& state) {
  // Code before the loop is not measured
  std::string x = "hello";
  for (auto _ : state) {
    std::string copy(x);
  }
}
BENCHMARK(StringCopy);

The link result says StringCreation is 2.6 times faster than StringCopy.

I did it locally on x86_64, and the result is following:

Run on (4 X 3192.01 MHz CPU s)
CPU Caches:
  L1 Data 32 KiB (x4)
  L1 Instruction 32 KiB (x4)
  L2 Unified 256 KiB (x4)
  L3 Unified 12288 KiB (x4)
Load Average: 0.15, 0.98, 0.63
---------------------------------------------------------
Benchmark               Time             CPU   Iterations
---------------------------------------------------------
StringCreation       1.27 ns         1.27 ns    552871452
StringCopy           3.99 ns         3.98 ns    177720572

I don’t understand what metric can be used to compare the two in a fair way ? If iterations are not same, we can’t just say the shortest Time is the best implementation, right ? How the 2.6 times faster come from ?