basic reimplementation of inet_ntop (ipv6 only) and benchmark against libc.
dataset:
auto data = std::array{
"2001:0db8:85a3:0000:0000:8a2e:0370:7334",
"2001:db8::1:0",
"2001:db8:0:1:1:1:1:1",
"2001:db8:1234:ffff:ffff:ffff:ffff:ffff",
"2001:db8:85a3:8d3:1319:8a2e:370:7348",
"fe80::1ff:fe23:4567:890a",
"64:ff9b::255.255.255.255",
"2001:db8:3333:4444:5555:6666:7777:8888",
"2001:db8::123.123.123.123",
"2001:db8::1234:5678:5.6.7.8",
"::1",
"::",
"::123.123.123.123",
};on my machine the result is:
libc, fmt ---- 2023-12-28T17:27:53+01:00 Running ./bench_inet_ntop Run on (12 X 5000 MHz CPU s) CPU Caches: L1 Data 32 KiB (x6) L1 Instruction 32 KiB (x6) L2 Unified 256 KiB (x6) L3 Unified 12288 KiB (x1) Load Average: 1.65, 1.12, 1.00 ***WARNING*** CPU scaling is enabled, the benchmark real time measurements may be noisy and will incur extra overhead. ----------------------------------------------------------- Benchmark Time CPU Iterations ----------------------------------------------------------- BM_runner/glibc 2769 ns 2769 ns 251399 BM_runner/fmt_v1 2775 ns 2774 ns 241027 BM_runner/fmt_v2 1950 ns 1950 ns 398019
tested only on linux. Needs cmake and a c++ compiler:
mkdir build
cmake -S . -B build -DCMAKE_BUILD_TYPE=Release
cmake --build build
build/bench_inet_ntop