"Fast inverse square root" benchmark
This is a benchmark of the "fast inverse square root" algorithm found in Quake 3 Arena source code.
I created this because a coworker "kindly suggested" I use this "very optimised function" in some non-performance-critical code, running on x86. I did not agree and asked them to provide benchmarks that would prove its superior performance on modern hardware. They eventually let it go, but I decided to go ahead with the benchmark anyway.
tl;dr using it does not seem to make sense on any post-SSE2 (so, practically any 64-bit) CPU
See RESULTS.md for the actual results.