@neogoose_btw Your disassembler is using a -O3 flag, the other person used none (which GCC/G++ defaults to -O0), thus their benchmarked C++ code doesn't optimize the loop.
Comparison:
@astro__pallas librewolf is a really good alternative to firefox in my opinion - it's based on an older version but with a lot of privacy features and without all the tracking, though it's exclusive to linux from what I know, unless you can compile it for other systems