Here's my full test case, built with VS 2022 17.6 Preview 7 x64 (note that x64 vs. x86 makes a HUGE difference to charconv). I didn't bother to compare the from_chars scenario (as MSVC doesn't use a highly optimized algorithm there; investigating Eisel-Lemire is on our todo list).

Hope this helps, STL

Stephan, Thank you for this. I can add benchmarks for Lemire’s algorithm vs. the current MSVC STL from_chars to see if it’s worth pursuing for your team. Matt