Comments by "MrAbrazildo" (@MrAbrazildo) on "Fastware - Andrei Alexandrescu" video.

  1. 54:34, I recently discovered that unrolling a loop of 'sscanf's to (I don't remember well) 50 numbers made it more than 3x faster, for both Clang and GCC. I thought they do that by default in -O2/3.
    1