3 ms·

that seems really tough for compilers.

• dmitrygr 2y ago
  Not really. Most modern x86 compilers already align jump targets to cache line boundaries since this helps x86 a lot. So it is doable. If you compile each function into a section (common), then the linker can be told to align them to 64 or 128 bytes easily. Code size would grow (but tetris can be played to reduce this by packing functions)