Are _L variants slower? Will I get code bloat if I mix _L and normal variants of the same function? Or are both functions implemented only once underneath?
The performance of platform-aware functions with _L variant should be comparable with the none _L functions for compatible systems. _L functions can overcome 32-bit size limitations. Can you explain why you'd use the same function with both variants? This link https://software.intel.com/en-us/node/684662 provides more information.
