Shared/IntrinsicsSse4.h file

Intrinsics for x86 SSE4.1, SSE4.2 and POPCNT instructions.

Equivalent to #include <smmintrin.h> and #include <nmmintrin.h> on most compilers except for:

• GCC 4.8, where it contains an additional workaround to make the instructions available with just the DEATH_ENABLE_SSE41 and DEATH_ENABLE_SSE42 function attributes instead of having to specify -msse4.1 or -msse4.2 for the whole compilation unit. This however can't reliably be done for -mpopcnt because then it could not be freely combined with other instruction sets, only used alone. You have to enable these instructions globally in order to use them on GCC 4.8.
• Clang < 7, where __POPCNT__ has to be explicitly defined in order to access the POPCNT instruction

Because GCC puts both the SSE4.1 and the SSE4.2 instructions into the same header and just guards each with a different macro, they have to be included together, unlike with other SSE variants.