ARM > Introduction to ARM > Not a Trivial Mapping
Remarks
SIMD = Single Instruction Multiple Data.
MAC = Multiply Accumulate.
C compilers aren’t capable of producing every operation which a CPU can perform. Many compilers have intrinsics, which look like functions but emit the desired operations into the instruction stream. But no language can provide complete coverage for every CPU.
The compiler is likely to be a more consistent assembly language programmer than you. You must weigh up the pros and cons of using it. It’s not worth wasting time on assembly language code which the compiler could have produced.
The 80-20 rule: 80% of the time is spent executing 20% of the program. Concentrate on the 20%. It’s a rule of thumb.