Some Intrinsic SUMS

[ivanpribec]

@RonShepard

Without optimization, I would have expected the loop to be faster due to the memory allocation and the extra memory references.

The compiler is smart enough that it doesn’t create an array temporary. See my previous response here: Size of long array - #44 by ivanpribec

Update:

@RonShepard

I am surprised that this code was faster for gfortran than the loop version. With optimization, I would have expected both to compile to exactly the same instructions.

Under -O2 they compile to the same instructions, but under -Og as used by @Harper the compiler uses extra move instructions to traverse the array. You can inspect the instruction generated in Compiler Explorer: Compiler Explorer