Re: the concern about the compiler optimizing away the loop:
1. (re)checked the assembly in Godbolt. Still unsure of SIMDing, but can
point to the exact instruction that adds exactly three inside the loop. The sum
is also not optimized away.
2. To make doubly sure (as this is compiler-specific), I added the following
code right after array initialization and before the initial time check. It
turns the value to add into a run-time variable. No effect on speed in Nim
static arrays, for sufficiently long test times.
# Save a user-specified value to add to each element in the array
stdout.write("Add how much to each element in the array, per cycle (must be
an integer): ")
#
var to_add:type_t
try:
let tmp = readLine(stdin)
to_add = type_t(parseInt(tmp))
except:
echo "Could not read number!"
(and, below, replace the addition of 3 with "to_add" in the loop)
Run
