Real Intel Skylake and AMD Zen cores expose 6-10 execution ports specifically to exploit the kind of independent, unrolled arithmetic you write here, and the restrict keyword you'll see recommended for avoiding pointer aliasing is a real, load-bearing hint that GCC and Clang use to enable more aggressive out-of-order and vectorized scheduling. Compilers routinely perform the same loop-unrolling and independent-accumulator transformation shown here as an -O2/-O3 optimization pass.