Looking for more library optimization patterns

[Ivan Kazmenko]

On Thursday, 30 January 2014 at 18:36:09 UTC, bearophile wrote:
> A de-optimization I'd really like in LDC (and dmd) is to not 
> call the run-time functions when you perform a verctor 
> operation on arrays statically known to be very short:
>
> void main() {
>     int[3] a, b;
>     a[] += b[];
> }
>
> And just rewrite that with a for loop (as done for cases where 
> the run time function is not available), and let ldc2 compile 
> it on its own.

For short loops, an unrolled version like
     a[0] += b[0];
     a[1] += b[1];
     a[2] += b[2];
may well be faster than a simple loop as the following one:
     foreach (immutable i; 0..3) {
         a[i] += b[i];
     }
At least on x86/64.
Will that optimization happen too, at a later stage?

Ivan Kazmenko.