How to squeeze more performance out of gcd?

[Uknown]

On Monday, 15 June 2020 at 16:45:01 UTC, H. S. Teoh wrote:
> On Sat, Jun 13, 2020 at 02:25:21PM +0000, Uknown via 
> Digitalmars-d wrote: [...]
> [...]
> need to think about some way of reducing the number of gcd 
> calculations, since the gcd computation itself is already maxed 
> out in terms of optimization AFAICT.
>
>
> T

I didn't notice your reply, but if you haven't solved this yet, 
you could try this:

Make a simple recursive gcd implementation and memoize it. It 
might work better than any other implementation (depending on the 
numbers in question and the data set size)