Re: Some optimisations for numeric division

On Fri, 25 Feb 2022 at 18:30, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>
> Dean Rasheed <dean(dot)a(dot)rasheed(at)gmail(dot)com> writes:
> > And another update following feedback from the cfbot.
>
> This patchset LGTM. On my machine there doesn't seem to be any
> measurable performance change for the numeric regression test,
> but numeric_big gets about 15% faster.
>

Yes, that matches my observations. Thanks for reviewing and testing.

> The only additional thought I have, based on your comments about
> 0001, is that maybe in mul_var we should declare var1digit as
> being NumericDigit not int. I tried that here and didn't see
> any material change in the generated assembly code (using gcc
> 8.5.0), so you're right that modern gcc doesn't need any help
> there; but I wonder if it could help on other compiler versions.
>

Yes, that makes sense. Done that way.

Regards,
Dean