How to I phrase RIP addressing in assembly under Linux AMD64 in D.  I can't seem to figure it out.
Normally I do something like:

movq rax,rndphi[rip]

Where rndphi is a label pointing to data.

I know I have change movq to mov in D and hope the operation works on all 64 bits.

If for some reason it is not possible in D I can still put all my special vectorized code in a shared library and access it that way.  In fact I should code it on a GPU where it likely would run 100 times faster. Maybe I should set up a Patreon account and see if there are any true believers in evolution out there.
https://github.com/S6Regen/Thunderbird

On Sunday, 19 August 2018 at 07:18:20 UTC, Sean O'Connor wrote:
> How to I phrase RIP addressing in assembly under Linux AMD64 in D.  I can't seem to figure it out.
> Normally I do something like:
>
> movq rax,rndphi[rip]
>
> Where rndphi is a label pointing to data.
>
> I know I have change movq to mov in D and hope the operation works on all 64 bits.
>
> If for some reason it is not possible in D I can still put all my special vectorized code in a shared library and access it that way.  In fact I should code it on a GPU where it likely would run 100 times faster. Maybe I should set up a Patreon account and see if there are any true believers in evolution out there.
> https://github.com/S6Regen/Thunderbird

Hello, there's an issue opened https://issues.dlang.org/show_bug.cgi?id=17617.
RIP is not available ATM.

Okay, I see.  I'll try with a shared library.  I only have 3 or 4 algorithms that don't auto-vectorize well with modern compilers and have to be written by hand.
Patreon have that terrible catch 22 tax form that stopped me writing for US magazines, rather unfortunate.