[fpc-devel] Current work

[J. Gareth Moreton]

Thanks Denis,

I have to be slightly careful though because the example I gave with 
"addps" is not actually correct, because (%r8) could be unaligned (it 
works fine with "vaddps" though due to a paradigm change in the design 
of AVX compared to SSE).  In this situation a more correct example would be:

     movups    (%rdx),%xmm0
     movups    (%r8),%xmm1
     addps    %xmm1,%xmm0
     movups    %xmm0,(%rcx)
     ret

This, like pure functions, is another one of those big changes that need 
a lot of testing and experimentation, and probably a means to turn off 
if it is actually merged into the main branch. It would be good to 
gather some input from the moderators and administrators on this.

Gareth aka. Kit

On 23/08/2024 20:59, denisgolovan wrote:
> Nice work, Gareth.
> Vectorization is definitely a nice improvement for the compiler.
>
>
> -- Regards,
> Denis Golovan
>

-- 
This email has been checked for viruses by Avast antivirus software.
www.avast.com