[sldev] SSE2 / Ouch

[Robin Cornelius]

On 7/9/07, Tateru Nino <tateru.nino at gmail.com> wrote:
> You could overload the class methods, but it still might not play well
> in the preconstruction phase or destruction.
> You can try abstracting it all through a pointer or a reference off into
> 'untainted' portions of the text segment.

Hi everyone,

Sorry to jump in on the discussion, i have not made my self know before. Hello

If we go down this type of route do we not end up with effectivly
twice the binary after compile one lot with SSE enabled one lot
without and some bootstrap code (non SSE) either runs the SSE or non
SSE versions much in the same way the mac executables can contain two
different versions of executables for different processor
architectures. What worries me here is unless the split is very clean
it may be uncertain where the code is jumping to and you could easily
end up in a situation where you are hitting a mix of the SSE and non
SSE functions due to simple mistakes or omissions.

Would it be easier rather than making complex auto detection code, to
simply distribute two versions with and without SSE enabled? It should
be trival to batch build different versions in visual studio with
different build options. May be if we want to be clever the/an
installer can detect the SSE capability and then download the correct
file.

-- 
Robin Cornelius
http://www.byteme.org.uk