[Tinycc-devel] Fast floating point & SSE2

[Daniel White]

Hi all,

I need to supply a compiler with my graphics software to allow
for super-fast user 'scripting'. However, speed is a concern.

Using the Microsoft compiler, I get about the same speed as
TCC. However the following MVSC flags really make a difference:

SSE2  (Streaming SIMD Extensions 2)
'Fast' floating point model (less accurate).

These speed up the code by a massive 3x. As you may have guessed,
it's quite mathsy (lots of trig), but the code in question is very
short (only about 10-20 lines).

Is there any hope of TCC supporting either of these? I think the
fast floating point model was the biggest contributor.

Thanks, Daniel